This PDF file contains the front matter associated with SPIE Proceedings Volume 7965, including the Title Page, Copyright information, Table of Contents, and the Conference Committee listing.

Functional neuroimaging research is moving from the study of "activations" to the study of "interactions" among brain regions. Granger causality analysis provides a powerful technique to model spatio-temporal interactions among brain regions. We apply this technique to full-brain fMRI data without aggregating any voxel data into regions of interest (ROIs). We circumvent the problem of dimensionality using sparse regression from machine learning. On a simple finger-tapping experiment we found that (1) a small number of voxels in the brain have very high prediction power, explaining the future time course of other voxels in the brain; (2) these voxels occur in small sized clusters (of size 1-4 voxels) distributed throughout the brain; (3) albeit small, these clusters overlap with most of the clusters identified with the non-temporal General Linear Model (GLM); and (4) the method identifies clusters which, while not determined by the task and not detectable by GLM, still influence brain activity.

Classical measures of functional connectivity assume that the stationarity of the time courses and the time-invariance of functional connectivity under investigation. These assumptions may not be valid in the real cases. Also, they are bivariate measures and may not provide the directional information flow between brain units. A new approach is proposed to tackle these problems. A statistics reasoning shows that the short-length time course is more likely to be stationary than the long-length time course. Thus, the entire time course under investigation is divided into short segments with the proper length. Magnitude squared coherence (in spectrum domain) is computed to assess functional connectivity on these segments, hence, provides a dynamic measure of functional connectivity. The averaged magnitude squared coherence over the segments gives an overall measure of functional connectivity. This approach has been applied to several neuroimaging data analysis. The results and the interpretations / predictions are in good agreement. Mutual coherence (in time domain) is computed to assess functional connectivity, hence, provides an insight on directional information flow. By using grid computing, this approach will be extended from the bivariate to the multivariate.

The disgust system arises phylogenetically in response to dangers to the internal milieu from pathogens and their toxic products. Functional imaging studies have demonstrated that a much wider range of neural structures was involved in triggering disgust reactions. However, less is known regarding how and what neural pathways these neural structures interact. To address this issue, we adopted an effective connectivity based analysis, namely the multivariate Granger causality approach, to explore the causal interactions within these brain networks. Results presented that disgust can induce a wide range of brain activities, such as the insula, the anterior cingulate cortex, the parahippocampus lobe, the dorsal lateral prefrontal cortex, the superior occipital gyrus, and the supplementary motor cortex. These brain areas constitute as a whole, with much denser connectivity following disgust stimuli, in comparison with that of the neutral condition. Moreover, the anterior insula, showing multiple casual interactions with limbic and subcortical areas, was implicated as a central hub in organizing multiple information processing in the disgust system.

It is well known that adults are experts at processing words and faces. Accordingly, adult research has identified two neural expertise systems involved in word processing and face processing within the fusiform gyrus, respectively, namely the visual word form area (VWFA) and fusiform face area (FFA). The present study used fMRI to explore whether similar differentiations exist for the FFA and VWFA in 10~11-aged children, by comparing the activation between faces, Chinese characters, and common objects. Our study identified adult-like Chinese character-preferential activation and common object-preferential activation in 10~11-aged children, especially with the fusiform gyrus, while fail to reveal a consistent region showing preferential response to faces. An inspection of individual activation of faces relative to Chinese characters and common objects revealed adults-like FFA in some of children, indicating that the absence of face-preferential activation at the group level may be mainly due to the considerable variability in the magnitude and locus of individual face-preferential activation. Our finds suggested that the Chinese character-preferential regions and common object-preferential regions within the fusiform gyrus may be formed earlier than that of faces. Especially, though the VWFA and FFA are both related to visual expertise, our findings indicated that the VWFA can be formed only through a 3~4-years' schooling; in contrast the formation of FFA appear to undergo a more prolonged development before it reaches the adult level.

Task-based neuroimaging studies revealed that different attention operations were carried out by the functional interaction and cooperation between two attention systems: the dorsal attention network (DAN) and the ventral attention network (VAN), which were respectively involved in the "top-down" endogenous attention orienting and the "bottomup" exogenous attention reorienting process. Recent focused resting functional MRI (fMRI) studies found the two attention systems were inherently organized in the human brain regardless of whether or not the attention process were required, but how the two attention systems interact with each other in the absence of task is yet to be investigated. In this study, we first separated the DAN and VAN by applying the group independent component analysis (ICA) to the resting fMRI data acquired from 12 healthy young subjects, then used Gaussian Bayesian network (BN) learning approach to explore the plausible effective connectivity pattern of the two attention systems. It was found regions from the same attention network were strongly intra-dependent, and all the connections were located in the information flow from VAN to DAN, which suggested that an orderly functional interactions and information exchanges between the two attention networks existed in the intrinsic spontaneous brain activity, and the inherent connections might benefit the efficient cognitive process between DAN and VAN, such as the "top-down" and "bottom-up" reciprocal interaction when attention-related tasks were involved.

Parameters extracted from the vasculature on the retina are correlated with various conditions such as diabetic retinopathy and cardiovascular diseases such as stroke. Segmentation of the vasculature on the retina has been a topic that has received much attention in the literature over the past decade. Analysis of the segmentation result, however, has only received limited attention with most works describing methods to accurately measure the width of the vessels. Analyzing the connectedness of the vascular network is an important step towards the characterization of the complete vascular tree. The retinal vascular tree, from an image interpretation point of view, originates at the optic disc and spreads out over the retina. The tree bifurcates and the vessels also cross each other. The points where this happens form the key to determining the connectedness of the complete tree. We present a supervised method to detect the bifurcations and crossing points of the vasculature of the retina. The method uses features extracted from the vasculature as well as the image in a location regression approach to find those locations of the segmented vascular tree where the bifurcation or crossing occurs (from here, POI, points of interest). We evaluate the method on the publicly available DRIVE database in which an ophthalmologist has marked the POI.

The detection of early-stage keratoconus is one of the most important safety issues in screening candidates for corneal refractive surgeries. We propose to use epithelial thickness maps to assist the diagnosis of keratoconus. The corneal epithelial thickness in normal and keratoconic eyes was mapped with optical coherence tomography (OCT). A Fourier-domain OCT system capable of acquiring 26,000 axial-scans per second was used. It has an axial resolution of 5μm in cornea. A pachymetry scan pattern (8 radials, 1024 axial-scans each, 6mm diameter, repeat 3 times) centered at the pupil center was used to image the cornea. The 3 repeated radial scans on each meridian were registered and averaged. Then the anterior corneal, posterior corneal and epithelial boundaries were segmented automatically with a computer algorithm by increased signal intensity at corresponding boundaries. The epithelial thickness map was generated by interpolating epithelial thickness profile calculated from each meridian. Normal and keratoconic eyes (24 eyes each) were scanned 3 times. The central epithelial thickness in normal eyes was thicker than those of keratoconic eyes (mean difference 2.1 μm, t-test p=0.05). The epithelium was thinner superiorly than inferiorly in normal eyes (mean difference -1.4±1.1μm, p<0.001) while thicker superiorly than inferiorly in keratoconic eyes (2.0±4.1 μm, p=0.02).

Assessing the hemodynamic status of the brain and its variations in response to stimulations is required to understand the local cerebral circulatory mechanisms. Dynamic contrast enhanced imaging of cerebral microvasculature provides information that can be used in understanding physiology of cerebral diseases. Bolus tracking is used to extract characteristic parameters that quantify local cerebral blood flow. However, post-processing of the data is needed to segment the field of view (FOV) and to perform deconvolution to remove the effects of input bolus profile and the path it travels to reach the imaging window. Finding the arterial input function (AIF) and dealing with the ill-posedness of deconvolution system make this process are the main challenges. We propose using ICA to segment the FOV and to extract a local AIF as well as the venous output function that is required for deconvolution. This also helps to stabilize the system as ICA suppresses noise efficiently. Tikhoniv regularization (with L-curve analysis to find the best regularization parameter) is used to make the system stable. In-vivo dynamic 2PLSM images of a rat brain in two conditions (when the animal is at rest and when it is stimulated) are used in this study. The experimental along with the simulation studies provided promising results that demonstrate the feasibility and importance of performing deconvolution.

Among many optical molecular imaging modalities, bioluminescence imaging (BLI) has more and more wide application in tumor detection and evaluation of pharmacodynamics, toxicity, pharmacokinetics because of its noninvasive molecular and cellular level detection ability, high sensitivity and low cost in comparison with other imaging technologies. However, BLI can not present the accurate location and intensity of the inner bioluminescence sources such as in the bone, liver or lung etc. Bioluminescent tomography (BLT) shows its advantage in determining the bioluminescence source distribution inside a small animal or phantom. Considering the deficiency of two-dimensional imaging modality, we developed three-dimensional tomography to reconstruct the information of the bioluminescence source distribution in transgenic mOC-Luc mice bone with the boundary measured data. In this paper, to study the osteocalcin (OC) accumulation in transgenic mOC-Luc mice bone, a BLT reconstruction method based on multilevel adaptive finite element (FEM) algorithm was used for localizing and quantifying multi bioluminescence sources. Optical and anatomical information of the tissues are incorporated as a priori knowledge in this method, which can reduce the ill-posedness of BLT. The data was acquired by the dual modality BLT and Micro CT prototype system that was developed by us. Through temperature control and absolute intensity calibration, a relative accurate intensity can be calculated. The location of the OC accumulation was reconstructed, which was coherent with the principle of bone differentiation. This result also was testified by ex vivo experiment in the black 96-plate well using the BLI system and the chemiluminescence apparatus.

In vivo bioluminescence imaging (BLI) has played a more and more important role in biomedical research of small animals. Tomographic bioluminescence imaging (TBI) further translates the BLI optical information into three-dimensional bioluminescent source distribution, which could greatly facilitate applications in related studies. Although the diffusion approximation (DA) is one of the most widely-used forward models, higher-order approximations are still needed for in vivo small animal imaging. In this work, as a relatively accurate and higher-order approximation theory, a simplified spherical harmonics approximation (SP_N) is applied for heterogeneous tomographic bioluminescence imaging in vivo. Furthermore, coupled with the SP_N, a generalized graph cuts optimization approach is utilized, making BLT reconstructions fast and suit for the whole body of small animals. Heterogeneous in vivo experimental reconstructions via the higher-order approximation model demonstrate higher tomographic imaging quality, which is shown the capability for practical biomedical tomographic imaging applications.

Personalized treatment of prostate cancer would be enhanced by an assessment of cancer stage and grade from imaging, the validation of which requires the accurate co-registration of in vivo images with a gold standard for stage and grade established by histopathology. We present a visualization tool supporting an image-guided approach enabling the acquisition of histopathology images parallel to the in vivo imaging planes, simplifying this registration. This tool decreases imaging-to-specimen landmark alignment error by 62%, and decreases the time required to mark the slicing plane on the specimen by 47%. Preliminary results from our method demonstrate the alignment of regions suspicious for cancer on T2w MRI with confirmed cancer foci on histopathology, and we calculate a sub-millimeter in-plane target registration error.

Automatic small animal whole-body organ registration is challenging because of subject's joint structure, posture and position difference and loss of reference features. In this paper, an improved 3D shape context based non-rigid registration method is applied for mouse whole-body skeleton registration and lung registration. A geodesic path based non-rigid registration method is proposed for mouse torso skin registration. Based on the above registration methods, a novel non-rigid registration framework is proposed for mouse whole-body organ mapping from an atlas to new scanned CT data. A preliminary experiment was performed to test the method on lung and skin registration. A whole-body organ mapping was performed on three target data and the selected organs were compared with the manual outlining results. The robust of the method has been demonstrated.

We present a novel method for three-dimensional (3D) shape parameterization. The approach is affine invariant and is applied to comparing local shape across abdominal organs. The inherent structure of the abdominal organs is used to generate a regular sampling of the organ's surface. 'Planar-convexity' is defined for a general 3D closed object as the property that there exists a set of parallel planes which cover the 3D space such that every intersection of a plane with the object is a singular closed planar curve. We show that this parameterization, combined with the 3D analogue of a 2D shape descriptor, successfully shown to be invariant under affine transformations and noise, can be effectively used to compare features of two closed 3D surfaces point-to-point. The technique avoids common problems with the parameterization of concave surfaces and shows great potential for analyzing and improving the automatic modeling and segmentation of abdominal organs.

Duchenne muscular dystrophy (DMD) is a progressive and fatal X-linked disease caused by mutations in the DMD gene. Magnetic resonance imaging (MRI) has shown potential to provide non-invasive and objective biomarkers for monitoring disease progression and therapeutic effect in DMD. In this paper, we propose a semi-automated scheme to quantify MRI features of golden retriever muscular dystrophy (GRMD), a canine model of DMD. Our method was applied to a natural history data set and a hydrodynamic limb perfusion data set. The scheme is composed of three modules: pre-processing, muscle segmentation, and feature analysis. The pre-processing module includes: calculation of T2 maps, spatial registration of T2 weighted (T2WI) images, T2 weighted fat suppressed (T2FS) images, and T2 maps, and intensity calibration of T2WI and T2FS images. We then manually segment six pelvic limb muscles. For each of the segmented muscles, we finally automatically measure volume and intensity statistics of the T2FS images and T2 maps. For the natural history study, our results showed that four of six muscles in affected dogs had smaller volumes and all had higher mean intensities in T2 maps as compared to normal dogs. For the perfusion study, the muscle volumes and mean intensities in T2FS were increased in the post-perfusion MRI scans as compared to pre-perfusion MRI scans, as predicted. We conclude that our scheme successfully performs quantitative analysis of muscle MRI features of GRMD.

The abnormal thermogram has been shown to be a reliable indicator of a high risk of breast cancer. Nevertheless, a major weakness of current infrared breast thermography is its poor sensitivity for deeper tumors. Numerical modeling for breast thermography provides an effective tool to investigate the complex relationships between the breast thermal behaviors and the underlying patho-physiological conditions. We have developed a set of new modeling techniques to take into account some subtle factors usually ignored in previous studies, such as gravity-induced elastic deformations of the breast, nonlinear elasticity of soft tissues, and dynamic behavior of thermograms. Conventional "forward problem" modeling cannot be used directly to improve tumor detectability, however, because the underlying tissue thermal properties are generally unknown. Therefore, we propose an "inverse problem" modeling technique that aims to estimate the tissue thermal properties from the breast surface thermogram. Our data suggest that the estimation of the tumor-induced thermal contrast can be improved significantly by using the proposed inverse problem solving techniques to provide the individual-specific thermal background, especially for deeper tumors. We expect the proposed new methods, taken together, to provide a stronger foundation for, and greater specificity and precision in, thermographic diagnosis, and treatment, of breast cancer.

Osteogenesis imperfecta (OI) is a genetic syndrome affecting collagen synthesis and assembly. Its symptoms vary widely but commonly include bone fragility, reduced stature, and bone deformity. Because of the small size and paucity of human specimens, there is a lack of biomechanical data for OI bone. Most literature has focused on histomorphometric analyses, which rely on assumptions to extrapolate 3-D properties. In this study, a micro-computed tomography (μCT) system was used to directly measure structural and mineral properties in pediatric OI bone collected during routine surgical procedures. Surface renderings suggested a poorly organized, plate-like orientation. Patients with a history of bone-augmenting drugs exhibited increased bone volume fraction (BV/TV), trabecular number (Tb.N), and connectivity density (Eu.Conn.D). The latter two parameters appeared to be related to OI severity. Structural results were consistently higher than those reported in a previous histomorphometric study, but these differences can be attributed to factors such as specimen collection site, drug therapy, and assumptions associated with histomorphometry. Mineral testing revealed strong correlations with several structural parameters, highlighting the importance of a dual approach in trabecular bone testing. This study reports some of the first quantitative μCT data of human OI bone, and it suggests compelling possibilities for the future of OI bone assessment.

Research on bone and teeth mineralization in animal models is critical for understanding human pathologies. Genetically modified mice represent highly valuable models for the study of osteo/dentinogenesis defects and osteoporosis. Current investigations on mice dental and skeletal phenotype use destructive and time consuming methods such as histology and scanning microscopy. Micro-CT imaging is quicker and provides high resolution qualitative phenotypic description. However reliable quantification of mineralization processes in mouse bone and teeth are still lacking. We have established novel CT imaging-based software for accurate qualitative and quantitative analysis of mouse mandibular bone and molars. Data were obtained from mandibles of mice lacking the Fibromodulin gene which is involved in mineralization processes. Mandibles were imaged with a micro-CT originally devoted to industrial applications (Viscom, X8060 NDT). 3D advanced visualization was performed using the VoxBox software (UsefulProgress) with ray casting algorithms. Comparison between control and defective mice mandibles was made by applying the same transfer function for each 3D data, thus allowing to detect shape, colour and density discrepencies. The 2D images of transverse slices of mandible and teeth were similar and even more accurate than those obtained with scanning electron microscopy. Image processing of the molars allowed the 3D reconstruction of the pulp chamber, providing a unique tool for the quantitative evaluation of dentinogenesis. This new method is highly powerful for the study of oro-facial mineralizations defects in mice models, complementary and even competitive to current histological and scanning microscopy appoaches.

We analyse μ-CT tomographic images of human trabecular bone in vitro. We consider a sample consisting of 201 bone specimens harvested from six different skeletal sites within a narrow range of bone fraction values. Using the characterization of the trabecular bone network given by local Minkowski Functionals, we apply classification algorithms in order to reveal structural similarities in the sample. Clusters show some interesting specific structural features, like compact, porous, and fragmented structures. The contribution of the different skeletal sites to these clusters indicate some variability due to intrinsic structural differences of the specific skeletal site.

Immunocytochemical and immunofluorescence staining are used for identifying the characteristics of metastasis in traditional ways. Micro-computed tomography (micro-CT) is a useful tool for monitoring and longitudinal imaging of tumor in small animal in vivo. In present study, we evaluated the feasibility of the detection for metastasis of gastric carcinoma by high-resolution micro-CT system with omnipaque accumulative enhancement method in the organs. Firstly, a high-resolution micro-CT ZKKS-MCT-sharp micro-CT was developed by our research group and Guangzhou Zhongke Kaisheng Medical Technology Co., Ltd. Secondly, several gastric carcinoma models were established through inoculating 2x10⁶ BGC-823 gastric carcinoma cells subcutaneously. Thirdly, micro-CT scanning was performed after accumulative enhancement method of intraperitoneal injection of omnipaque contrast agent containing 360 mg iodine with a concentration of 350 mg I/ml. Finally, we obtained high-resolution anatomical information of the metastasis in vivo in a BALB/c NuNu nude mouse, the 3D tumor architecture is revealed in exquisite detail at about 35 μm spatial resolution. In addition, the accurate shape and volume of the micrometastasis as small as 0.78 mm³ can be calculated with our software. Overall, our data suggest that this imaging approach and system could be used to enhance the understanding of tumor proliferation, metastasis and could be the basis for evaluating anti-tumor therapies.

Background: Evaluation of cardiovascular function in mice using micro-CT requires that a contrast agent (CA) be administered to differentiate the blood from the myocardium. eXIA 160, an aqueous colloidal poly-disperse CA with a high concentration of iodine (160mg I/mL), creates strong contrast between blood and tissue with a low injection volume. In this study, the blood-pool enhancement time-course of eXIA 160 is monitored over a 24-hour period to determine its optimal use during cardiac function studies. Methods/Results: 8-second scans were performed (80kVp, 110mA) using the GE Locus Ultra micro-CT scanner. Male mice (black, 22-24g) were injected via tail vein with 5 μL/g body weight eXIA 160 (Binitio Biomedical Inc.). A precontrast scan was performed; following injection, mice were scanned at 15, 30, 45, and 60 minutes, 2, 4, 8, 24, and 48 hours. Overall, the highest contrast in the left ventricle occurred at 5 minutes (687 HU). Uptake of the CA by the myocardium was also observed: myocardial tissue showed increasing enhancement over a 4-hour period, remaining even once the contrast was eliminated from the vasculature. Conclusion: eXIA 160 provided high contrast between blood and myocardial tissue for a period of 30 minutes following injection. Notably, this CA was also taken up by the myocardium and provided continued enhancement when the contrast agent was eliminated from the blood, making LV wall function studies possible. In conclusion, eXIA 160, with its high iodine concentration and targeted tissue uptake characteristics, make it an ideal agent to use when evaluating cardiovascular function in mice.

Advances in laboratory imaging systems for CT, SPECT, MRI, and PET facilitate routine micro-imaging during pre-clinical investigations. Challenges still arise when dealing with immune-compromised animals, biohazardous agents, and multi-modality imaging. These challenges can be overcome with an appropriate animal management system (AMS), with the capability for supporting and monitoring a rat or mouse during micro-imaging. We report the implementation and assessment of a new AMS system for mice (PRA-3000 / AHS-2750, ASI Instruments, Warren MI), designed to be compatible with a commercial micro-CT / micro-SPECT imaging system (eXplore speCZT, GE Healthcare, London ON). The AMS was assessed under the following criteria: 1) compatibility with the imaging system (i.e. artifact generation, geometric dimensions); 2) compatibility with live animals (i.e. positioning, temperature regulation, anesthetic supply); 3) monitoring capabilities (i.e. rectal temperature, respiratory and cardiac monitoring); 4) stability of co-registration; and 5) containment. Micro-CT scans performed using a standardized live-animal protocol (90 kVp, 40 mA, 900 views, 16 ms per view) exhibited low noise (±19 HU) and acceptable artifact from high-density components within the AMS (e.g. ECG pad contacts). Live mice were imaged repeatedly (with removal and replacement of the AMS) and spatial registration was found to be stable to within ±0.07 mm. All animals tolerated enclosure within the AMS for extended periods (i.e. > one hour) without distress, based on continuous recordings of rectal temperature, ECG waveform and respiratory rate. A sealed AMS system extends the capability of a conventional micro-imaging system to include immune-compromised and biosafety level 2 mouse-imaging protocols.

Deformation Based Morphometry (DBM) is a relatively new method used for characterizing anatomical differences among populations. DBM is based on the analysis of the deformation fields generated by non-rigid registration algorithms, which warp the individual volumes to one standard coordinate system. Although several studies have compared non-rigid registration algorithms for segmentation tasks, few studies have compared the effect of the registration algorithm on population differences that may be uncovered through DBM. In this study, we compared DBM results obtained with five well established non-rigid registration algorithms on the corpus callosum (CC) in thirteen subjects with Williams Syndrome (WS) and thirteen Normal Control (NC) subjects. The five non-rigid registration algorithms include: (1) The Adaptive Basis Algorithm (ABA); (2) Image Registration Toolkit (IRTK); (3) FSL Nonlinear Image Registration Tool (FSL); (4) Automatic Registration Tools (ART); and (5) the normalization algorithm available in SPM8. For each algorithm, the 3D deformation fields from all subjects to the atlas were obtained and used to calculate the Jacobian determinant (JAC) at each voxel in the mid-sagittal slice of the CC. The mean JAC maps for each group were compared quantitatively across different nonrigid registration algorithms. An ANOVA test performed on the means of the JAC over the Genu and the Splenium ROIs shows the JAC differences between nonrigid registration algorithms are statistically significant over the Genu for both groups and over the Splenium for the NC group. These results suggest that it is important to consider the effect of registration when using DBM to compute morphological differences in populations.

Accurate ventricle volume estimates could improve the understanding and diagnosis of postoperative communicating hydrocephalus. For this category of patients, associated changes in ventricle volume can be difficult to identify, particularly over short time intervals. We present an automated segmentation algorithm that evaluates ventricle size from serial brain MRI examination. The technique combines serial T1- weighted images to increase SNR and segments the means image to generate a ventricle template. After pre-processing, the segmentation is initiated by a fuzzy c-means clustering algorithm to find the seeds used in a combination of fast marching methods and geodesic active contours. Finally, the ventricle template is propagated onto the serial data via non-linear registration. Serial volume estimates were obtained in an automated robust and accurate manner from difficult data.

Clinical and pre-clinical studies show that deep brain stimulation (DBS) of targeted brain regions by neurosurgical techniques ameliorate psychiatric disorder such as anorexia nervosa. Neurosurgical interventions in preclinical rodent brain are mostly accomplished manually with a 2D atlas. Considering both the large number of animals subjected to stereotactic surgical experiments and the associated imaging cost, feasibility of sophisticated pre-operative imaging based surgical path planning and/or robotic guidance is limited. Here, we spatially normalize vasculature information and assess the intra-strain variability in cerebral vasculature for a neurosurgery planning. By co-registering and subsequently building a probabilistic vasculature template in a standard space, we evaluate the risk of a user defined electrode trajectory damaging a blood vessel on its path. The use of such a method may not only be confined to DBS therapy in small animals, but also could be readily applicable to a wide range of stereotactic small animal surgeries like targeted injection of contrast agents and cell labeling applications.

Understanding the effects of adolescent binge drinking that persist into adulthood is a crucial public health issue. Adolescent intermittent ethanol exposure (AIE) is an animal model that can be used to investigate these effects in rodents. In this work, we investigate the application of a particular image analysis technique, tensor-based morphometry, for detecting anatomical differences between AIE and control rats using Diffusion Tensor Imaging (DTI). Deformation field analysis is a popular method for detecting volumetric changes analyzing Jacobian determinants calculated on deformation fields. Recent studies showed that computing deformation field metrics on the full deformation tensor, often referred to as tensor-based morphometry (TBM), increases the sensitivity to anatomical differences. In this paper we conduct a comprehensive TBM study for precisely locating differences between control and AIE rats. Using a DTI RARE sequence designed for minimal geometric distortion, 12-directional images were acquired postmortem for control and AIE rats (n=9). After preprocessing, average images for the two groups were constructed using an unbiased atlas building approach. We non-rigidly register the two atlases using Large Deformation Diffeomorphic Metric Mapping, and analyze the resulting deformation field using TBM. In particular, we evaluate the tensor determinant, geodesic anisotropy, and deformation direction vector (DDV) on the deformation field to detect structural differences. This yields data on the local amount of growth, shrinkage and the directionality of deformation between the groups. We show that TBM can thus be used to measure group morphological differences between rat populations, demonstrating the potential of the proposed framework.

Examining the spontaneous activity to understand the neural mechanism of brain disorders and establish neuroimaging-based disease-related biomarkers is a focus in recent resting-state functional MRI (fMRI) studies. The present study hypothesized that resting activity in the default mode network (DMN), which was used for characterizing the resting-state human brain might be different in patients with depressed Parkinson disease (dPD) compared with non-depressed Parkinson disease (ndPD) patients. To test the hypothesis, we firstly employed the Group independent component analysis (ICA) approach to isolate the DMN for the two groups by analyzing the resting-state fMRI data from a group of 12 patients with dPD and a group of 12 age-matched ndPD subjects. Between-group comparison of the functional connectivity in the DMN was then performed to examine the impact of depression on the intrinsic activity in PD. We found 1) the core region from the network the medial prefrontal cortex (MPFC) show significant decreased activity in dPD group compared with ndPD group; 2) the activity in MPFC has significant negative correlation with behavioral measure; 3) the resting activity intensity of MPFC is suggested to be a promising biomarker for distinguishing dPD from ndPD.

Magnetic Particle Imaging (MPI) has been introduced as a modality that allows for the acquisition of three-dimensional functional images with high sensitivity in real time. Here, alternative coil topologies are presented that differ significantly from the original set-up. Two novel coil topologies will be presented. Beside an asymmetric coil topology, where all field generating coils are arranged on a single side, an effective coil assembly has been accomplished that creates a field-free line for spatial coding. The alternative coil topologies may overcome the problem of a confined measurement field or lead to an increase of the sensitivity of MPI.

We have developed a new systems theory for analyzing Magnetic Particle Imaging (MPI), x-space MPI, that motivates an image reconstruction technique requiring only a simple scaling and mapping of the raw signal to image space. As a result, reconstruction time scales linearly with image size. Moreover, we expect there to be no noise gain due to performing a matrix inverse. Reconstruction is simple and can be performed in real-time. Furthermore, there is no "System Function" and no requirement to pre-characterize the nanoparticle in the imaging system before imaging. So far, there has not been an experimental test of the x-space theory on an imaging system. In this paper we begin by describing the theoretical signal that we would expect from the imager for a SPIO nanoparticle source. We then construct a small scale MPI imager to test x-space. We conclude with images from the x-space scanner using the iron oxide tracer Resovist. Our preliminary images show 1.6 mm spatial resolution using a 6.0 T/m gradient that closely matches the expected image resolution from x-space theory. In conclusion, our data supports the theory that MPI can be considered a linear, shift invariant system. This bodes well for the future as we scale MPI scanners to human sizes as large images can be reconstructed simply and rapidly.

Molecular imaging technologies are advancing rapidly and optical techniques in particular are set to play a large role in preclinical pharmaceutical testing. These approaches, however, are generally unable to quantify the level of expression of imaging probe reporters. In this study a novel method of quantification is presented using dual-probe fluorescence imaging, where an endothelial growth factor receptor (EGFR) fluorescent probe was paired with a non-targeted probe before being injected, and tracer kinetic compartmental modeling was used to determine EGFR expression in a region of interest from the uptake curves of the two drugs in that region. The approach was tested out in a simulation experiment and then applied in an in vivo study in one mouse to investigate EGFR expression in various tissue types (pancreas, pancreas tumor, and leg). The binding potentials (a unitless correlate of target availability) of EGFR expression in the various tissue types were 8.57, 25.64, and 0.11 for the pancreas, pancreas tumor, respectively. For the pancreas and leg, these results correlate well with expected levels of EGFR expression, with the pancreas demonstrating a much higher expression than the skin and also as expected, the tumor expressed much more EGFR than either healthy tissue.

We report the use of a nanoparticle magnetization spectrometer to measure the suitability of magnetite nanoparticles for magnetic particle imaging (MPI). The spectrometer, which measures the strength of harmonics in the magnetization of magnetic nanoparticles, produces a driving field at 25 kHz and measures up to 40 harmonics (1Mhz bandwidth). Large, uniform, single-core magnetite particles with diameter ranging from 17 to 22 nm were synthesized for this work and their magnetization spectra were compared with a commercially available option, Resovist. The driving field amplitude was varied and spectra were acquired for field amplitudes ranging from 9.7 to 18.6 mTμ₀^-1. Compared with Resovist, synthesized particles showed increased harmonic power over the full measured frequency range at each driving field.

The field free line (FFL) magnet has the potential to greatly increase signal to noise ratio (SNR) or to decrease scan time for magnetic particle imaging (MPI). The use of an FFL will decrease scan time by reducing image dimensionality from a 3D image to a projection image. Alternatively, in comparison to a 3D scan of equal scan time, an FFL scanner will increase SNR through more signal averages. An FFL magnet would enable projection imaging as is used in projection x-ray and is common in angiography. The Philips and Lubeck groups have pioneered the design of field free line magnets for MPI and have shown that they can achieve power efficiency similar to that of a field free point, the standard in MPI. Current FFL magnet designs have not been optimized for characteristics such as gradient efficiency and gradient magnitude homogeneity. This work shows a 2.25 T/m Halbach quadrupole permanent magnet design that produces a homogeneous magnetic field along the field free line. Along the FFL, we experimentally measured a field maximum of 2mT within the imaging field of view (FOV), and we experimentally measured that the gradient perpendicular to the FFL deviates by a maximum of 3.4%. In future work, we plan to produce an x-space MPI image using the FFL magnet. We also plan to improve upon this design using optimization techniques.

Binding affinity can be estimated in several ways in the laboratory but there is no viable way to estimate binding affinity in vivo without assumptions on the number of binding sites. Magnetic spectroscopy of nanoparticle Brownian motion, MSB, measures the rotational Brownian motion. The MSB signal is affected by nanoparticle binding affinity so it provides a mechanism to measure the chemical binding affinity. We present a possible mechanism to quantify the binding affinity and test that mechanism using viscous solutions.

Magnetic resonance imaging (MRI) cell tracking has become an important non-invasive technique to interrogate the fate of cells upon transplantation. At least 6 clinical trials have been published at the end of 2010, all of which have shown that real-time monitoring of the injection procedure, initial engraftment, and short-term biodistribution of cells is critical to further advance the field of cellular therapeutics. In MRI cell tracking, cells are loaded with superparamagnetic iron oxide (SPIO) particles that provide an MRI contrast effect through microscopic magnetic field disturbances and dephasing of protons. Magnetic particle imaging (MPI) has recently emerged as a potential cellular imaging technique that promises to have several advantages over MRI, primarily linear quantification of cells, a higher sensitivity, and "hot spot" tracer identification without confounding background signal. Although probably not fully optimized, SPIO particles that are currently used as MRI contrast agent can be employed as MPI tracer. Initial studies have shown that cells loaded with SPIO particles can give a detectable MPI signal, encouraging further development of MPI cell tracking.

Magnetic Particle Imaging (MPI) is a high-potential new medical imaging modality that has been introduced in 2005. MPI uses the non-linear magnetization behavior of iron-oxide based nano-particles, named tracer, to perform quantitative measurements of their local concentration. Previous publications demonstrated the feasibility of real-time in vivo 3D imaging with clinical concentration of Resovist®. Given MPI's fast and sensitive imaging as well as its overall versatility, it has potential to support various medical applications spanning from diagnostics to therapy. As an example, ongoing research investigates the use of MPI in cardiovascular diagnostics for myocardial perfusion measurement. While previous publications reported results from experimental systems with limited bore size (3cm), this contribution presents first phantom and in vivo images acquired on the next hardware generation, an experimental system with an effective bore size of 12cm. The system is designed for pre-clinical studies and can capture image data from an extended field of view compared to the previous, experimental system. The contribution introduces concepts for the encoding of a larger field of view by means of additional magnetic fields, named focus-fields, and outlines the path to stitching of images from multiple focus field settings, called "multi-station reconstruction". To prove the feasibility of imaging of an extended field of view, volumetric images of a moving phantom as well as of a living rat were acquired.

Multi-modality PET-CT imaging is playing an important role in the field of oncology. While PET imaging facilitates functional interrogation of tumor status, the use of CT imaging is primarily limited to anatomical reference. In an attempt to extract comprehensive information about tumor cells and its microenvironment, we used a nanoparticle xray contrast agent to image tumor vasculature and vessel 'leakiness' and 18F-FDG to investigate the metabolic status of tumor cells. In vivo PET/CT studies were performed in mice implanted with 4T1 mammary breast cancer cells.Early-phase micro-CT imaging enabled visualization 3D vascular architecture of the tumors whereas delayedphase micro-CT demonstrated highly permeable vessels as evident by nanoparticle accumulation within the tumor. Both imaging modalities demonstrated the presence of a necrotic core as indicated by a hypo-enhanced region in the center of the tumor. At early time-points, the CT-derived fractional blood volume did not correlate with 18F-FDG uptake. At delayed time-points, the tumor enhancement in 18F-FDG micro-PET images correlated with the delayed signal enhanced due to nanoparticle extravasation seen in CT images. The proposed hybrid imaging approach could be used to better understand tumor angiogenesis and to be the basis for monitoring and evaluating anti-angiogenic and nano-chemotherapies.

Optical imaging (OI) is a relatively new method in detecting active inflammation of hand joints of patients suffering from rheumatoid arthritis (RA). With the high number of people affected by this disease especially in western countries, the availability of OI as an early diagnostic imaging method is clinically highly relevant. In this paper, we present a newly in-house developed OI analyzing tool and a clinical evaluation study. Our analyzing tool extends the capability of existing OI tools. We include many features in the tool, such as region-based image analysis, hyper perfusion curve analysis, and multi-modality image fusion to aid clinicians in localizing and determining the intensity of inflammation in joints. Additionally, image data management options, such as the full integration of PACS/RIS, are included. In our clinical study we demonstrate how OI facilitates the detection of active inflammation in rheumatoid arthritis. The preliminary clinical results indicate a sensitivity of 43.5%, a specificity of 80.3%, an accuracy of 65.7%, a positive predictive value of 76.6%, and a negative predictive value of 64.9% in relation to clinical results from MRI. The accuracy of inflammation detection serves as evidence to the potential of OI as a useful imaging modality for early detection of active inflammation in patients with rheumatoid arthritis. With our in-house developed tool we extend the usefulness of OI imaging in the clinical arena. Overall, we show that OI is a fast, inexpensive, non-invasive and nonionizing yet highly sensitive and accurate imaging modality.-

Quantitative analysis of lymphatic function is crucial for understanding the lymphatic system and diagnosing the associated diseases. Recently, a near-infrared (NIR) fluorescence imaging system is developed for real-time imaging lymphatic propulsion by intradermal injection of microdose of a NIR fluorophore distal to the lymphatics of interest. However, the previous analysis software3, 4 is underdeveloped, requiring extensive time and effort to analyze a NIR image sequence. In this paper, we develop a number of image processing techniques to automate the data analysis workflow, including an object tracking algorithm to stabilize the subject and remove the motion artifacts, an image representation named flow map to characterize lymphatic flow more reliably, and an automatic algorithm to compute lymph velocity and frequency of propulsion. By integrating all these techniques to a system, the analysis workflow significantly reduces the amount of required user interaction and improves the reliability of the measurement.

Current methods for detecting source number in magnetic source image (MSI) have some shortcomings: i) they are somewhat subjective, ii) various MSIs employ different indexes, and iii) the number of the potential sources in MSI via Beamforming may be underestimated. The new approach extracts a principal component from each trial which serves as a (one) representative virtual channel for that trial. Principal components from all trials are combined to form a (one) synthesized trial. The information criterion is applied to this synthesized trial to select an optimal estimate of the source number. PCA reduces noise and captures the greatest variability in each trial, the synthesized trial reduces data fluctuations among trials and also reduces the dimensions of the covariance matrix. The approach is entirely data-driven and is suitable for detecting seizures and interictal spikes in epilepsy patients.

Prior work has shown that white matter fiber integrity decreases in Alzheimer's disease (AD) and mild cognitive impairment (MCI). This can be achieved by quantifying anisotropic water movement in the brain using diffusion tensor imaging techniques. Since less than half (but still a considerable amount) of the MCI patients convert to AD it is important to identify features that can predict the chance of conversion to AD within a certain time frame. In this study we applied tract-based spatial statistics (TBSS) in order to perform this task, which overcomes limitations that are commonly associated with ROI-based approaches and voxel-based morphometry (VBM). Diffusion weighted images were taken from 15 healthy controls, 15 AD patients and 17 MCI patients. 8 MCI patients remained stable within 3 years of follow-up investigations ("non-converters" or MCI-nc) and 9 converted to AD ("converters" or MCI-c). Significant differences between the MCI-nc and MCI-c groups were found in large parts of the fornix, the corpus callosum and the cingulum. In comparison, the MCI-c group did not differ significantly from the AD group and the MCI-nc group exhibited features similar to the control group in most parts of the structures. These results demonstrate that, although MCI-c and MCI-nc patients were clinically similar at time of inclusion, the MCI-c group already exhibited pathologic features of fiber integrity associated with AD. This finding could lead to more powerful techniques in the early identification of AD and thus support an earlier and more successful treatment.

Hypoxic-Ischemic Encephalopathy (HIE) is the brain manifestation of systemic asphyxia that occurs in 20 out of 1000 births. HIE triggers an immediate neuronal and glial injury leading to necrosis secondary to cellular edema and lysis. Because of this destructive neuronal injury, up to 25% of neonates exhibit severe permanent neuropsychological handicaps in the form of cerebral palsy, with or without associated mental retardation, learning disabilities, or epilepsy. Due to the devastating consequences of HIE, much research has focused on interrupting the cascade of events triggered by HIE. To date, none of these therapies, with the exception of hypothermia, have been successful in the clinical environment. Even in the case of hypothermia, only neonates with mild to moderate HIE respond to therapy. Stem cell therapy offers an attractive potential treatment for HIE. The ability to replace necrotic cells with functional cells could limit the degree of long-term neurological deficits. The neonatal brain offers a unique milieu for stem cell therapy due to its overall plasticity and the continued division of cells in the sub-ventricular zones. New powerful imaging tools allow researchers to track stem cells in vivo post-transplant, as shown in Figure 1. However, neuroimaging still leaves numerous questions unresolved: How can we identify stem cells without using tracking agents, what cells types are destroyed in the brain post injury? What is the final phenotypic fate of transplanted cells? Are the transplanted cells still viable? Do the transplanted cells spare endogenous neuronal tissue? We hypothesize that magnetic resonance spectroscopy (MRS), a broadly used clinical technique that can be performed at the time of a standard MRI scan, can provide answers to these questions when coupled with advanced computational approaches. MRS is widely available clinically, and is a relative measure of different metabolites within the sampled area. These measures are presented as a series of peaks at a particular bandwidth that corresponds to an individual metabolite, such as lactate or creatine, as shown in Figure 2. Currently, the data are only subjectively interpreted by a neuro-radiologist, but hold great potential if they were analyzed in a more objective manner. The overall purpose of the research described here is to develop pattern recognition algorithms for MRS data as a means to detect novel biomarkers or fingerprints of stem cells. Once identified, this technique will be used to identify in vivo transplanted stem cells within the brain.

Sparse partial correlation is a useful connectivity measure for brain networks when it is difficult to compute the exact partial correlation in the small-n large-p setting. In this paper, we formulate the problem of estimating partial correlation as a sparse linear regression with a l1-norm penalty. The method is applied to brain network consisting of parcellated regions of interest (ROIs), which are obtained from FDG-PET images of the autism spectrum disorder (ASD) children and the pediatric control (PedCon) subjects. To validate the results, we check their reproducibilities of the obtained brain networks by the leave-one-out cross validation and compare the clustered structures derived from the brain networks of ASD and PedCon.

The analysis of real-time biomolecular interactions (observation is performed as the biological interaction occurs) provides information on the formation of target/probe complexes, particularly on their dynamic behaviours. Namely, it allows the determination of the affinity constant, a static value that characterizes the interaction properties, using two dynamic values, the association and dissociation constants. Such dynamic behaviour can be assessed either with surface plasmon resonance (SPR) or uorescence-based biosensors. The challenging issue is the automatic extraction and analysis of the interaction signal for each spotted probe on the biosensor in a highthroughput framework (hundreds of probes). This paper addresses such issue and develops a uniffied approach for analyzing the image data provided by the above-mentioned technologies. A mathematical modelling of the image data allowed building-up a virtual biosensor able to simulate biologic experiences related to various possible parameters (level of signal and noise, presence of artefacts, surface functionalization, spotting heterogeneity). Based on such simulation, a generic and automated approach combining 3D mathematical morphology and spatio-temporal classiffication is proposed for detecting the interacting probes, segmenting the regions of effective signal, and characterizing the associated affinity constants. The developed method has been assessed both qualitatively and quantitatively on simulated and experimental datasets and showed accurate results (maximum error of 7% for the most difficult cases in terms of noise and surface functionalization).

Targeted fluorophore-labeled contrast agents are moving toward translation to human surgical use. To prepare for future clinical use, we examined the performance of potential ligands targeting the epidermal growth factor receptor, α5β3 integrins, and GLUT transporters for their suitability as directed contrast agents. Each agent was labeled with IRDye 800CW, and near-infrared dye with excitation/emission wavelengths of 789/805 nm, which we determined had favorable toxicity characteristics. The probe molecules examined consisted of Affibodies, nanobodies, peptides, and the sugar 2-deoxy-D-glucose. Each probe was tested for specific and non-specific binding in cell based assays. All probe types showed good performance in mouse models for detecting either spontaneous tumors or tumor xenografts in vivo. Each of the probes tested show promise for future human clinical studies.

Fluorescence molecular tomography (FMT) has become a promising imaging modality for in vivo small animal molecular imaging, and has many successful applications. This is partly due to the wealth of the fluorescent probes. By labeling the regions of interest with fluorescent probes, FMT can achieve non-invasive investigation of the biological process by localizing the targeted probes based on certain inverse mathematical models. However, FMT is usually an illposed problem, and some form of regularization should be included to stabilize the problem, which can be considered as the a priori information of the fluorescent probe bio-distribution. When FMT is used for the early detection of tumors, an important characteristic is the sparsity of the fluorescent sources. This is because tumors are usually very small and sparse at this stage. Considering this, general sparsity-promoting Lp-norm regularization is utilized in this paper. The iterated shrinkage based reconstruction method is adopted to solve the general Lp regularization problem. However, the original iterated shrinkage method is proved to have a linear convergence rate, and a large number of iterations are needed to obtain satisfactory results. In this paper, an improved iterated shrinkage based FMT reconstruction algorithm is proposed. By using the solutions from two previous iterations to determine the current solution, the convergence rate can be greatly increased. Heterogeneous simulation experiment shows that the proposed method can obtain comparable results with greatly reduced number of iterations compared with the original iterated shrinkage based method, which makes it a practical reconstruction algorithm.

As a newly emerged optical imaging method, fluorescence molecular imaging technique has been receiving increasing attention for its ability of non-invasive visualization of the cellular and molecular activities. However, as a kind of background noise, autofluorescence is a major disturbing factor in fluorescence molecular imaging. In this paper, we proposed a novel method to eliminate autofluorescence of small animals. The method is based on the fact that most autofluorescent signal has a broad excitation and emission spectrum, whereas specific fluorescent probe has a narrow one. First, two fluorescent images are obtained at two different excitation wavelengths. Then we divide the two obtained fluorescent images into blocks with the size of 8×8 pixel. The two blocks from the same position of the two different images respectively constitute a block pair. The ratio of one block's summation of total pixel value to that of ther other block belonging to the same block pair is calculated. After that, we classify all block pairs into fluorescent and nonfluorescent ones by ratio. The former are considered to be actual fluorescent regions. In next step, we adopt an adaptive cluster analysis method to classify all fluorescent block pairs into multiple interest regions. A general centroid algorithm is then applied to locate the center of each interest regions. We recover the fluorescent interest regions using flood filling algorithm. Finally, we choose a GFP-transfected tumor mouse model and a GFP-transplanted mouse skin model to validate our algorithm.

The tumor extracellular matrix has been focused on by newer approaches to cancer therapy owing to its important functions in the process of drug delivery and cellular metastasis. This study aims to characterize tumor extracellular matrix structures in the presence and absence of therapy, as observed on second harmonic generation (SHG) images through both gray-level co-occurrence matrix (GLCM) derived texture features as well as Minkowski Functionals (MF) that focus on the underlying gray-level topology and geometry of the texture patterns. Thirteen GLCM texture features and three MF texture features were extracted from 119 regions of interest (ROI) annotated on SHG images of treated and control samples of tumor extracellular matrix. These texture features were then used in a machine learning task to classify ROIs as belonging to treated or control samples. A fuzzy k-nearest neighbor classifier was optimized using random sub-sampling cross-validation for each texture feature and the classification performance was calculated on an independent test set using the area under the ROC curve (AUC); AUC distributions of different features were compared using a Mann-Whitney U-test. Two GLCM features f3 and f13 exhibited a significantly higher classification performance when compared to other GLCM features (p < 0.05). The MF feature Area exhibited the best classification performance among the MF features while also being comparable to that obtained with the best GLCM features. These results show that both statistical and topological texture features can be used as quantitative measures is evaluating the effects of therapy on the tumor extracellular matrix.

Cerebral aneurysm rupture is a leading cause of hemorrhagic strokes. Because they are being more frequently diagnosed before rupture and the prognosis of subarachnoid hemorrhage is poor, clinicians are often required to judge which aneurysms are prone to progression and rupture. Unfortunately, the processes of aneurysm initiation, growth and rupture are not well understood. Multiple factors associated to these processes have been identified. Our goal is to investigate two of them, arterial hemodynamics (using computational fluid dynamics) and the peri-aneurysmal environment, by studying a group of growing cerebral aneurysms that are followed longitudinally in time. Six patients with unruptured untreated brain aneurysms which exhibited growth during the observation period were selected for the study. Vascular models of each aneurysm at each observation time were constructed from the corresponding computed tomography angiography (CTA) images. Subsequently, models were aligned, and geometrical differences quantified. Blood flow was modeled with the 3D unsteady incompressible Navier-Stokes equation for a Newtonian fluid, and wall shear stress distribution and flow patterns were calculated and visualized. Analysis of the simulations and changes in geometry revealed asymmetric growth patterns and suggests that areas subject to vigorous flows, i.e. relative high wall shear stress and concentrated streamlines patterns; correspond to regions of aneurysm growth. Furthermore, in some cases the geometrical evolution of aneurysms is clearly affected by contacts with bone structures and calcifications in the wall, and as a consequence the hemodynamics is greatly modified. Thus, in these cases the peri-aneurysmal environment must be considered when analyzing aneurysm evolution.

To treat or prevent some of the 795,000 annual strokes in the U.S., self-expanding endo-vascular stents deployed under fluoroscopic image guidance are often used. Neuro-interventionalists need to know the deployment behavior of each stent in order to place them in the correct position. Using the Micro-Angiographic Fluoroscope (MAF) which has about 3 times higher resolution than commercially available flat panel detectors (FPD) we studied the deployment mechanics of two of the most important commercially available nitinol stents: the Pipeline embolization device (EV3), and the Enterprise stent (Codman). The Pipeline stent's length extends to about 3 times that of its deployed length when it is contained inside a catheter. From the high-resolution images with the MAF we found that upon the sudden release of the distal end of the Pipeline from a helical wire cap, the stent expands radially but retracts to about 30% (larger than for patient deployments) of its length. When released from the catheter proximally, it retracts additionally about 50% contributing to large uncertainty in the final deployed location. In contrast, the MAF images clearly show that the Enterprise stent self expands with minimal length retraction during deployment from its catheter and can be retrieved and repositioned until the proximal markers are released from clasping structures on its guide-wire thus enabling more accurate placement at the center of an aneurysm or stenosis. The high-resolution imaging demonstrated in this study should help neurointerventionalists understand and control endovascular stent deployment mechanisms and hence perform more precise treatments.

Aneurysm treatment using flow diversion could become the treatment of choice in the near future. While such side-wall aneurysm treatments have been studied in many publications and even implemented in selected clinical cases, bifurcation aneurysm treatment using flow diversion has not been addressed in detail. Using angiographic imaging, we evaluated treatment of such cases with several stent designs using patient-specific aneurysm phantoms. The aim is to find a way under fluoroscopic image guidance to place a low-porosity material across the aneurysm orifice while keeping the vessel blockage minimal. Three pre-shaped self-expanding stent designs were developed: the first design uses a middle-flap wing stent, the second uses a two-tapered-wing-ended stent, and the third is a slight modification of the first design in which the middle-flap is anchored tightly against the aneurysm using a standard stent. Treatment effects on flow were evaluated using high-speed angiography (30 fps) and compared with the untreated aneurysm. Contrast inflow was reduced in all the cases: 25% for Type 1, 63% for type 2 and 88% for Type 3. The first and the second stent design allowed some but substantially-reduced flow inside the aneurysm neck as indicated by the time-density curves. The third stent design eliminated almost all flow directed at the aneurysm dome, and only partial filling was observed. In the same time Type 1 and 3 delayed the inflow in the branches up to 100% compared to the untreated phantom. The results are quite promising and warrant future study.

Dynamic contrast enhanced CT has been successfully applied in cardiac imaging for the estimation of myocardial blood flow (MBF). In general, these acquisitions impart a relatively high radiation dose because they require continuous or gated imaging of the heart for 15-40 seconds. At present, there is no consensus on the appropriate estimation method to derive MBF and on the appropriate acquisition technique to minimize dose while maintaining MBF estimation accuracy and precision. This work explores the tradeoff of accuracy and precision of MBF estimates with several estimation methods and acquisition techniques in support of the fundamental goal of optimizing dynamic cardiac CT in terms of radiation dose and MBF estimation fidelity. We simulated time attenuation curves (TACs) for a range of flow states (Flow = [0.8, 1.6, 2.4, 3.2] ml/g/min) and several acquisition techniques. We estimated MBF with 5 different methods for each simulated TAC. From multiple independent realizations, we assessed the accuracy and precision of each method. Results show that acquisition techniques with 1/3 tube current or 1/3 temporal sampling permits accurate MBF estimates with most methods with reduction in MBF estimate precision by on average 30%. Furthermore, reduction in model complexity can be beneficial for improving the precision of MBF estimates.

Dynamic Contrast-Enhanced Magnetic Resonance Imaging (DCE-MRI) is becoming an indispensable tool to non-invasively study tumor characteristics. However, many different DCE-analysis methods are currently being used. To compare and validate different methods, histology is the gold standard. For this purpose, exact co-localization between histology and MRI images is a prerequisite. In this study a methodology is developed to validate DCE-data with histology with an emphasis on correct registration of DCE-MRI and histological data. A pancreatic tumor was grown in a rat model. The tumor was dissected after MR imaging, embedded in paraffin, and cut into thin slices. These slices were stained with haematoxylin and eosin, digitized and stacked in a 3D volume. Next, the 3D histology was registered to ex-vivo SWI-weighted MR images, which in turn were registered to in-vivo SWI and DCE images to achieve correct co-localization. Semi-quantitative and quantitative parameters were calculated. Preliminary results suggest that both pharmacokinetic and heuristic DCE-parameters can discriminate between vital and non-vital tumor regions. The developed method offers the basis for an accurate spatial correlation between DCE-MRI derived parametric maps and histology, and facilitates the evaluation of different DCE-MRI analysis methods.

Two-dimensional echocardiography continues to be the most widely used modality for the assessment of cardiac function due to its effectiveness, ease of use, and low costs. Echocardiographic images are derived from the mechanical interaction between the ultrasound field and the contractile heart tissue. Previously, in [6], based on B-mode echocardiographic simulations, we showed that motion estimation errors are significantly higher in shift-varying simulations when compared to shift-invariant simulations. In order to ascertain the effect of the spatial variance of the Ultrasonic field point spread function (PSF) and the transducer geometry on motion estimation, in the current paper, several simple canonical cardiac motions such as translation in axial and horizontal direction, and out-of-plane motion were simulated and the motion estimation errors were calculated. For axial motions, the greatest angular errors occurred within the lateral regions of the image, irrespective of the motion estimation technique that was adopted. We hypothesize that the transducer geometry and the PSF spatial-variance were the underlying sources of error for the motion estimation methods. No similar conclusions could be made regarding motion estimation errors for azimuthal and out-of-plane ultrasound simulations.

Current optical imaging techniques can successfully measure tumor load in murine models of lung carcinoma but lack structural detail. We demonstrate that respiratory gated micro-CT imaging of such models gives information about structure and correlates with tumor load measurements by optical methods. Four mice with multifocal, Kras-induced tumors expressing firefly luciferase were imaged against four controls using both optical imaging and respiratory gated micro-CT. CT images of anesthetized animals were acquired with a custom CNT-based system using 30 ms x-ray pulses during peak inspiration; respiration motion was tracked with a pressure sensor beneath each animal's abdomen. Optical imaging based on the Luc+ signal correlating with tumor load was performed on a Xenogen IVIS Kinetix. Micro-CT images were post-processed using Osirix, measuring lung volume with region growing. Diameters of the largest three tumors were measured. Relationships between tumor size, lung volumes, and optical signal were compared. CT images and optical signals were obtained for all animals at two time points. In all lobes of the Kras+ mice in all images, tumors were visible; the smallest to be readily identified measured approximately 300 microns diameter. CT-derived tumor volumes and optical signals related linearly, with r=0.94 for all animals. When derived for only tumor bearing animals, r=0.3. The trend of each individual animal's optical signal tracked correctly based on the CT volumes. Interestingly, lung volumes also correlated positively with optical imaging data and tumor volume burden, suggesting active remodeling.

Chronic Obstructive Pulmonary Disease (COPD) affects almost 600 million people and is currently the fourth leading cause of death worldwide. COPD is an umbrella term for respiratory symptoms that accompany destruction of the lung parenchyma and/or remodeling of the airway wall, the sum of which result in decreased expiratory flow, dyspnea and gas trapping. Currently, x-ray computed tomography (CT) is the main clinical method used for COPD imaging, providing excellent spatial resolution for quantitative tissue measurements although dose limitations and the fundamental spatial resolution of CT limit the measurement of airway dimensions beyond the 5^th generation. To address this limitation, we are piloting the use of bronchoscopic Optical Coherence Tomography (OCT), by exploiting its superior spatial resolution of 5-15 micrometers for in vivo airway imaging. Currently, only manual segmentation of OCT airway lumen and wall have been reported but manual methods are time consuming and prone to observer variability. To expand the utility of bronchoscopic OCT, automatic and robust measurement methods are required. Therefore, our objective was to develop a fully automated method for segmenting OCT airway wall dimensions and here we explore several different methods of image-regeneration, voxel clustering and post-processing. Our resultant automated method used K-means or Fuzzy c-means to cluster pixel intensity and then a series of algorithms (i.e. cluster selection, artifact removal, de-noising) was applied to process the clustering results and segment airway wall dimensions. This approach provides a way to automatically and rapidly segment and reproducibly measure airway lumen and wall area.

We demonstrate the application of our cardiac- and respiratory-gated carbon nanotube (CNT) micro-CT system by evaluating murine myocardial infarction models with a delayed contrast enhancement technique. Myocardial infarction was induced in 8 wild-type male mice. The ischemia reperfusion model was achieved by surgical occlusion of the LAD artery for 30 minutes followed by 24 hours of reperfusion. Free-breathing subjects were anesthetized with isoflurane during imaging. Respiratory and cardiac signals were monitored externally to gate the scan. Micro-CT data was obtained at 50kV, 3mA cathode current for 15ms per projection. All images were acquired during end exhalation at either 0msec or 55msec after the R-wave (diastole or systole, respectively). Following administration of Omnipaque 300mgI/mL at 0.1ml/5g, images were obtained at 0msec after the R-wave. Fenestra VC was then administered at a 0.1ml/5g dose, followed by images 0 and 55msec after the R-wave. Hearts were then harvested, sliced 1mm thick and stained with TTC. All animals survived surgery and imaging; all demonstrated obvious delayed contrast enhancement in the left ventricular wall in Omnipaque images. Fenestra VC revealed cardiac functional changes quantified by low ejection fractions. All subjects demonstrated areas of myocardial infarct in the LAD distribution on both TTC staining and micro-CT imaging. CNT enabled gated cardiac micro-CT imaging demonstrates the ability to consistently identify areas of myocardial infarct in mice, providing a powerful tool for the study of cardiovascular biology. Further work is ongoing to streamline the imaging protocol and perform more quantitative analysis of the images.

In this study, an automated analysis of pulmonary ventilation (AAPV) was developed to visualize the ventilation in pediatric lungs using dynamic Xe-133 scintigraphy. AAPV is a software algorithm that converts a dynamic series of Xe- 133 images into four functional images: equilibrium, washout halftime, residual, and clearance rate by analyzing pixelbased activity. Compared to conventional methods of calculating global or regional ventilation parameters, AAPV provides a visual representation of pulmonary ventilation functions.

The recognition of abnormalities relative to the lobular anatomy has become increasingly important in the diagnosis and differential diagnosis of lung abnormalities at clinical routines of CT examinations. This paper aims for a 3-D microstructural analysis of the pulmonary acinus with isotropic spatial resolution in the range of several micrometers by using micro CT. Previously, we demonstrated the ability of synchrotron radiation micro CT (SRμCT) using offset scan mode in microstructural analysis of the whole part of the secondary pulmonary lobule. In this paper, we present a semi-automatic method to segment the acinar and subacinar airspaces from the secondary pulmonary lobule imaged by the SRμCT. The method began with a segmentation of the tissues such as pleural surface, interlobular septa, alveola wall, or vessel using threshold technique and 3-D connected component analysis. Follow-on stages then constructed 3-D air space separated by tissues and represented branching patterns of airways and airspaces distal to the terminal bronchiole. Finally, a graph-partitioning approach isolated acini whose stems were interactively defined as the terminal bronchiole in the secondary pulmonary lobule. Additionally, the isolated acinar airspace was segmented into subacini in which the airway was considered as the stem using the graph-partitioning approach. Results demonstrate that the proposed method can extract several acinar airspaces from the 3-D SRμCT image of secondary pulmonary lobule and that the extracted acinar airspace enable an accurate quantitative description of the anatomy of the human acinus for interpretation of the basic unit of pulmonary structure and function.

Acupoint specificity, lying at the core of the Traditional Chinese Medicine, underlies the theoretical basis of acupuncture application. However, recent studies have reported that acupuncture stimulation at nonacupoint and acupoint can both evoke similar signal intensity decreases in multiple regions. And these regions were spatially overlapped. We used a machine learning based Support Vector Machine (SVM) approach to elucidate the specific neural response pattern induced by acupuncture stimulation. Group analysis demonstrated that stimulation at two different acupoints (belong to the same nerve segment but different meridians) could elicit distinct neural response patterns. Our findings may provide evidence for acupoint specificity.

This fMRI study used Psychophysiological interaction (PPI) to investigate top-down letter processing with an illusory letter detection task. After an initial training that became increasingly difficult, participant was instructed to detect a letter from pure noise images where there was actually no letter. Such experimental paradigm allowed for isolating top-down components of letter processing and minimizing the influence of bottom-up perceptual input. A distributed cortical network of top-down letter processing was identified by analyzing the functional connectivity patterns of letter-preferential area (LA) within the left fusiform gyrus. Such network extends from the visual cortex to high level cognitive cortexes, including the left middle frontal gyrus, left medial frontal gyrus, left superior parietal gyrus, bilateral precuneus, and left inferior occipital gyrus. These findings suggest that top-down letter processing contains not only regions for processing of letter phonology and appearance, but also those involved in internal information generation and maintenance, and attention and memory processing.

Real-time functional magnetic resonance imaging (rtfMRI) is a new technique which can present (feedback) brain activity during scanning. Through fast acquisition and online analysis of BOLD signal, fMRI data are processed within one TR. Current rtfMRI provides an activation map under specific task mainly through the GLM analysis to select region of interest (ROI). This study was based on independent component analysis (ICA) and used the result of fast ICA analysis to select the node of the functional network as the ROI. Real-time brain activity within the ROI was presented to the subject who needed to find strategies to control his brain activity. The whole real-time processes involved three parts: pre-processing (including head motion correction and smoothing), fast ICA analysis and feedback. In addition, the result of fast head motion correction was also presented to the experimenter in a curve diagram. Based on the above analysis processes, a real time feedback experiment with a motor imagery task was performed. An overt finger movement task as localizer session was adopted for ICA analysis to get the motor network. Supplementary motor area (SMA) in such network was selected as the ROI. During the feedback session, the average of BOLD signals within ROI was presented to the subjects for self-regulation under a motor imagery task. In this experiment, TR was 1.5 seconds, and the whole time of processing and presentation was within 1 second. Experimental results not only showed that the SMA was controllable, but also proved that the analysis method was effective.

Spatial Independent component analysis (sICA) has been successfully used to analyze functional magnetic resonance (fMRI) data. However, the application of ICA was limited in multi-task fMRI data due to the potential spatial dependence between task-related components. Long et al. (2009) proposed ICA with linear projection (ICAp) method and demonstrated its capacity to solve the interaction among task-related components in multi-task fMRI data of single subject. However, it's unclear that how to perform ICAp over a group of subjects. In this study, we proposed a group analysis framework on multi-task fMRI data by combining ICAp with the temporal concatenation method reported by Calhoun (2001). The results of real fMRI experiment containing multiple visual processing tasks demonstrated the feasibility and effectiveness of the group ICAp method. Moreover, compared to the GLM method, the group ICAp method is more sensitive to detect the regions specific to each task.

Motor imagery training, as an effective strategy, has been more and more applied to mental disorders rehabilitation and motor skill learning. Studies on the neural mechanism underlying motor imagery have suggested that such effectiveness may be related to the functional congruence between motor execution and motor imagery. However, as compared to the studies on motor imagery, the studies on motor imagery training are much fewer. The functional alterations associated with motor imagery training and the effectiveness of motor imagery training on motor performance improvement still needs further investigation. Using fMRI, we employed a sequential finger tapping paradigm to explore the functional alterations associated with motor imagery training in both motor execution and motor imagery task. We hypothesized through 14 consecutive days motor imagery training, the motor performance could be improved and the functional congruence between motor execution and motor imagery would be sustained form pre-training phase to post-training phase. Our results confirmed the effectiveness of motor imagery training in improving motor performance and demonstrated in both pre and post-training phases, motor imagery and motor execution consistently sustained the congruence in functional neuroanatomy, including SMA (supplementary motor cortex), PMA (premotor area); M1( primary motor cortex) and cerebellum. Moreover, for both execution and imagery tasks, a similar functional alteration was observed in fusiform through motor imagery training. These findings provided an insight into the effectiveness of motor imagery training and suggested its potential therapeutic value in motor rehabilitation.

In this study we present a texture-based method of emphysema segmentation depicted on CT examination consisting of two steps. Step 1, a fractal dimension based texture feature extraction is used to initially detect base regions of emphysema. A threshold is applied to the texture result image to obtain initial base regions. Step 2, the base regions are evaluated pixel-by-pixel using a method that considers the variance change incurred by adding a pixel to the base in an effort to refine the boundary of the base regions. Visual inspection revealed a reasonable segmentation of the emphysema regions. There was a strong correlation between lung function (FEV1%, FEV1/FVC, and DLCO%) and fraction of emphysema computed using the texture based method, which were -0.433, -.629, and -0.527, respectively. The texture-based method produced more homogeneous emphysematous regions compared to simple thresholding, especially for large bulla, which can appear as speckled regions in the threshold approach. In the texture-based method, single isolated pixels may be considered as emphysema only if neighboring pixels meet certain criteria, which support the idea that single isolated pixels may not be sufficient evidence that emphysema is present. One of the strength of our complex texture-based approach to emphysema segmentation is that it goes beyond existing approaches that typically extract a single or groups texture features and individually analyze the features. We focus on first identifying potential regions of emphysema and then refining the boundary of the detected regions based on texture patterns.

The ability to estimate the volume of pleural effusions is desirable as it can provide information about the severity of the condition and the need for thoracentesis. We present here an improved version of an automated program to measure the volume of pleural effusions using regular chest CT images. First, the lungs are segmented using region growing, mathematical morphology, and anatomical knowledge. The visceral and parietal layers of the pleura are then extracted based on anatomical landmarks, curve fitting and active contour models. The liver and compressed tissues are segmented out using thresholding. The pleural space is then fitted to a Bezier surface which is subsequently projected onto the individual two-dimensional slices. Finally, the volume of the pleural effusion is quantified. Our method was tested on 15 chest CT studies and validated against three separate manual tracings. The Dice coefficients were 0.74±0.07, 0.74±0.08, and 0.75±0.07 respectively, comparable to the variation between two different manual tracings.

Quantitative computed tomography (CT) has been widely used to detect and evaluate the presence (or absence) of emphysema applying the density masks at specific thresholds, e.g., -910 or -950 Hounsfield Unit (HU). However, it has also been observed that subjects with similar density-mask based emphysema scores could have varying lung function, possibly indicating differences of disease severity. To assess this possible discrepancy, we investigated whether density distribution of "viable" lung parenchyma regions with pixel values > -910 HU correlates with lung function. A dataset of 38 subjects, who underwent both pulmonary function testing and CT examinations in a COPD SCCOR study, was assembled. After the lung regions depicted on CT images were automatically segmented by a computerized scheme, we systematically divided the lung parenchyma into different density groups (bins) and computed a number of statistical features (i.e., mean, standard deviation (STD), skewness of the pixel value distributions) in these density bins. We then analyzed the correlations between each feature and lung function. The correlation between diffusion lung capacity (DLCO) and STD of pixel values in the bin of -910HU ≤ PV < -750HU was -0.43, as compared with a correlation of -0.49 obtained between the post-bronchodilator ratio (FEV1/FVC) measured by the forced expiratory volume in 1 second (FEV1) dividing the forced vital capacity (FVC) and the STD of pixel values in the bin of -1024HU ≤ PV < -910HU. The results showed an association between the distribution of pixel values in "viable" lung parenchyma and lung function, which indicates that similar to the conventional density mask method, the pixel value distribution features in "viable" lung parenchyma areas may also provide clinically useful information to improve assessments of lung disease severity as measured by lung functional tests.

Pulmonary ventilation and blood flow are reflected in dynamic chest radiographs as changes in X-ray translucency, i.e., pixel values. This study was performed to investigate the feasibility of ventilation-perfusion (V/Q) study based on the changes in pixel value. Sequential chest radiographs of a patient with ventilation-perfusion mismatch were obtained during respiration using a dynamic flat-panel detector (FPD) system. The lung area was recognized and average pixel value was measured in each area, tracking and deforming the region of interest. Inter-frame differences were then calculated, and the absolute values were summed in each respiratory phase. The results were visualized as ventilation, blood flow, V/Q ratio distribution map and compared to distribution of radioactive counts on ventilation and perfusion scintigrams. In the results, abnormalities were appeared as a reduction of changes in pixel values, and a correlation was observed between the distribution of changes in pixel value and those of radioactivity counts (Ventilation; r=0.78, Perfusion; r=0.77). V/Q mismatch was also indicated as mismatch of changes in pixel value, and a correlation with V/Q calculated by radioactivity counts (r=0.78). These results indicated that the present method is potentially useful for V/Q study as an additional examination in conventional chest radiography.

Obesity has become widespread in America and has been associated as a risk factor for many illnesses. Adipose tissue (AT) content, especially visceral AT (VAT), is an important indicator for risks of many disorders, including heart disease and diabetes. Measuring adipose tissue (AT) with traditional means is often unreliable and inaccurate. CT provides a means to measure AT accurately and consistently. We present a fully automated method to segment and measure abdominal AT in CT. Our method integrates image preprocessing which attempts to correct for image artifacts and inhomogeneities. We use fuzzy cmeans to cluster AT regions and active contour models to separate subcutaneous and visceral AT. We tested our method on 50 abdominal CT scans and evaluated the correlations between several measurements.

Cardiac motion analysis can play an important role in cardiac disease diagnosis. Tagged magnetic resonance imaging (MRI) has the ability to directly and non-invasively alter tissue magnetization and produce tags on the deforming tissue. This paper proposes an approach to analysis of tagged MR images using a multilevel B-splines fitting model incorporating phase information. The novelty of the proposed technique is that phase information is extracted from SinMod.¹ By using real tag intersections extracted directly from tagged MR image data and virtual tag intersections extracted from phase information, both considered to be scattered data, multilevel B-spline fitting can result in accurate displacement motion fields. The B-spline approximation which also serves to remove noise in the displacement measurements is performed without specifying control point locations explicitly and is very fast. Dense virtual tag intersections based on SinMod were created and incorporated into the multilevel B-spline fitting process. Experimental results on simulated data from the 13- parameter kinematic model of Arts et al.² and in vivo canine data demonstrate further improvement in accuracy and effectiveness of the proposed method.

This paper describes a non-linear medical image registration algorithm that aligns lung CT images scanned at different respiratory phases. The method uses landmarks obtained from the airway tree to find the airway branch extension lines and where the lines intersect the lung surface. The branch extension and lung intersection voxels on the surface were the crucial landmarks that initialize the non-rigid registration process. The advantage of these landmarks is that they have high correspondence between the matching patterns in the template images and deformed images. This method was developed and tested on CT examinations from participants in an asthma study. The registration accuracy was evaluated by the average distance between the corresponding airway tree branch points in the pair of images. The mean value of the distance between landmarks in template images and deformed matching images for subjects 1 and 2 were 8.44 mm (±4.46 mm) and 4.33 mm (± 3.78 mm), respectively. The results show that the lung image registration technique developed in this study may prove useful in quantifying longitudinal changes, performing regional analysis, tracking lung tumors, and compensating for subject motion across CT images.

This work presents novel methods to accurately placing landmarks inside the vessel lumen. This task is an important prerequisite to automatic centerline tracing. Methods have been proposed in the past to determine the location of organ landmarks, and yet several challenges remain for vascular landmarks. First, placing landmarks inside the lumen could be challenging for narrow vessels. Second, contrast-enhanced arteries could be tightly surrounded by bones with similar intensity profiles, making detection difficult compared to arteries surrounded only by darker tissues. Third, landmarks not located at bifurcations could be ill-defined as they have high uncertainty in position. We first present a method to detect landmarks that are located at vessel bifurcations. Such landmarks have well-defined positions, and we detect them using machine learning techniques. We then present a method to detect vascular landmarks not located at bifurcations. First, a segment detector is created to detect a vessel segment. Annotating multiple points along a vessel segment is easier than annotating a single landmark position, as there is no well-defined position along a vessel. This resolves the ambiguity issue mentioned above. Second, spatial features are computed from the segment detector's response map, and a regression model is created which takes as input the local spatial features surrounding a voxel, and outputs a confidence score of how likely this voxel is inside the lumen. We evaluate the system on a set of 94 3D CT datasets.

Changes in intraretinal layer thickness occur in a variety of diseases such as glaucoma, macular edema and diabetes. To segment the intraretinal layers from macular spectral-domain OCT (SD-OCT) scans, we previously introduced an automated multiscale 3-D graph search method and validated its performance by computing unsigned border positioning differences when compared with human expert tracings. However, it is also important to study the reproducibility of resulting layer thickness measurements, as layer thickness is a commonly used clinical parameter. In this work, twenty eight (14 x 2) repeated macular OCT volumes were acquired from the right eyes of 14 normal subjects using two Zeiss-Cirrus SD-OCT scanners. After segmentation of 10 intraretinal layers and rigid registration of layer thickness maps from the repeated OCT scans, the thickness difference of each layer was calculated. The overall mean global and regional thickness differences of 10 intraretinal layers were 0.46 ± 0.25 μm (1.70 ± 0.72 %) and 1.16 ± 0.84 μm (4.03 ± 2.05 %), respectively. No specific local region showed a consistent thickness difference across the layers.

Bioluminescence tomography (BLT) is a novel optical molecular imaging (MI) modality. It can reconstruct the inner bioluminescent light source distribution, according to the surface light distribution. The trust region method (TRM) can overcome the ill-posedness of BLT for its regularization property. As there exists a "TRUST" function that can solve the trust region subproblem in Matlab and Matlab's powerful matrix operation ability suited for TRM, the TRM is implemented in Matlab. Then the Matlab code of TRM is transformed into a dynamic linked library (DDL) and mixed together with the C++ code of the adaptive finite element (AFE) framework, using the mixed-language programming technology (MLPT). There are two main advantages of the MLPT. The first is taking advantages of all the participated programming languages. The second is time efficient. The usual way of transferring data between programmes written in different programming languages is to write the data first into files that are stored in the hard discs in one programme, and then read the files from another programme. Besides wasting time on writing and reading, it is difficult to keep the precision of the data. The DLL based MLPT can eliminate the need of installing code compilers in the platform running the software. Furthermore, in DLL, the code is implemented in C/C++ with high time efficiency, while the code in Matlab remains relatively low time efficiency. Finally, a numerical experiment is carried out to show MLPT's usage in the source reconstruction procedure of BLT, using the MLPT based on DLL.

Bone microarchitecture is the predictor of bone quality or bone disease. It can only be measured on a bone biopsy, which is invasive and not available for all clinical situations. Texture analysis on radiographs is a common way to investigate bone microarchitecture. But relationship between three-dimension histomorphometric parameters and two-dimension texture parameters is not always well known, with poor results. The aim of this study is to performed angulated radiographs of the same region of interest and see if a better relationship between texture analysis on several radiographs and histomorphometric parameters can be developed. Computed radiography images of dog (Beagle) mandible section in molar regions were compared with high-resolution micro-CT (Computed-Tomograph) volumes. Four radiographs with 27° angle (up, down, left, right, using Rinn ring and customized arm positioning system) were performed from initial radiograph position. Bone texture parameters were calculated on all images. Texture parameters were also computed from new images obtained by difference between angulated images. Results of fractal values in different trabecular areas give some caracterisation of bone microarchitecture.

With recent developments, digital mammograms can be obtained with a small pixel size, i.e., high resolution; however, the matrix size increases. Therefore, when the image is thinned out, image information is lost when the image is displayed on a liquid crystal display (LCD). To resolve this issue, we have developed a super high resolution liquid crystal display (SHR-LCD) by using a novel resolution enhancement technology for independent subpixel driving (ISD) with three subpixels in each pixel element. However, the lack of image information caused by thinning of the image cannot be ignored because the matrix size of a phase contrast mammogram (PCM) is very large as compared to that of a conventional mammogram. We obtained noise and edge images by using the geometrical layouts of the PCM (7080 x 9480). We measured the Wiener spectrum (WS), modulation transfer function (MTF), and noise-equivalent number of quanta (NEQ) of the images reduced by the nearest-neighbor, bilinear, and bicubic (sharpness and smooth) interpolations. The reduction rate was approximately 0.14. We measured the WS and MTF when the PCM image was displayed on a 5-megapixel (MP) and 15-MP LCD. The bilinear interpolation technique gave the best image quality. The image quality was further improved by using a 15-MP SHR-LCD.

In This study, we used Fractional anisotropy (FA), mean diffusivity (D), parallel diffusivity (D//) and perpendicular diffusivity (D), to localize the regions where occur axonal lesion and demyelization. TBSS was applied to analyze the FA data. After, the regions with alteration were studied with D, D// and D maps. Patients exhibited widespread degradation of FA. With D, D// and D maps analysis we found alterations in corpus callosum, corticospinal tract, fornix, internal capsule, corona radiate, Sagittal stratum, cingulum, fronto-occipital fasciculus and uncinate fasciculus. Our results are consistent with the hypothesis that exist demyelization and axonal damage in patients with TLE.

Performance comparison of functional Magnetic Resonance Imaging (fMRI) software tools is a very difficult task. In this paper, a framework for comparison of fMRI analysis results obtained with different software packages is proposed. An objective evaluation is possible only after pre-processing steps that normalize input data in a standard domain. Segmentation and registration algorithms are implemented in order to classify voxels belonging to brain or not, and to find the non rigid transformation that best aligns the volume under inspection with a standard one. Through the definitions of intersection and union of fuzzy logic an index was defined which quantify information overlap between Statistical Parametrical Maps (SPMs). Direct comparison between fMRI results can only highlight differences. In order to assess the best result, an index that represents the goodness of the activation detection is required. The transformation of the activation map in a standard domain allows the use of a functional Atlas for labeling the active voxels. For each functional area the Activation Weighted Index (AWI) that identifies the mean activation level of whole area was defined. By means of this brief, but comprehensive description, it is easy to find a metric for the objective evaluation of a fMRI analysis tools. Trough the first evaluation method the situations where the SPMs are inconsistent were identified. The result of AWI analysis suggest which tool has higher sensitivity and specificity. The proposed method seems a valid evaluation tool when applied to an adequate number of patients.

A great deal of current literature regarding functional neuroimaging has elucidated the relationships of neurons distributed all over the brain. Modern neuroimaging techniques, such as the functional MRI (fMRI), provide a convenient tool for people to study the correlation among different voxels as well as the spatio-temporal patterns of brain activity. In this study, we present a computational model using radial basis function neural network (RBF-NN) to predict the fMRI voxel activation with the activation of other voxels acquired at the same time. The fMRI data from a visual images stimuli presentation experiment was separated into two sets; one was used to train the model, and the other to validate the accuracy or generalizability of the model. In the visual stimuli presentation experiment, the subject did simple one-back-repetition tasks when four categories of stimuli (houses, faces, cars, and cats) were presented. Voxel sets A and B were selected from fMRI data by two different voxel selection criterion: (1) Voxel set A are those activated for any kind of object stronger than the other three objects in regions of interest (ROIs) without correction (P=0.001); (2) Voxel set B are those activated for at least one of the categories of stimuli within the ROIs (FWE correction, P=0.05). RBF-NN regression models construct the nonlinear relationship between the activation of voxels in A and B. Our test results showed that RBF-NN can capture the nonlinear relationship existing in neurons and reveal the relationship between voxel's activation from different brain regions.

Functional magnetic resonance imaging (fMRI) relies on detecting small changes in signal during brain activities, in presence of various noise, including those caused by respiration and cardiac pulsation. In the resting state, there is no explicit task event except the baseline neuroactivities of awakeness and other unknowns. However, the resting state is accompanied with the cardiac and respiration pulsations, which are the explicit non-neuronal physiological sources of fMRI signals. By recording the respiration and cardiac waveforms in synchrony with the fMRI scanning, we may estimate the physiological modulation artifacts in the fMRI dataset by the temporal correlations between the waveforms and the fMRI signal. In this work, we demonstrate that the respiration and cardiac modulation effects on the magnitude and phase components of the complex fMRI signal, including temporal correlation and time latency. In particular, our results show that: 1) the fMRI phase is slightly more modulated by the physiological modulations than its magnitude counterpart; 2) the fMRI signal (both magnitude and phase) shows 1 to 2s latency to respiration stimulus, and 0 to 1s latency to cardiac stimulus. For physiological artifact removal, we compare the band-stop filtering method with the RETROICOR method and find the former can remove the physiological modulations in a stable and consistent manner in frequency domain (stopping the signature frequencies irrespective of asynchrony.

N170 is an important neurophysiological index to study the underlying mechanisms of face and object perception. In this study, we proposed a mean-sensitive spatial filtering (MSF) method for linear transformation of event-related potentials (ERP) that is sensitive to mean differences between stimuli conditions and applied it to N170 component to extract category-specific spatio-temporal features contained in EEG. MSF method estimated a set of optimal projecting vectors according to the spatial distribution patterns of N170 means. Then, we applied these spatial filters to single-trial ERP data and perform classification on the extracted features. In this way, the presence of a larger negative component in EEG time courses evoked by faces can be detected robustly in single trial EEG, and hereby we can infer the category of every presented stimulus from faces and objects. Furthermore, we also successfully extracted the unobvious distinct spatial patterns between cars and cats with MSF and separated them correctly. Our remarkable and robust classification performances suggest that MSF works well in extracting stable spatial patterns from N170. Therefore, MSF provides a promising solution for decoding presented visual information basing on single-trial N170 component.

The perfusion of the brain is essential to maintain brain function. Stroke is an example of a decrease in blood flow and reduced perfusion. During ischemic stroke the blood flow to tissue is hampered due to a clot inside a vessel. To investigate the recovery of stroke patients, follow up studies are necessary. MRI is the preferred imaging modality for follow up because of the absence of radiation dose concerns, contrary to CT. Dynamic Susceptibility Contrast (DSC) MRI is an imaging technique used for measuring perfusion of the brain, however, is not standard applied in the clinical routine due to lack of immediate patient benefit. Several post processing algorithms are described in the literature to obtain cerebral blood flow (CBF). The quantification of CBF relies on the deconvolution of a tracer concentration-time curve in an arterial and a tissue voxel. There are several methods to obtain this deconvolution based on singular-value decomposition (SVD). This contribution describes a comparison between the different approaches as currently there is no best practice for (all) clinical relevant situations. We investigate the influence of tracer delay, dispersion and recirculation on the performance of the methods. In the presence of negative delays, the truncated SVD approach overestimates the CBF. Block-circulant and reformulated SVD are delay-independent. Due to its delay dependent behavior, the truncated SVD approach performs worse in the presence of dispersion as well. However all SVD approaches are dependent on the amount of dispersion. Moreover, we observe that the optimal truncation parameter varies when recirculation is added to noisy data, suggesting that, in practice, these methods are not immune to tracer recirculation. Finally, applying the methods to clinical data resulted in a large variability of the CBF estimates. Block-circulant SVD will work in all situations and is the method with the highest potential.

MRI velocimetry (also known as phase-contrast MRI) is a powerful tool for quantification of cerebrospinal fluid (CSF) flow in various regions of the brain and craniospinal junction and has been accepted as a diagnostic tool to assist with the diagnosis of certain conditions such as hydrocephalus and chiari malformations. Cerebrospinal fluid is continually produced in the ventricles of the brain, flows through the ventricular system and then out and around the brain and spinal cord and is reabsorbed over the convexity of the brain. Any disease process which either impedes the normal pattern of flow or restricts the area where flow occurs can change the pattern of these waveforms with the direction and velocity of flow being determined by the pressure transmitted from the pulsation of the heart and circulation of blood within the central nervous system. Therefore, we hypothesized that phase-contrast MRI could eventually be used as a diagnostic aid in determining the degree of spinal cord compression following injury to the cervical or thoracic spine. In this study, we examined CSF flow in 3 normal subjects and 2 subjects with non-acute injuries in the cervical spine using Cine phasecontrast MRI. CSF flow analysis was performed using an in-house developed software. The flow waveform was calculated in both normal subjects (n=3) as well as subjects with spinal cord injury in the cervical spine (n=2). The bulk flow at C2 was measured to be 0.30 +/- 0.05 cc, at 5 cm distal to C2, it was 0.19+/- 0.07 cc, and at 10 cm distal to C2, it was 0.17+/- 0.05 cc. These results were in good agreement with previously published results. In patients with spinal cord injury, at the site of injury in the cervical spine, bulk flow was found to be 0.08 +/- 0.12 cc, at 5 cm proximal to the site of injury it was found to be 0.18 +/- 0.07 cc, and at 5 cm distal to the site of injury, it was found to be 0.12 +/- 0.01 cc.

Gray matter volume and cortical thickness are two indices of concern in brain structure magnetic resonance imaging research. Gray matter volume reflects mixed-measurement information of cerebral cortex, while cortical thickness reflects only the information of distance between inner surface and outer surface of cerebral cortex. Using Scaled Subprofile Modeling based on Principal Component Analysis (SSM_PCA) and Pearson's Correlation Analysis, this study further provided quantitative comparisons and depicted both global relevance and local relevance to comprehensively investigate morphometrical abnormalities in cerebral cortex in Alzheimer's disease (AD). Thirteen patients with AD and thirteen age- and gender-matched healthy controls were included in this study. Results showed that factor scores from the first 8 principal components accounted for ~53.38% of the total variance for gray matter volume, and ~50.18% for cortical thickness. Factor scores from the fifth principal component showed significant correlation. In addition, gray matter voxel-based volume was closely related to cortical thickness alterations in most cortical cortex, especially, in some typical abnormal brain regions such as insula and the parahippocampal gyrus in AD. These findings suggest that these two measurements are effective indices for understanding the neuropathology in AD. Studies using both gray matter volume and cortical thickness can separate the causes of the discrepancy, provide complementary information and carry out a comprehensive description of the morphological changes of brain structure.

Temporal lobe epilepsy (TLE) is one of the most common epilepsy syndromes with focal seizures generated in the left or right temporal lobes. With the magnetic resonance imaging (MRI), many evidences have demonstrated that the abnormalities in hippocampal volume and the distributed atrophies in cortical cortex. However, few studies have investigated if TLE patients have the alternation in the structural networks. In the present study, we used the cortical thickness to establish the morphological connectivity networks, and investigated the network properties using the graph theoretical methods. We found that all the morphological networks exhibited the small-world efficiency in left TLE, right TLE and normal groups. And the betweenness centrality analysis revealed that there were statistical inter-group differences in the right uncus region. Since the right uncus located at the right temporal lobe, these preliminary evidences may suggest that there are topological alternations of the cortical anatomical networks in TLE, especially for the right TLE.

Hippocampal sclerosis (HS) is the most common damage seen in the patients with temporal lobe epilepsy (TLE). In the present study, the hippocampal-cortical connectivity was defined as the correlation between the hippocampal volume and cortical thickness at each vertex throughout the whole brain. We aimed to investigate the differences of ipsilateral hippocampal-cortical connectivity between the unilateral TLE-HS patients and the normal controls. In our study, the bilateral hippocampal volumes were first measured in each subject, and we found that the ipsilateral hippocampal volume significantly decreased in the left TLE-HS patients. Then, group analysis showed significant thinner average cortical thickness of the whole brain in the left TLE-HS patients compared with the normal controls. We found significantly increased ipsilateral hippocampal-cortical connectivity in the bilateral superior temporal gyrus, the right cingulate gyrus and the left parahippocampal gyrus of the left TLE-HS patients, which indicated structural vulnerability related to the hippocampus atrophy in the patient group. However, for the right TLE-HS patients, no significant differences were found between the patients and the normal controls, regardless of the ipsilateral hippocampal volume, the average cortical thickness or the patterns of hippocampal-cortical connectivity, which might be related to less atrophies observed in the MRI scans. Our study provided more evidence for the structural abnormalities in the unilateral TLE-HS patients.

Magnetic particle imaging (MPI) has emerged as a new imaging modality that uses the nonlinear magnetization behavior of superparamagnetic particles. Due to the need to avoid contamination of particle signals with the simultaneous excitation signal, MPI transmit systems require different design considerations from those in MRI, where excitation and detection are temporally decoupled. Specifically, higher order harmonic distortion in the transmit spectrum can feed through to and contaminate the received signal spectrum. In a prototype MPI scanner, this distortion needs to be attenuated by 90 dB at all frequencies. In this paper, we describe two methods of filtering out harmonic distortion in the transmit spectrum. The first method uses a Butterworth topology while the second a cascaded Butterworth-elliptic topology. We show that whereas the Butterworth filter alone achieves around 16 and 32 dB attenuation at the second and third harmonics, the cascaded filter can achieve around 65 and 73 dB at these harmonics. Finally, we discuss how notch placement in the stopband can also be applied to design highpass filters for MPI detection systems.

Magnetic particle imaging (MPI) is a new medical imaging modality that maps the instantaneous response of superparamagnetic particles under an applied magnetic field. In MPI, the excitation and detection of the nanoparticles occur simultaneously. Therefore, when a sinusoidal excitation field is applied to the system, the received signal spectrum contains both harmonics from the particles and a direct feedthrough signal from the source at the fundamental drive frequency. Removal of the induced feedthrough signal from the received signal requires significant filtering, which also removes part of the signal spectrum. In this paper, we present a method to investigate the impact of temporally filtering out individual lower order harmonics on the reconstructed x-space image. Analytic and simulation results show that the loss of particle signal at low frequency leads to a recoverable loss of low spatial frequency information in the x-space image. Initial experiments validate the findings and demonstrate the feasibility of the recovery of the lost signal. This builds on earlier work that discusses the ideal one-dimensional MPI system and harmonic decomposition of the MPI signal.

Magnetic particle imaging (MPI) using the nonlinear interaction between internally administered magnetic nanoparticles (MNPs) and electromagnetic waves irradiated from outside of the body has attracted attention for the early diagnosis of diseases such as cancer. In MPI, the local magnetic field distribution is scanned, and the magnetization signal from MNPs inside an object region is detected. However, the signal sensitivity and image resolution are degraded by interference from the magnetization signal generated by MNPs that exist outside of the desired region, owing to nonlinear responses. Earlier, we proposed an image reconstruction method for suppressing the interference component while emphasizing the signal component using the property of the higher harmonic components generated by the MNPs. However, edge areas in the reconstructed image were emphasized excessively owing to the high-pass-filter effect of this method. Here, we propose a new method based on correlation information between the observed signal and a system function. We performed a numerical analysis and found that, although the image was somewhat blurred, the detection sensitivity can clearly be improved without the inverse-matrix operation used in conventional image reconstruction.

Magnetic Particle Imaging (MPI) is a new imaging modality promising high sensitivity and high-resolution imaging of ultra-small superparamagnetic iron oxide (USPIO) nanoparticles.1 A new mathematical theory for MPI based on x-space was recently developed that indicates the spatial resolution of MPI improves with the cube of the USPIO iron core diameter.2 A system that can accurately measure the USPIO point spread function and relaxation time constants would enable MPI researchers to decouple magnetic particle development and imaging system development. This system would enable magnetic nanoparticle manufacturers and MPI researchers to measure the intrinsic spatial resolution of the USPIOs to be used in the imaging system without an imager. Therefore, we have developed a magnetic nanoparticle relaxometer that estimates nanoparticle diameter and relaxation constant using a modified form of the x-space theory of MPI. Fitting of the measured signal to the theoretical signal uses nonnegative least squares with an optimal Tikhonov regularization fitting scheme. The technique estimates magnetic nanoparticle diameter, relaxation time constants from the nanoparticle signal. Our measurements have excellent sensitivity and change little with independent, repeated measurements. While more experiments are necessary, our data lends the first experimental evidence to support the cubic dependence of spatial resolution on magnetic nanoparticle diameter.

Magnetic particle imaging (MPI) is an emerging medical imaging modality that is predicted to have improved sensitivity and contrast as compared to existing technologies. MPI uses a strong magnetic field gradient (6.5 T/m) to spatially localize the induction response of ultra-small superparamagnetic iron oxide nanoparticles (USPIOs), which are currently approved as a contrast agent for MRI. This new imaging modality has excellent contrast and will be safe for human use. MPI relies on USPIO dipole moments aligning quickly with the applied magnetic field, but Neel and Brownian relaxation mechanisms can significantly retard this alignment. By causing this lag in magnetization alignment, relaxation ultimately degrades the resolution and accuracy of the MPI method. Our early simulation results indicate that relaxation effects in larger USPIOs could degrade spatial resolution by a factor of two or more. Our goal here is to develop a rigorous and predictive mathematical model for the imaging process including relaxation effects. The x-space formulation of MPI previously developed by our group details the theoretical signal, bandwidth, resolution, SNR, and SAR of MPI; however this theory was formulated assuming negligible relaxation times. Here we updated the x-space analysis of MPI to include relaxation effects and have demonstrated that this inclusion is essential for excellent agreement with experimental MPI data. We have also shown that relaxation degrades image resolution and accuracy, necessitating an improved understanding of relaxation for future mitigation of these consequences through careful USPIO and MPI system optimization.

The detection of tumors in soft tissues, such as breast cancer, is important to achieve at the earliest stages of the disease to improve patient outcome. Tumors often exhibit a greater elastic modulus compared to normal tissues. In this paper, we report our first study to measure elastic properties of soft tissues by mapping the surface acoustic waves (SAWs) with image plane digital holography. The experimental results show that the SAW velocity is proportional to the square root of elastic modulus over a range from 3.7-122kPa in homogeneous tissue phantoms, consistent with Rayleigh wave theory. This technique also permits detection of the interface of two-layer phantoms 10mm deep under surface and the interface depth by quantifying the SAW dispersion.