When medicine asks a machine to compare two images of the same body taken months apart, it is asking a computational system to understand how living tissue moves — a question that is, at its root, a question about physics. Researchers have now answered it more honestly, embedding the mechanical laws of bone, soft tissue, and sliding organ boundaries directly into the neural networks that perform this alignment. The result is a system that does not merely match pixels, but reasons about anatomy — and carries that reasoning forward to patients it has never encountered before.