Show all abstracts
View Session
- Front Matter: Volume 6920
- Beamforming/Image Formation I
- Beamforming/Image Formation II
- Frontiers in Ultrasound
- Cardiovascular
- Image Processing
- Tissue Characterization
- Elastography
- Poster Session
Front Matter: Volume 6920
Front Matter: Volume 6920
Show abstract
This PDF file contains the front matter associated with SPIE Proceedings Volume 6920, including the Title Page, Copyright information, Table of Contents, Introduction (if any), and the Conference Committee listing.
Beamforming/Image Formation I
Beamformer enhancement by post-processing for improved spatial resolution and signal-to-noise ratio
Show abstract
Various methods are used in ultrasound beamforming to increase signal-to-noise ratio (SNR) and improve
spatial resolution. SNR is typically improved by exploiting coherence in the RF channel data, for example summing
channel data after applying focal delays in the delay-and-sum (DAS) beamformer, and summing channel data after
applying a per-channel matched filter for the spatial matched filter beamformer[1]. Inverse filter methods are
capable of improving spatial resolution at the cost of SNR [2],[3], or can trade resolution for SNR using a
regularization parameter, but in general are very computationally intensive due to the large RF data sets used. We
propose a post-processing method operating on post-summed but pre-envelope detected beamformed image data
that can improve the pixel SNR and spatial resolution of any beamformer with low computational cost. This is
achieved by forming a new pixel for each point in the image as a linear combination of the surrounding
beamformed pixels. The weights for each pixel are calculated in advance using a quadratically constrained least
squares method to reduce PSF energy outside the mainlobe and noise energy. Simulations indicate that this method
can increase cystic contrast by up to 20dB without any cost in SNR, and can increase pixel SNR can by up 16dB
without affecting contrast. Alternatively, simultaneous gains in contrast and SNR can be achieved. Experimental
results show smaller performance improvements yet validate the feasibility of this technique.
Medical ultrasound digital beamforming on a massively parallel processing array platform
Show abstract
Digital beamforming has been widely used in modern medical ultrasound instruments. Flexibility is the key advantage
of a digital beamformer over the traditional analog approach. Unlike analog delay lines, digital delay can be
programmed to implement new ways of beam shaping and beam steering without hardware modification. Digital
beamformers can also be focused dynamically by tracking the depth and focusing the receive beam as the depth
increases. By constantly updating an element weight table, a digital beamformer can dynamically increase aperture size
with depth to maintain constant lateral resolution and reduce sidelobe noise. Because ultrasound digital beamformers
have high I/O bandwidth and processing requirements, traditionally they have been implemented using ASICs or
FPGAs that are costly both in time and in money.
This paper introduces a sample implementation of a digital beamformer that is programmed in software on a Massively
Parallel Processor Array (MPPA). The system consists of a host PC and a PCI Express-based beamformer accelerator
with an Ambric Am2045 MPPA chip and 512 Mbytes of external memory. The Am2045 has 336 asynchronous RISCDSP
processors that communicate through a configurable structure of channels, using a self-synchronizing
communication protocol.
Real-time implementation of the echo signal processing and digital scan conversion for medical ultrasound imaging with a single TMS320C6416 DSP
Show abstract
Software implementation of a medical ultrasound imaging system using commercial DSPs (Digital Signal Processor) has
advantages over FPGA- or ASIC-based system in development cost and time. The authors have developed a full
software-based ultrasound scanner consisting of a typical analog front-end block and a DSP system. In this work, we
present efficient methods for software realization of an echo processor to perform all the ultrasound signal processing
functions following the receive beamforming. For implementation with a single TMS320C6416 DSP, the most
computationally demanding functions such as dynamic filtering, quadrature demodulation, decimation, magnitude
calculation, and log compression are implemented using modified algorithms and structures optimized to best match the
DSP architecture for fast computation. The DSC (digital scan converter) is realized with an LUT for generating memory
addresses and interpolation coefficients for each display point. The LUT table is stored in a single external SDRAM so
that the internal DSP memory can be fully utilized by the DSP core to maximize the processing speed. The possible
memory stall that can be caused by the external memory access is removed by properly employing the enhanced direct
memory access channels. Experimental results show that the proposed implementation can support up to 4 kHz PRF
(pulse repetition frequency) when the input data rate is 40 MHz.
Projection-reflection ultrasound images using PE-CMOS sensor: a preliminary bone fracture study
Show abstract
In this study, we investigated the characteristics of the ultrasound reflective image obtained by a CMOS sensor
array coated with piezoelectric material (PE-CMOS). The laboratory projection-reflection ultrasound prototype
consists of five major components: an unfocused ultrasound transducer, an acoustic beam splitter, an acoustic
compound lens, a PE-CMOS ultrasound sensing array (Model I400, Imperium Inc. Silver Spring, MD), and a
readout circuit system. The prototype can image strong reflective materials such as bone and metal. We found
this projection-reflection ultrasound prototype is able to reveal hairline bone fractures with and without intact skin
and tissue. When compared, the image generated from a conventional B-scan ultrasound on the same bone
fracture is less observable. When it is observable with the B-scan system, the fracture or crack on the surface
only show one single spot of echo due to its scan geometry. The corresponding image produced from the
projection-reflection ultrasound system shows a bright blooming strip on the image clearly indicating the fracture
on the surface of the solid material. Speckles of the bone structure are also observed in the new ultrasound
prototype. A theoretical analysis is provided to link the signals as well as speckles detected in both systems.
Beamforming/Image Formation II
An ideal observer approach to beamforming
Show abstract
Many signal processing issues in ultrasonic imaging can be viewed as attempts to focus signal energy while preserving
the diagnostic information they contain. We have been developing a task-based ideal-observer approach to signal
processing with the goal of better understanding the factors that influence the transfer of diagnostic information and
improving signal processing algorithms for optimal transfer. We treat the scattering medium as a Gaussian random field
with non-uniform variance that encodes the properties necessary for accurate task performance. Using measured pointspread
functions for a given system we propagate the scattering statistics through to various stages of the acquisition
process along with acquisition noise. In this work we focus on the role of beamforming in this process. We consider the
efficiency of information transfer by analyzing the ideal observer acting on individual receive elements and then
considering different strategies for preserving diagnostic information in beamformed echo signals. Optimal
beamforming strategies suggested by the analysis are approximated by applying spatial filtering techniques to fixedfocus
echo data. These results are compared with a standard delay-and-sum beamformer.
Globally optimized Fourier finite-difference method for ultrasound breast imaging
Show abstract
Ultrasound reflection imaging is a promising imaging modality for detecting small, early-stage breast cancers. Properly
accounting for ultrasound scattering from heterogeneities within the breast is essential for high-resolution and high-quality
ultrasound breast imaging. We develop a globally optimized Fourier finite-difference method for ultrasound reflectivity
image reconstruction. It utilizes an optimized solution of acoustic-wave equation and a heterogeneous sound-speed distribution
of the breast obtained from tomography to reconstruct ultrasound reflectivity images. The method contains a
finite-difference term in addition to the split-step Fourier implementation, and minimizes ultrasound phase errors during
wavefield inward continuation while maintaining the advantage of high computational efficiency. The accuracy analysis indicates
that the optimized method is much more accurate than the split-step Fourier method. The computational efficiency
of the optimized method is one to two orders of magnitude faster than time-reversal imaging using a finite-difference
time-domain wave-equation scheme. Our new optimized method can accurately handle ultrasound scattering from breast
heterogeneities during reflectivity image reconstruction. Our numerical imaging examples demonstrate that the optimized
method has the potential to produce high-quality and high-resolution ultrasound reflectivity images in combination with a
reliable ultrasound sound-speed tomography method.
Discrete echo signal modeling of ultrasound imaging systems
Show abstract
In this paper, a discrete model representing the pulse-tissue interaction in the medical ultrasound scanning and
imaging process is developed. The model is based on discretizing the acoustical wave equation and is in terms
of convolution between the input ultrasound pulses and the tissue mass density variation. Such a model can
provide a useful means for ultrasound echo signal processing and imaging.
Most existing models used for ultrasound imaging are based on frequency domain transform. A disadvantage
of the frequency domain transform is that it is only applicable to shift-invariant models. Thus it has ignored the
shift-variant nature of the original acoustic wave equation where the tissue compressibility and mass density distributions
are spatial-variant factors. The discretized frequency domain model also obscures the compressibility
and mass density representations of the tissue, which may mislead the physical understanding and interpretation
of the image obtained. Moreover, only the classical frequency domain filtering methods have been applied to the
frequency domain model for acquiring some tissue information from the scattered echo signals. These methods
are non-parametric and require a prior knowledge of frequency spectra of the transmitted pulses.
Our proposed model technique will lead to discrete, multidimensional, shift-variant and parametric difference or convolution equations with the transmitted pulse pressure as the input, the measurement data of the echo signals as the output, and functions of the tissue compressibility and mass density distributions as shift-variant parameters that can be readily identified from input-output measurements. The proposed model represents the entire multiple scattering process, and hence overcomes the key limitation in the current ultrasound imaging methods.
Clinical breast imaging using sound-speed reconstructions of ultrasound tomography data
Show abstract
To improve clinical breast imaging, a new ultrasound tomography imaging device (CURE) has been built at the
Karmanos Cancer Institute. The ring array of the CURE device records ultrasound transmitted and reflected ultrasound
signals simultaneously. We develop a bent-ray tomography algorithm for reconstructing the sound-speed distribution of
the breast using time-of-flights of transmitted signals. We study the capability of the algorithm using a breast phantom
dataset and over 190 patients' data. Examples are presented to demonstrate the sound-speed reconstructions for different
breast types from fatty to dense on the BI-RADS categories 1-4. Our reconstructions show that the mean sound-speed
value increases from fatty to dense breasts: 1440.8 m/ s (fatty), 1451.9 m/ s (scattered), 1473.2 m/ s(heterogeneous), and 1505.25 m/ s (dense). This is an important clinical implication of our reconstruction. The mean
sound speed can be used for breast density analysis. In addition, the sound-speed reconstruction, in combination with
attenuation and reflectivity images, has the potential to improve breast-cancer diagnostic imaging. The breast is not
compressed and does not move during the ultrasound scan using the CURE device, stacking 2D slices of ultrasound
sound-speed tomography images forms a 3D volumetric view of the whole breast. The 3D image can also be projected
into a 2-D "ultrasound mammogram" to visually mimic X-ray mammogram without breast compression and ionizing
radiation.
3D PSF analysis for arbitrary transducer geometries and SAFT-based image reconstruction
Show abstract
The point spread function (PSF) of an imaging system may be used as measure for the imaging quality. The PSF usually depends on position and an several other system parameters. Our current 3D imaging system for ultrasound computer tomography consists of a rotatable cylinder with approx. 2000 ultrasound transducers. 3D images are reconstructed by means of synthetic aperture focusing technique (SAFT) using all available emitter-receiver-combinations. No analytical solution exists for determining the spatially varying PSF for arbitrary placement of the transducers.
This work derives a new numerical approach for the approximation of the 3D PSF for arbitrary transducer geometries including the beam pattern of the ultrasound transducers, a directional point scatterer model, damping of the breast and arbitrary pulse shapes.
As an exemplary application the spatially varying 3D PSF of the current cylindrical geometry is analyzed under idealized conditions (point sources, no damping, and isotropic scattering) and compared to non-idealized results of the PSF analysis. The results show the necessity to take the system specific parameters into account for a realistic
prognosis of 3D imaging performance.
Frontiers in Ultrasound
Implantable ultrasound devices
Show abstract
Using medical implants to wirelessly report physiological data is a technique that is rapidly growing. Ultrasound is
well-suited for implants -- it requires little power and this form of radiated energy has no ill effects on the body. We
report here on techniques we have developed in our experience gained in implanting over a dozen Doppler ultrasound flow-measuring implants in dogs. The goal of our implantable device is to measure flow in an arterial graft. To accomplish this, we place a Doppler transducer in the wall of a graft and an implant unit under the skin that energizes the 20 MHz Doppler transducer system, either when started by external command or by internal timetable. The implant records the digitized Doppler real and imaginary channels and transmits the data to a nearby portable computer for storage and evaluation. After outlining the overall operation of the system, we will concentrate on three areas of implant design where special techniques are required: ensuring safety, including biocompatibility to prevent the body from reacting to its invasion; powering the device, including minimizing energy used so that a small battery can provide long-life; and transmitting the data obtained.
Cardiovascular
Cardiac phase detection in intravascular ultrasound images
Show abstract
Image gating is related to image modalities that involve quasi-periodic moving organs. Therefore, during intravascular
ultrasound (IVUS) examination, there is cardiac movement interference. In this paper, we aim to obtain IVUS gated
images based on the images themselves. This would allow the reconstruction of 3D coronaries with temporal accuracy
for any cardiac phase, which is an advantage over the ECG-gated acquisition that shows a single one. It is also important
for retrospective studies, as in existing IVUS databases there are no additional reference signals (ECG). From the
images, we calculated signals based on average intensity (AI), and, from consecutive frames, average intensity difference
(AID), cross-correlation coefficient (CC) and mutual information (MI). The process includes a wavelet-based filter step
and ascendant zero-cross detection in order to obtain the phase information. Firstly, we tested 90 simulated sequences
with 1025 frames each. Our method was able to achieve more than 95.0% of true positives and less than 2.3% of false
positives ratio, for all signals. Afterwards, we tested in a real examination, with 897 frames and ECG as gold-standard.
We achieved 97.4% of true positives (CC and MI), and 2.5% of false positives. For future works, methodology should be
tested in wider range of IVUS examinations.
Evaluation of a level set segmentation method for cardiac ultrasound images
Show abstract
This paper evaluates the quality of segmentation achieved by a level set evolution strategy call Tunneling Descent. Level sets often evolve and become stationary at the nearest local minimum of an energy function. A
comparison of the local level set minimum with a global minimum is important for many applications. This is
especially true of ultrasound segmentation where ultrasound speckle can introduce many local minima which trap
the level set. In this paper, we compare the quality of the level set segmentation with the quality of segmentation
achieved by (1) simulated annealing (with three different cooling schedules), and (2) random sampling, and (3)
three experts (manual segmentation). Simulated annealing and random sampling offer global minimization.
In this paper, the quality of the segmentation is compared for 21 clinically-obtained images. The comparison
is carried out using two performance measures: the amount by which the global minimizers can further decrease
the level set energy, and the contour distance between the segmentations themselves. The results show that level
set segmentation is within one ultrasound resolution cell of the global minimum. The results also show that the
level set segmentation is quite close to manual segmentation.
Detection of artery interfaces: a real-time system and its clinical applications
Show abstract
Analyzing the artery mechanics is a crucial issue because of its close relationship with several cardiovascular risk
factors, such as hypertension and diabetes. Moreover, most of the work can be carried out by analyzing image sequences
obtained with ultrasounds, that is with a non-invasive technique which allows a real-time visualization of the observed
structures. For this reason, therefore, an accurate temporal localization of the main vessel interfaces becomes a central
task for which the manual approach should be avoided since such a method is rather unreliable and time consuming.
Real-time automatic systems are advantageously used to automatically locate the arterial interfaces. The automatic
measurement reduces the inter/intra-observer variability with respect to the manual measurement which unavoidably
depends on the experience of the operator. The real-time visual feedback, moreover, guides physicians when looking for
the best position of the ultrasound probe, thus increasing the global robustness of the system. The automatic system
which we developed is a stand-alone video processing system which acquires the analog video signal from the
ultrasound equipment, performs all the measurements and shows the results in real-time. The localization algorithm of
the artery tunics is based on a new mathematical operator (the first order absolute moment) and on a pattern recognition
approach. Various clinical applications have been developed on board and validated through a comparison with gold-standard
techniques: the assessment of intima-media thickness, the arterial distension, the flow-mediated dilation and
the pulse wave velocity. With this paper, the results obtained on clinical trials are presented.
A new fringeline-tracking approach for color Doppler ultrasound imaging phase unwrapping
Show abstract
Color Doppler ultrasound imaging is a powerful non-invasive diagnostic tool for many clinical applications that involve
examining the anatomy and hemodynamics of human blood vessels. These clinical applications include cardio-vascular diseases, obstetrics, and abdominal diseases. Since its commercial introduction in the early eighties, color Doppler ultrasound imaging has been used mainly as a qualitative tool with very little attempts to quantify its images. Many imaging artifacts hinder the quantification of the color Doppler images, the most important of which is the aliasing
artifact that distorts the blood flow velocities measured by the color Doppler technique. In this work we will address the
color Doppler aliasing problem and present a recovery methodology for the true flow velocities from the aliased ones. The problem is formulated as a 2D phase-unwrapping problem, which is a well-defined problem with solid theoretical foundations for other imaging domains, including synthetic aperture radar and magnetic resonance imaging. This paper documents the need for a phase unwrapping algorithm for use in color Doppler ultrasound image analysis. It describes a new phase-unwrapping algorithm that relies on the recently developed cutline detection approaches. The algorithm is novel in its use of heuristic information provided by the ultrasound imaging modality to guide the phase unwrapping process. Experiments have been performed on both in-vitro flow-phantom data and in-vivo human blood flow data. Both data types were acquired under a controlled acquisition protocol developed to minimize the distortion of the color Doppler data and hence to simplify the phase-unwrapping task. In addition to the qualitative assessment of the results, a quantitative assessment approach was developed to measure the success of the results. The results of our new algorithm have been compared on ultrasound data to those from other well-known algorithms, and it outperforms all of them.
Automatic detection of blood versus non-blood regions on intravascular ultrasound (IVUS) images using wavelet packet signatures
Show abstract
Intravascular ultrasound (IVUS) has been proven a reliable imaging modality that is widely employed in cardiac
interventional procedures. It can provide morphologic as well as pathologic information on the occluded plaques in the
coronary arteries. In this paper, we present a new technique using wavelet packet analysis that differentiates between
blood and non-blood regions on the IVUS images. We utilized the multi-channel texture segmentation algorithm based
on the discrete wavelet packet frames (DWPF). A k-mean clustering algorithm was deployed to partition the extracted
textural features into blood and non-blood in an unsupervised fashion. Finally, the geometric and statistical information
of the segmented regions was used to estimate the closest set of pixels to the lumen border and a spline curve was fitted
to the set. The presented algorithm may be helpful in delineating the lumen border automatically and more reliably prior
to the process of plaque characterization, especially with 40 MHz transducers, where appearance of the red blood cells
renders the border detection more challenging, even manually. Experimental results are shown and they are
quantitatively compared with manually traced borders by an expert. It is concluded that our two dimensional (2-D)
algorithm, which is independent of the cardiac and catheter motions performs well in both in-vivo and in-vitro cases.
Image Processing
Adaptive spatial compounding for improving ultrasound images of the epidural space on human subjects
Show abstract
Administering epidural anesthesia can be a difficult procedure, especially for inexperienced physicians. The use of ultrasound
imaging can help by showing the location of the key surrounding structures: the ligamentum flavum and the lamina
of the vertebrae. The anatomical depiction of the interface between ligamentum flavum and epidural space is currently
limited by speckle and anisotropic reflection. Previous work on phantoms showed that adaptive spatial compounding with
non-rigid registration can improve the depiction of these features. This paper describes the development of an updated
compounding algorithm and results from a clinical study. Average-based compounding may obscure anisotropic reflectors
that only appear at certain beam angles, so a new median-based compounding technique is developed. In order to
reduce the computational cost of the registration process, a linear prediction algorithm is used to reduce the search space
for registration. The algorithms are tested on 20 human subjects. Comparisons are made among the reference image plus
combinations of different compounding methods, warping and linear prediction. The gradient of the bone surfaces, the
Laplacian of the ligamentum flavum, and the SNR and CNR are used to quantitatively assess the visibility of the features
in the processed images. The results show a significant improvement in quality when median-based compounding
with warping is used to align the set of beam-steered images and combine them. The improvement of the features makes
detection of the epidural space easier.
Measuring shape complexity of breast lesions on ultrasound images
Show abstract
The shapes of malignant breast tumors are more complex than the benign lesions due to their nature of infiltration into
surrounding tissues. We investigated the efficacy of shape features and presented a method using polygon shape
complexity to improve the discrimination of benign and malignant breast lesions on ultrasound. First, 63 lesions (32
benign and 31 malignant) were segmented by K-way normalized cut with the priori rules on the ultrasound images.
Then, the shape measures were computed from the automatically extracted lesion contours. A polygon shape complexity
measure (SCM) was introduced to characterize the complexity of breast lesion contour, which was calculated from the
polygonal model of lesion contour. Three new statistical parameters were derived from the local integral invariant
signatures to quantify the local property of the lesion contour. Receiver operating characteristic (ROC) analysis was
carried on to evaluate the performance of each individual shape feature. SCM outperformed the other shape measures,
the area under ROC curve (AUC) of SCM was 0.91, and the sensitivity of SCM could reach 0.97 with the specificity
0.66. The measures of shape feature and margin feature were combined in a linear discriminant classifier. The
resubstitution and leave-one-out AUC of the linear discriminant classifier were 0.94 and 0.92, respectively. The
distinguishing ability of SCM showed that it could be a useful index for the clinical diagnosis and computer-aided
diagnosis to reduce the number of unnecessary biopsies.
Image-based method for in-vivo freehand ultrasound calibration
Show abstract
For freehand ultrasound systems, a calibration method is necessary to locate the position and orientation of a
2D B-mode ultrasound image plane with respect to a position sensor attached to the transducer. In addition,
the acquisition time discrepancy between the position measurements and the image frames has the be computed.
We developed a new method that adresses both of these problems, based on the fact that a freehand ultrasound
system establishes consistent 3D data of an arbitrary object. Two angulated sweeps of any object containing
well visible structures are recorded, each at a different orientation. A non-linear optimization strategy maximizes
the similarity of 2D ultrasound images from one sweep to reconstructions computed from the other sweep. No
designated phantom is required for this calibration. The process can be performed in vivo on the patient. We
evaluated our method using freehand acquisitions on both a phantom and the human liver. The accuracy of the
approach was validated using a 3D ultrasound probe as a known reference geometry.
Spatial compounding for rotating linear probe in the presence of parameter error using image registration
Show abstract
Spatial compounding has been used in ultrasonic imaging for suppressing speckle noise. The technique generally
involves electronically steering the ultrasonic beams. In this paper, we present a spatial compounding approach where
the component image is acquired by mechanically rotating the probe element. A linear transducer array is rotated about
an axis in the plane of the image. The goal is to avoid the problems associated with the electronic beam steering at a
large angle such as decrease in the effective aperture size, grating lobe effect, and decrease in transducer sensitivity
caused by obliquity factor. In the computation of the ultrasound image, we need the values of the axis of rotation and
angular position of the transducer array. However, the construction of the rotation mechanism and the control system
accompanies the inevitable uncertainties in these values. These parameter errors result in the target position error, and
the consequence is the blurry compounded image. We present the spatial compounding for rotating linear probe in the
presence of parameter error using image registration. The effect of the uncertainty in the mechanical parameters was
compensated by registering the wire target images before spatial compounding. An efficient registration algorithm was
developed to compute the transformation matrix required for the registration. The component images were registered by
the transformation matrix before spatial compounding and the effect of the parameter errors were removed.
Tissue Characterization
Lamb waves detection in a bovine cortical tibia using a scanning laser vibrometry
Show abstract
Most of the techniques for generating and detecting ultrasonic Lamb waves (e.g. angle-beam piezoelectric transducers,
micro-electro mechanical systems (MEMS), comb and interdigital transducers, phased array transducers, and
piezoceramic transducers) require a firm physical contact with the measured objects. For objects with highly irregular
surfaces such as bones, it will be very difficult to produce a good contact. Thus, a non-contact Lamb wave measurement
technique, the scanning laser vibrometry, is proposed in this paper to examine a bovine cortical tibia in vitro. The
ultrasonic Lamb waves used had the center frequency of 84KHz. The waves were generated using a planar transducer
which was coupled with a cone-shaped resonant vibrator. Only the fundamental modes of a0 and s0 were expected to
occur. 2-Dimensional images of the Lamb waves traveling in the bone were recorded. The scan results represent out-of-plane
vibration of the surface of the bone. Lamb wave modes were verified with further post-processing analyses. In
time-domain, time-history prediction of the modes is fitted onto the original detected signal as to confirm their common
rising time for each mode. A frequency-domain method, i.e. wavelet analysis, is also employed to define the traveling
modes and their group velocity. The expected modes can be clearly defined at the center frequency. Additionally, what
seemed to be a new mode, a1, was generated and detected at the higher frequency of the responses.
Multi-modal breast imaging with ultrasound tomography
Show abstract
We report and discuss clinical breast imaging results obtained with operator independent ultrasound tomography. A
series of breast exams are carried out using a recently upgraded clinical prototype designed and built on the principles of
ultrasound tomography. The in-vivo performance of the prototype is assessed by imaging patients at the Karmanos
Cancer Institute. Our techniques successfully demonstrate in-vivo tomographic imaging of breast architecture in both
reflection and transmission imaging modes. These initial results indicate that operator-independent whole-breast imaging
and the detection of cancerous breast masses are feasible using ultrasound tomography techniques. This approach has
the potential to provide a low cost, non-invasive, and non-ionizing means of evaluating breast masses. Future work will
concentrate on extending these results to larger trials.
A novel software-based technique for quantifying placental calcifications and infarctions from ultrasound
Show abstract
In obstetrics, antenatal ultrasound assessment of placental morphology comprises an important part of the estimation of
fetal health. Ultrasound analysis of the placenta may reveal abnormalities such as placental calcification and infarcts.
Current methods of quantification of these abnormalities are subjective and involve a grading system of Grannum stages
I-III. The aim of this project is to develop a software tool that quantifies semi-automatically placental ultrasound images
and facilitates the assessment of placental morphology. We have developed a 2D ultrasound imaging software tool that
allows the obstetrician or sonographer to define the placental region of interest. A secondary reference map is created for
use in our quantification algorithm. Using a slider technique the user can easily define an upper threshold based on high
intensity for calcification classification and a lower threshold to define infarction regions based on low intensity within
the defined region of interest. The percentage of the placental area that is calcified and also the percentage of infarction is
calculated and this is the basis of our new metric. Ultrasound images of abnormal and normal placentas have been
acquired to aid our software development. A full clinical prospective evaluation is currently being performed and we are
currently applying this technology to the three-dimensional ultrasound domain. We have developed a novel software-based
technique for calculating the extent of placental calcification and infarction, providing a new metric in this field.
Our new metric may provide a more accurate measurement of placental calcification and infarction than current
techniques.
A novel ultrasonic method for measuring breast density and breast cancer risk
Show abstract
Women with high mammographic breast density are at 4- to 6-fold increased risk of developing breast cancer compared
to women with fatty breasts. However, current breast density estimations rely on mammography, which cannot provide
accurate volumetric breast representation. Therefore, we explored two techniques of breast density evaluation via
ultrasound tomography. A sample of 93 patients was imaged with our clinical prototype; each dataset contained 45-75
tomograms ranging from near the chest wall through the nipple. Whole breast acoustic velocity was determined by
creating image stacks and evaluating the sound speed frequency distribution. Ultrasound percent density (USPD) was
determined by segmenting high sound speed areas from each tomogram using k-means clustering, integrating over the
entire breast, and dividing by total breast area. Both techniques were independently evaluated using two mammographic
density measures: (1) qualitative, determined by a radiologist's visual assessment using BI-RADS Categories, and (2)
quantitative, via semi-automatic segmentation to calculate mammographic percent density (MPD) for craniocaudal and
medio-lateral oblique mammograms. ~140 m/s difference in acoustic velocity was observed between fatty and dense BI-RADS
Categories. Increased sound speed was found with increased BI-RADS Category and quantitative MPD.
Furthermore, strong positive associations between USPD, BI-RADS Category, and calculated MPD were observed.
These results confirm that utilizing sound speed, both for whole-breast evaluation and segmenting locally, can be
implemented to evaluate breast density.
Elastography
Estimation of 3D cardiac deformation using spatio-temporal elastic registration of non-scanconverted ultrasound data
Show abstract
Current ultrasound methods for measuring myocardial strain are often limited to measurements in one or two
dimensions. Spatio-temporal elastic registration of 3D cardiac ultrasound data can however be used to estimate
the 3D motion and full 3D strain tensor. In this work, the spatio-temporal elastic registration method was
validated for both non-scanconverted and scanconverted images. This was done using simulated 3D pyramidal
ultrasound data sets based on a thick-walled deforming ellipsoid and an adapted convolution model. A B-spline
based frame-to-frame elastic registration method was applied to both the scanconverted and non-scanconverded
data sets and the accuracy of the resulting deformation fields was quantified. The mean accuracy of the estimated
displacement was very similar for the scanconverted and non-scanconverted data sets and thus, it was shown
that 3D elastic registration to estimate the cardiac deformation from ultrasound images can be performed on
non-scanconverted images, but that avoiding of the scanconversion step does not significantly improve the results
of the displacement estimation.
Real-time visualization of pulsatile tissue-motion in B-mode ultrasonogram for assistance in bedside diagnosis of ischemic diseases of neonatal cranium
Show abstract
By developing a real-time visualization system, pulsatile tissue-motion caused by artery pulsation of blood flow has been
visualized continuously from a video stream of ultrasonogram in brightness mode. The system concurrently executes the
three processes: (1) capturing an input B-mode video stream (640×480 pixels/frame, 30 fps) into a ring buffer of 256
frames, (2) detecting intensity and phase of pulsatile tissue-motion at each pixel from a heartbeat-frequency component
in Fourier transform of a series of pixel value through the latest 64 frames as a function of time, and (3) generating an
output video-stream of pulsatile-phase image, in which the motion phase is superimposed as color gradation on an input
video-stream when the motion intensity exceeds a proper threshold. By optimizing the visualization software with the
streaming SIMD extensions, the pulsatile-phase image has been continuously updated at more than 10 fps, which was
enough to observe pulsatile tissue-motion in real time. Compared to the retrospective technique, the real-time
visualization had clear advantages not only in enabling bedside observation and quick snapshot of pulsatile tissue-motion
but also in giving useful feedback to probe handling for avoiding unwanted motion-artifacts, which may strongly assist
pediatricians in bedside diagnosis of ischemic diseases.
Image-based speckle tracking for tissue motion characterization in a deformable cardiovascular phantom
Show abstract
We present and validate image-based speckle-tracking calipers for quantification of tissue deformation and rotation
in dynamic cardiovascular phantom models. Lagrangian strain was computed from the change in distance
between caliper regions-of-interest (ROIs) positioned within the wall of a pulsatile phantom and compared with
reference measurements derived from cardiac CT imaging. In a torsion phantom, rotational tissue excursion
in a 2D plane was estimated and compared with reference values from CT-scan data. Tissue deformation and
rotation measurements correlated well with their respective reference measurements. Our algorithm is capable
of estimating strain and rotation from distinct tissue regions without requiring explicit cardiac border detection,
a step which can be especially challenging in patients with poor acoustic windows.
IVUS catheter motion: implications for ultrasonic speckle and correction in IVUS palpography
Show abstract
Intravascular Ultrasound (IVUS) palpography is a techniques that depicts the distribution of the mechanical
strain over the luminal surface of coronary arteries. It utilizes conventional radiofrequency (RF) signals acquired
at two different levels of a compressional load. The signals are cross-correlated to obtain the microscopic tissue
displacements, which can be directly translated into local strain of the vessel wall. However, (apparent) tissue
motion and nonuniform deformation of the vessel wall due to catheter jolting and rotation reduce signal correlation
and result in void strain estimates. Implications of probe motion were studied on the tissue-mimicking
phantom. The measured circumferential tissue displacement and level of the speckle decorrelation amounted to
12° and 0.58 for the catheter displacement of 800 μm, respectively. To compensate for the motion artifacts in
IVUS palpography, a novel method, based on the feature-based scale-space Optical Flow (OF) was employed.
The computed OF vector field quantifies the amount of the local tissue misalignment in consecutive frames.
Subsequently, the extracted motion pattern is used to realign the signals prior to the cross-correlation analysis,
reducing signal decorrelation and increasing the number of valid strain estimates. The advantage of applying the
motion compensation algorithms was demonstrated in a mid-scale validation study on 14 in-vivo pullbacks. Both
methods substantially increase the number of valid strain estimates in the partial and compounded palpograms.
A mean relative improvement amounts to 28% and 14%, respectively. Implementation of motion compensation
method increase the diagnostic value of IVUS palpography.
Strain index: a new visualizing parameter for US elastography
Show abstract
Elastography, an ultrasound modality based on the relation between tissue strain and its mechanical properties, has a strong potential in the diagnosis and prognosis of tumors. For instance, tissue affected by breast and prostate cancer undergoes a change in its elastic properties. These changes can be measured using ultrasound signals. The standard way to visualize the elastic properties of tissues in elastography is the representation of the axial strain. Other approaches investigate the information contained in shear strain elastograms, vorticity or the representation of the full strain tensor. In this paper, we propose to represent the elastic behaviour of tissues through the visualization of the Strain Index, related with the trace of the strain tensor. Based on the mathematical interpretation of the strain tensor, this novel parameter is equivalent to the sum of the eigenvalues of the strain tensor, and constitutes a measure of the total amount of strain of the soft tissue. In order to show the potential of this visualization approach, a tissue-mimicking phantom was modeled as a 10x10x5 cm region containing a centered 10mm cylindrical inclusion three times stiffer than the surrounding material, and its elastic behavior was simulated using finite elements software. Synthetic pre- and post-compression (1.25%) B-mode images were computer-generated with ultrasound simulator. Results show that the visualization of the tensor trace significantly improves the representation and detection of inclusions, and can help add insight in the detection of different types of tumors.
A new approach to elastography using a modified demons registration algorithm
Show abstract
Changes in tissue stiffness correlate with pathological phenomena that can aid the diagnosis of several diseases
such as breast and prostate cancer. Ultrasound elastography measures the elastic properties of soft tissues using
ultrasound signals.
The standard way to estimate the displacement field from which researchers obtain the strain in elastography
is the time-domain cross-correlation estimator (TDE). Optical flow (OF) methods have been also characterized
and their use keeps increasing.
We introduce in this paper the use of a Modified Demons Algorithm (MDA) to estimate the displacement
field and we compare it with OF. A least-squares strain estimator (LSE) is applied to estimate the strain from the
displacement. The input for the algorithm comes from the ultrasound scanner standard video output; therefore,
its clinical implementation is immediate.
To test the algorithm, a tissuemimicking phantom was modeled as a 10x10x5 cm region containing a centered
10mm cylindrical inclusion three times stiffer than the surrounding material, and its elastic behavior was simulated
using COMSOL Multiphysics 3.2 software. Synthetic pre- and post-compression (1.25%) B-mode images
were computer generated using FIELD II ultrasound simulator. Afterward, the algorithm was tested with a
commercial CIRS breast elastography phantom, applying a 2% freehand compression.
Axial displacement fields and strain figures are presented and in the case of the synthetic one compared
to the ground truth given by the FE software. Although other researchers have used registration methods for
elastography, as far as we know, they have not been used as stand alone but together with elastic modulus
reconstruction or FE which iteratively varies material properties to improve registration.
Poster Session
Software implementation of ultrasound beamforming using ADSP-TS201 DSPs
Show abstract
We present a software-based ultrasound beamformer to build a fully software-based ultrasound scanner which performs
real-time delay-sum beamforming using fractional delay filters using ADSP-TS201 DPSs (Analog Device Inc.). Receive
dynamic focusing generally requires fast transfer of a large amount of data for inter-channel summation and for delay
control. The DSPs are connected in pipelined network architecture without a glue logic using the DPS's parallel ports,
which allows the connection of unlimited DSPs. Each DSP has a small input FIFO which takes as input the data samples
from each ADC and can take delay values either from a memory (to use pre-calculated delay values) or an external
FPGA (for real-time delay calculation) via its LVDS channel. Two fractional delay beamformer (FDBF) schemes are
implemented on the DSP system. Each scheme is programmed in assembly code optimized for speed; instruction level
parallelism is secured to maximally utilize the four execution units of each DSP, pipeline scheduling is employed to
avoid pipeline stall, and instruction reordering techniques are used to prevent memory contention while preserving the
program semantics. It is found that dynamic focusing is carried out faster when delay filtering is performed prior to interchannel
summation, whereas hardware implementation of the FDBF favors performing delay filtering after interchannel
summation. The frame rate achievable with 16 DSPs is up to 28Hz when the sampling rate is 40MHz, the view depth is
20cm, the number of scanline is 128, and the number of channel is 64.
High throughput SAFT for an experimental USCT system as MATLAB implementation with use of SIMD CPU instructions
Show abstract
At Forschungszentrum Karlsruhe an Ultrasound Computer Tomography system USCT) is under development for
early breast cancer detection. To detect morphological indicators in sub-millimeter resolution, the visualization
is based on a SAFT algorithm (synthetic aperture focusing technique). The current 3D demonstrator system
consists of approx. 2000 transducers, which are arranged in layers on a cylinder of 18 cm diameter and 15 cm
height. With 3.5 millions of acquired raw data sets and up to one billion voxels for an image, a reconstruction
may last up to months.
In this work a performance optimized SAFT algorithm is developed. The used software environment is
MathWorks' MATLAB. Several approaches were analyzed: a plain M-code (MATLAB's native language), an
optimized M-code, a C-code implementation, and a low-level assembler implementation. The fastest found
solution uses an SIMD enhanced assembler code wrapped in the C-interface of MATLAB. Additionally a 10%
speed up is gained by reducing the function call overhead. The overall speed up is more than one order of
magnitude. The resulting computational efficiency is near the theoretical optimum. The reconstruction time is
significantly reduced without losing MATLAB's comfortable development environment.
A method to design an optimum pair of transmit and receive periodic sparse arrays
Show abstract
In such applications as fast 3D imaging with 2D arrays and point-of-care imaging with an ultra portable devices, periodic
sparse arrays(PSA) can be efficiently used to increase the effective aperture size with less number of active elements
than the conventional method. Generally, PSA can be represented as sub-arrays distributed uniformly in P -element
intervals, each with L consecutive elements, where L < P. Since the continuous wave beam pattern in the far-field is
given by Fourier transform of aperture function, the beam pattern of PSA is a multiplication of beam patterns of the
upsampled dense array by the ratio of P and L -elements sub-array. A recent method to design a PSA pair provides
analytically the values of P and L for transmit and receive arrays to eliminate the dominant grating lobes, which occur
at the same position on both transmit and receive. In this work, we present a method to design a PSA pair with improved
performance by further suppressing the residual grating lobes of PSA. It can be accomplished by properly shading
amplitude of the transmit and receive sparse arrays. This shading window function is also obtained by signal analysis of
aperture functions. The beam patterns of various PSA pairs based on the proposed design method are evaluated through
computer simulations. The simulation results show that the residual grating lobes are reduced by about 10dB more in all
cases. Consequently, our method can be used to improve the performance of beam pattern or enhance the periodicity of
sparse array.
Photoacoustic 3D visualization of tumor angiogenesis
Show abstract
Photoacoustic imaging is used to obtain a range of three-dimensional images representing tumor neovascularization
over a 10-day period after subcutaneous inoculation of pancreatic tumor cells in a rat. The images are
reconstructed from data measured with a double-ring photoacoustic detector. The ultrasound data originates
from the optical absorption by hemoglobin of 14 ns laser pulses at a wavelength of 1064 nm. Three-dimensional
data is obtained by using two dimensional linear scanning. Scanning and motion artifacts are reduced using a
correction method. The data is used to visualize the development of the individual blood vessels around the
growing tumor, blood concentration changes inside the tumor and growth in depth of the neovascularized region.
The three-dimensional vasculature reconstruction is created using VTK, which enables us to create a composition
of the vasculature on day seven, eight and ten and to interactively measure tumor growth in the near future.
Imaging of acoustic attenuation and speed of sound maps using photoacoustic measurements
Show abstract
Photoacoustic imaging is an upcoming medical imaging modality with the potential of imaging both optical and
acoustic properties of objects. We present a measurement system and outline reconstruction methods to image
both speed of sound and acoustic attenuation distributions of an object using only pulsed light excitation. These
acoustic properties can be used in a subsequent step to improve the image quality of the optical absorption
distribution. A passive element, which is a high absorbing material with a small cross-section such as a carbon
fiber, is introduced between the light beam and the object. This passive element acts as a photoacoustic source
and measurements are obtained by allowing the generated acoustic signal to propagate through the object. From
these measurements we can extract measures of line integrals over the acoustic property distribution for both
the speed of sound and the acoustic attenuation. Reconstruction of the acoustic property distributions then
comes down to the inversion of a linear system relating the obtained projection measurements to the acoustic
property distributions. We show the results of applying our approach on phantom objects. Satisfactory results
are obtained for both the reconstruction of speed of sound and the acoustic attenuation.
A study of 3-way image fusion for characterizing acoustic properties of breast tissue
Show abstract
A novel clinical prototype, CURE (Computed Ultrasound Risk Evaluation), is used to collect breast tissue image data of
patients with either benign or malignant masses. Three types of images, reflection, sound speed and attenuation, are
generated from the raw data using tomographic reconstruction algorithms. Each type of image, usually presented as a
gray scale image, maps different characteristics of the breast tissue. This study is focused on fusing all three types of
images to create true color (RGB) images by assigning a different primary color to each type of image. The resulting
fused images display multiple tissue characteristics that can be viewed simultaneously. Preliminary results indicate that
it may be possible to characterize breast masses on the basis of viewing the superimposed information. Such a
methodology has the potential to dramatically reduce the time required to view all the acquired data and to make a
clinical assessment. Since the color scale can be quantified, it may also be possible to segment such images in order to
isolate the regions of interest and to ultimately allow automated methods for mass detection and characterization.
Comparison of ultrasound attenuation tomography methods for breast imaging
Show abstract
Ultrasound attenuation parameters of breast masses are closely related to their types and pathological states, therefore, it
is essential to reliably estimate attenuation parameters for quantitative breast tissue characterization. We study the
applicability of three different attenuation tomography methods for ultrasound breast imaging using a ring transducer
array. The first method uses the amplitude decays of signals transmitted through the breast to reconstruct attenuation
coefficients. The second method employs the spectral ratios between the pulse propagating through the breast and that
through water to obtain attenuation parameters. The third method makes use of the complex energy ratios estimated
using the amplitude envelopes of transmitted signals. We use in vitro and in vivo breast data acquired with a clinical
ultrasound breast imaging system (CURE) to compare these tomography methods. Our results show that the amplitude
decay method yields attenuation coefficients with more artifacts than the other two methods. There is bias and
variability in the estimated attenuation using the spectral ratio due to its sensitivity to different temporal band-widths and
signal-to-noise-ratios of the data. The method based on the complex signal energy ratio is more robust than the other
two methods and yields images with fewer artifacts.
Simulation of microbubble response to ambient pressure changes
Show abstract
The theory on microbubbles clearly indicates a relation between the ambient pressure and the acoustic behavior
of the bubble. The purpose of this study was to optimize the sensitivity of ambient pressure measurements,
using the subharmonic component, through microbubble response simulations. The behaviour of two different
contrast agents was investigated as a function of driving pulse and ambient overpressure, pov. Simulations of
Levovist using a rectangular driving pulse show an almost linear reduction in the subharmonic component as
pov is increased. For a 20 cycles driving pulse, a reduction of 4.6 dB is observed when changing pov from 0 to
25 kPa. Increasing the pulse duration makes the reduction even more clear. For a pulse with 64 cycles, the
reduction is 9.9 dB. This simulation is in good correspondence with measurement results presented by Shi et al.
1999, who found a linear reduction of 9.6 dB. Further simulations of Levovist show that also the shape and the
acoustic pressure of the driving pulse are very important factors. The best pressure sensitivity of Levovist was
found to be 0.88 dB/kPa. For Sonazoid, a sensitivity of 0.71 dB/kPa has been found, although the reduction is
not completely linear as a function of the ambient pressure.
Harmonic quadrature demodulation for extracting the envelope of the second harmonic component
Show abstract
An efficient method for separating the harmonic component (2f0) from the fundamental component (f0) using harmonic
quadrature demodulation is presented. In the proposed method, the focused ultrasound signal is mixed with cosine and
sine signal waveforms of harmonic frequency 2f0 to produce the inphase and quadrature components, respectively. The
quadrature component is Hilbert-transformed and then added to the inphase component. This process cancels out both
the high and low frequency components of the mixed fundamental signal and the high frequency component of the
mixed harmonic signal, leaving only the envelope of the harmonic signal at the base band. This signal is then fed to a
low-pass filter to remove out of band noise. In summary, this method can extract the harmonic signal after a single
transmit-receive event even when there exists frequency overlap between the f0 and 2f0 components. Hence, the
proposed method is superior to the pulse inversion method which requires twice as many transmit-receive cycles as well
as the conventional filtering method which has a bandwidth limitation. Therefore, one can find the proposed method
useful not only for tissue harmonic imaging but also for contrast agent imaging in applications where high frame rate or
low motion artifact is important. The proposed method is verified by both the analytic and computer simulation studies.
For a stationary target, the difference between the estimated harmonic signals by the proposed and the pulse inversion
methods is within 0.1%.
Measurement of thermally ablated lesions in sonoelastographic images using level set methods
Show abstract
The capability of sonoelastography to detect lesions based on elasticity contrast can be applied to monitor the creation of
thermally ablated lesion. Currently, segmentation of lesions depicted in sonoelastographic images is performed manually
which can be a time consuming process and prone to significant intra- and inter-observer variability. This work presents
a semi-automated segmentation algorithm for sonoelastographic data. The user starts by planting a seed in the perceived
center of the lesion. Fast marching methods use this information to create an initial estimate of the lesion. Subsequently,
level set methods refine its final shape by attaching the segmented contour to edges in the image while maintaining
smoothness. The algorithm is applied to in vivo sonoelastographic images from twenty five thermal ablated lesions
created in porcine livers. The estimated area is compared to results from manual segmentation and gross pathology
images. Results show that the algorithm outperforms manual segmentation in accuracy, inter- and intra-observer
variability. The processing time per image is significantly reduced.