Computer Vision News - January 2021

13 Challenges of AI in urology • previously taken MRI images during a live feed from US simultaneously. • RSIP Vision offers an alternative to manual registration (cognitive fusion): accurate, safe and robust software for quick and fully automated image fusion (registration). This assistance in the critical navigation during interventional procedures helps physicians ensure better overall clinical outcome. • Video analysis - During some interventional procedures, small sensors are inserted in order to get a live video feed . For example, when a prostate gland is enlarged due to a malignant or a benign tumor, tissue often needs to be excised. The main challenge involves accurately reaching the targeted tissue using live video feed. • During a recent conversation with Ilya Kovler , CTO at RSIP Vision , Ilya enthusiastically told us about the significant progress made in tackling this challenge. He explains that the video feed taken by the sensor is used in order to reconstruct a three- dimensional image . Ilya says that the algorithm developed at RSIP Vision combined both classical computer vision and deep learning in order to complete the process. The algorithm starts with a rigid approximation, which is then updated to account for elastic deformations of the tissue. When dealing with excision • of prostate tissue, RSIP Vision used deep learning to reliably detect subtle anatomical cues for use as spatial landmarks. The result is better guidance and improved accuracy during interventional excision of prostate tissue. While the challenges presented here relate to urological diseases of the prostate, they most certainly also apply for other urological diseases , particularly those that demand interventional procedures. Biopsies of different kinds and treatment of kidney and bladder cancers demand better image analysis and AI tools for improved patient treatment and outcome. MRI imaging of the prostate gland A segmented Ultrasound image of the prostate gland