This PDF file contains the front matter associated with SPIE Proceedings Volume 11232 including the Title Page, Copyright information, Table of Contents, Introduction, and Conference Committee listing.

Ovarian cancer is one of the most lethal gynecological conditions in the developed world. Current screening methods have only made marginal differences in overall survival over the past 30 years. The deficit of early-stage detection methods is a critical factor in the mortality associated with this disease. Recent evidence has shown that the fallopian tubes are a critical site in carcinogenesis of ovarian cancers. We present the first endoscopic co-registered OCT-AFI imaging of ex vivo fallopian tubes. This work aims to evaluate the potential of OCT-AFI to identify pre-cancerous lesions in the fallopian tubes. The BC Cancer Research Centre’s Optical Imaging Lab has developed a multimodal imaging system and catheter which enables both optical coherence tomography (OCT) and autofluorescence imaging (AFI). The imaging probe consists of a dual-clad fiber optical core inside a 0.9mm diameter sterile sheath. This system allows for resolutions of 20-30μm and imaging depths of up to 1.5mm. Samples are collected from patients consented through the OVCARE Gynecological Cancer Tissue Bank banking protocol. Volumetric OCT-AFI images are acquired for the entire catheterizable length of the sample at pullback speeds of 1mm/s. After imaging, histology is conducted according to the “sectioning and extensively examining the fimbriated end” protocol to serve as a gold standard. We present methods for obtaining scans of the ex vivo fallopian tubes, sample cases correlated with histology, and our preliminary results. As of January 2020, we have imaged 21 patients and 27 fallopian tubes including 6 cancerous specimens.

Enhanced visualization of choroidal neovascularization (CNV) is critical to the precise diagnosis and treatment for patients with neovascular eye diseases such as wet macular degeneration. Currently, eye clinics lack an imaging modality that can precisely detect and visualize the position and margins of CNV in three-dimensions. This study shows that a unique multimodal photoacoustic microscopy, optical coherence tomography, and florescence microscopy system can accurately identify CNV in living rabbits with high resolution and image contrast at a sub-10-micron scale. In addition, to better visualize CNV and distinguish it from normal blood vessels, the peptide RGD targeting neovascularization was conjugated with gold nanostars (GNS) capped with photostable near-infrared (NIR) fluorophore alexa fluor 790 (AF790). Four New Zealand white rabbits with laser-induced CNV were intravenously administered with 400µL of GNS at a concentration of 5 mg/mL. The accumulation of GNS at CNV was monitored by multimodal OCT, PAM, color fundus photographs, fluorescein angiography (FA), and indocyanine green angiography (ICGA) at various time points: 1 h, 2 h, 4 h, 8 h, 24 h, 48 h, 72 h, day 4, day 7, day 9, day 11, and day 14. The experimental results show that GNS accumulation occurs in the region of CNV. The GNS were detected by all three imaging modalities both in vitro and in living rabbits. The PA signal was increased 19-fold 24 h post-injection. In addition, fluorescence signal gradually decreased over time. Histological analysis and TUNEL assay show no toxicity in the rabbit at the administered concentrations. Therefore, GNS-assisted multimodal imaging has the possibility to improve microvasculature imaging.

Multipath artifacts in double clad fiber (DCF) based endoscopic optical coherence tomography (OCT) imaging systems are investigated and a novel mechanism for artifact generation is proposed. We present evidence that the characteristic image artifacts found in DCF OCT images are partially due to the existence of an index dip within the core of double clad optical fibers. This core dip is shown to affect the modal quality of the light propagating through the core of the DCF, causing additional peaks or ghost images to be generated within the point spread function of the OCT system. Through these investigations we hope to gain a better understanding of how modal artifacts degrade OCT image quality, allowing for the design of more ideal optical fibers which can restore the quality of the OCT imaging domain.

The most common treatment for early-stage breast cancer involves breast-conserving surgery (BCS), which is a local resection of malignant tissue. BCS success relies on the presence of healthy tissue margins surrounding each resection. However, approximately 20% of BCS patients require follow-up surgery to remove residual tumor. Micro-computed tomography and optical structured light imaging were investigated for intraoperative margin assessment. Specimens were imaged immediately after resection and before surgical inking, and image data were compared to postoperative histopathology. The study demonstrated the feasibility of imaging during BCS and generated a baseline multimodal dataset of whole BCS resections, including positive margin signatures.

Endoscopic optical coherence tomography (OCT) and near-infrared (NIR) fluorescence imaging system was developed for characterization of colorectal cancer (CRC). NIR fluorescence is able to highlight the cancer-suspected area based on significant change of tumor vascular density and morphology. Co-registered OCT images has the capability of visualizing subsurface tissue layer architecture, so the suspected regions can be further investigated by the altered light scattering resulted from the morphological abnormality. The imaging result from in vivo rat experiment has demonstrated the enhanced capability of identification and classification of CRC compared to using any of these technologies alone, thus has the potential to provide a new clinical tool to advance gastroenterology practice.

Advanced optical endoscopic imaging techniques, including hyperspectral and holographic endoscopy, have shown promise in the improved diagnosis of the early stage of cancer. However, clinical applications of these imaging systems are still limited due to unclear diagnostic optical properties. Here, we developed a compact multimodal imaging system that enables hyperspectral imaging, spatial-frequency domain imaging, and 3D profilometric imaging to characterise the optimal optical features for the early detection of lesions. Optical properties of fresh specimens obtained from patients were measured within 30 minutes, and then histopathological assessment of specimens was performed to link extracted optical features to gold-standard diagnosis. With further sample collection and system refinements, this system can be used for high-throughput optical characterisation of fresh tissue specimens, allowing us to determine the optical signatures of early-stage disease.

A portable proof-of-concept prototype device for single snapshot capturing of four spectral line images has been designed, assembled and laboratory-tested. It comprises optical unit that ensures even illumination of the skin target area simultaneously at four laser wavelengths - 450 nm, 523 nm, 638 nm and 850 nm, double-camera image recording system, micro-computer managed operation system and a touch-screen display for image control and displaying the concentration distribution maps of four skin chromophores - melanin, oxy-hemoglobin, deoxy-hemoglobin and bilirubin. Besides, the device captures skin auto fluorescence image at 405 nm laser excitation to separate seborrheic keratosis from other pigmented skin lesions. Skin chromophore maps are calculated offline by an external computer.

There is considerable interest in using convolutional neural networks for computer aided medical diagnostics. In this paper, we present some recent results on brain lesion segmentation. While algorithms have been developed to automate the process, the results often lack accuracy in tracing the lesions. Here, we present our results on the publically available ATLAS R1.1 dataset using deep CNN architectures. We utilize a deep U-net architecture with 3D convolutions, consisting of an encoder and a corresponding decoder along with dense residual connections between them, leveraging both low and high level features from the encoder at each step. We also analyze the use of Inception, Residual and Inception-Res blocks for the encoder. The architecture also makes use of an attention gate for the decoder part to suppress irrelevant parts of the incoming input while highlighting the salient features. The dataset had 239 3D MRI images of dimensions (197 x 233 x 189) with their corresponding segmentation maps. We used a train-validation-test split of 7:2:1, employing mean normalization, histogram equalization and suitable data augmentation techniques. We used SGD optimizer with focal Tversky loss function. The Dice score and AUC were used as metrics. We were able to achieve a Dice coefficient of 0.53 with an AUC around 0.95 with a single end-to-end model, unlike previous models that either focus only on one part of the brain or use other metadata, and matched their performance in every aspect. We further discuss the possibilities of improving the results through denoising and data cleansing using standard machine learning and computer vision methodologies.

Magnetic Resonance Images are reconstructed from a finite number of samples in k-space. The accuracy of these reconstructions are crucial for segmentation and diagnosis. However, the nature of the reconstruction leads inevitably to Gibbs ringing. In this paper, we quantitatively compare the filtered Fourier and Gegenbauer ringing-suppressing reconstruction methods. The Gegenbauer method yields an order of magnitude better MSE than the other approaches we consider, and a 10 dB improvement in PSNR. These results confirm the Gegenbauer reconstruction as the most accurate choice in inverse problems where data is reconstructed from a finite number of Fourier coefficients.

The objective of the research was a multifaceted study of strangulated small intestine to reveal the optical, morphofunctional and biochemical signs of small bowel ischemia. The study was carried out in vivo using an artificially induced strangulation model of the small intestine (together with its mesentery and blood vessels) in 12 Wistar rats. Over a period of 120 minutes following the bowel ligation, changes in the density of the intramural vasculature and intestinal wall microstructure were detected using multimodal optical coherence tomography (MM OCT). Fluorescence lifetime changes of endogenous fluorophores were also measured using macro-FLIM of the strangulated loop and the adjacent segments of the intestine. At the end of the experiment, a morphometric study of the thickness of the layers and the prevalence of necrosis in the intestinal wall was carried out. A comprehensive analysis of the results of the OCT, FLIM and morphometry of the ischemic wall of the small intestine made it possible to determine the correlating morphofunctional and biochemical manifestations that are specific to this model of mesenteric blood flow disturbance.

To explore challenges for further improvement of diagnostic performance, a project aimed at development of technology for tri-modal skin imaging by combining multispectral, fluorescence lifetime and Raman band imaging was initiated. In this study, each of the mentioned imaging modalities has been preliminary tested and updated. Four different multispectral imaging devices were tested on color standards. Picosecond laser-excited fluorescence lifetime imaging equipment was examined on ex-vivo skin samples. Finally, a new Raman spectroscopy setup with 785 nm laser was launched and tested on cell cultures and ex-vivo skin. Advantages and specific features of the tri-modal skin imaging are discussed.