Computer Vision News - May 2023

29 Yuan Zhang and even on the medical side, it’s still very challenging, ” she tells us. “ I needed to use some cross-field knowledge. Doctors and surgeons have helped me and explained everything. They’re super busy but can always spare me 30 minutes of their time. They know how to diagnose this disease from MRI, so we can mimic what they do in our algorithm. ” ultrasound to check if their organs freely glide over each other. However, this is highly dependent on the operators’ skills . MRI provides more information, but again, diagnosis requires an experienced clinician , and it is generally more challenging to detect POD obliteration from MRI than TVUS. To address this problem, Yuan has developed an AI model to improve endometriosis diagnosis from MRI, using knowledge distilled from unpaired TVUS data . To achieve this, she recognized the need for multidisciplinary expertise. The work is backed by a team of researchers and clinicians working together on a large project called IMAGENDO. “ I’ma computer science student, but this is a huge medical issue, IEEE ISBI IEEE ISBI