Swin transformer
A hierarchical vision transformer that processes images in shifted local windows, making it efficient and well suited to dense tasks like medical image segmentation. Swin UNETR pairs a Swin encoder with a UNet-style decoder for 3D segmentation.
A radiotherapy department evaluates an auto-contouring tool whose technical documentation says it uses a Swin-based segmentation network. That explains how it can process a full planning CT efficiently, by computing attention within local 3D windows at several scales instead of across the whole volume at once. How closely its contours match the ones your team draws is a separate matter that only a local comparison settles. Run it on twenty recent cases your team has already contoured and record how much editing each organ's contour needed.
Terms like this come up in real clinical scenarios across the HelloAI courses: bite-sized modules with verifiable certificates. An account takes one minute, no password needed.
Sign in →