Computer Vision News - October 2021

30 AI for Urology Bladder cancer is a relatively common disease, affecting approximately 1 in 27 men (less common in women). When detected early, survival rates are high. Therefore, early and precise detection is crucial for bladder cancer healthcare. Typical symptoms includeblood inurine, painful urination, and even back pain. When these appear, common practice dictates that a cystoscopy is performed – a scope is inserted into the patient’s urinary tract until the bladder, which is then scanned in search for lesions and tumors. Additionally, bladder cancer has high recurrence rate, requiring periodical examinations for patients in remission. The main challenge in bladder cystoscopy is navigation within the bladder. It is vital that the bladder is scanned in its entirety, without leaving unexamined tissue, ruling out missed lesions. Currently, the physician manually scans the bladder repeatedly until they are certain it was completely viewed, and all suspicious areas were examined. RSIP Vision has recently implemented “Shape-from-Motion” algorithms and created a solution for the aforementioned challenge. This set of algorithms extracts key points from the cystoscopy video used for two tools: For both tools: • Panorama – Several key points coincide in two or more images. Using these points, the images are stitched into a large, clear, panoramic image of the bladder wall by finding the mapping between a reference image to another • Sparse Reconstruction – the key points are used in a shape-from- motion framework to get the 3D location of each key point and image camera. This point cloud is a sparse 3D model of the bladder. This model is essentially a map of the bladder. The physician can use it to verify complete coverage of the scan, or as a tool to re- visit suspicious areas. Additionally, as key points are shared in both tools, matched information is available to the surgeon. Bladder Panorama Generator and Sparse Reconstruction Tool Panorama of a bladder model