Computer Vision News - March 2022

34 AI for Surgical Orthopedics standard or reverse TSA is needed. RSIP Vision applies advanced Artificial Intelligence (AI) and computer vision algorithms to reduce radiation exposure to the patient, time to treatment, and procedural costs. Our new TSA planning tool accurately segments the shoulder bones - scapula and humerus - from a shoulder MRI scan. This tool is a pre-trained fully convolutional network (FCN) , trained on real shoulder MRI scans. The output of this network has one main detriment - it has a relatively low resolution compared to CT-based models and does not allow proper planning. To overcome this challenge, we developed Shoulder pain due to osteoarthritis or trauma is common, specifically in the elderly. To rule out soft tissue damage - tendon, muscle, or labrum tear - a shoulder MRI scan is conducted. Once soft-tissue damage is eliminated and bone damage is diagnosed, the physician may decide that TSA, or shoulder replacement , is required. Current practice dictates that for adequate planning, a shoulder CT scan is required. This scan is used to reconstruct a 3Dmodel of the shoulder bones - the humerus and scapula . From the model, the surgeon can extract accurate dimensions of the shoulder bone and fit a model of the implant to verify sizing. Additionally, this model can be used to determine whether TOTAL SHOULDER ARTHROPLASTY (TSA) PLANNING BASED ON SHOULDER MRI