(D-132) Feasibility of using polarization-sensitive optical coherence tomography endoscope in liver tumor biopsy guidance

Introduction:: Liver tumors grows on or inside the liver. It is a life-threatening disease, and the incidence rate is still growing. In the United States, around 36,000 people get liver cancer each year and about 28,000 people die from it each year. Liver biopsy is commonly used to assess the liver disease and monitor the treatment. It helps the diagnosis of the liver tumor and make the treatment plan. In liver biopsy, a needle is inserted inside the liver to take out the tissue for histology. Imaging methods are required to guide the biopsy needle during insertion. Ultrasound, computed tomography (CT), and magnetic resonance imaging (MRI) have been utilized in liver biopsy guidance. However, due to the limited spatial resolution, it is still challenging to accurately locate the tumor position or recognize small tumors.


Materials and Methods:: In this study, we propose to use optical coherence tomography (OCT) to recognize the tissue in front of the biopsy needle during the biopsy procedure. OCT is a novel imaging modality which can visualize the subsurface tissue structure. Its resolution is 10-100 times higher than conventional imaging modalities, so it has the potential to guide the biopsy needle more efficiently. In addition, polarization-sensitive OCT (PS-OCT) can provide polarization information from the tissue, so it helps provide more features to recognize the tissue types. We established an endoscopic PS-OCT system by integrating a gradient index (GRIN) lens to provide the forward view of endoscope. Five human liver tumor samples were utilized in our experiments. Normal liver tissue and tumor tissue were imaged. In addition, we used deep learning methods to automatically recognize the tissue type. Convolutional neural network (CNN) models were applied in normal tissue and tumor tissue classification.


Results, Conclusions, and Discussions:: From the imaging results, OCT imaging results from intensity mode and PS mode showed different features between normal and tumor tissues, thus our PS-OCT endoscope was feasible to recognize the liver tumor during liver biopsy procedure. Furthermore, the CNN based models provided over 95% classification accuracy. Currently, we are collecting more data from additional liver samples, and further training our CNN models for tissue identification.


Acknowledgements (Optional): : This work was supported by grants from the University of Oklahoma Health Sciences Center (3P30CA225520), Faculty Investment Program from University of Oklahoma, Institutional Research Grant number IRG-19-142-01 from the American Cancer Society, National Science Foundation (OIA-2132161, 2238648), National Institute of Health (R01DK133717), Oklahoma Shared Clinical and Translational Resources (NIGMS U54GM104938), Oklahoma Center for the Advancement of Science and Technology (HR23-071), and the medical imaging COBRE (P20 GM135009). Histology service provided by the Tissue Pathology Shared Resource was supported in part by the National Institute of General Medical Sciences COBRE Grant P20GM103639 and National Cancer Institute Grant P30CA225520 of the National Institutes of Health.


References (Optional): :