Computer Vision News - April 2021

Medical Imaging Projects 14 module that utilizes state-of-the- art deep learning (DL) algorithms designed and trained in advance to deal with the presence of metals in CT scans. This module accounts for the possible presence of metals by using an accurate, physical-based, and robust simulation tool of these metal artifacts. By adding synthetically produced data to the training process, the degrading effect of metals in CT scans has been dramatically reduced. “This module has a vital and important contribution to the treatment of the patients undergoing orthopedic procedures” , says RSIP Vision CEO Ron Soferman, “since many patients undergo additional or follow-up procedures throughout their lifetime”. The result is accurate discernment of the location of metals in the bones and exact delineation of their outlines from the bones. Using RSIP Vision's metal segmentation toolbox, accurate bone modelling is made easy even in the presence of preexisting metallic implants. The result is improved orthopedic surgical planning, safer and more accurate procedures and overall better patient treatment and outcome. Knee joint segmentation 2: A knee CT scan with image artifacts due to an existing metal implant (left) segmented to all bone and metal components (right)