Medical scans are maps of tissue density. The brightness of each pixel reveals what’s there, much like photometry measures light intensity. But conventional algorithms analyze pixels in isolation, producing jagged organ boundaries.
A new approach compels the program to prioritize smoothness. During training, a geometric rule is embedded: the contour must not only be accurate but also gracefully smooth. This mirrors the behavior of water — a droplet always assumes a shape with minimal surface area, striving for the lowest energy state, akin to reducing entropy in physical systems.
Experiments on liver and spleen scans confirmed a leap in accuracy. Crisp boundaries ease diagnosis, surgical planning, and the training of medical robots. Surprisingly, this simple geometric constraint boosted recognition quality more than adding thousands of new scans to the training set.
In zero gravity, a water droplet naturally forms a perfect sphere: this self-assembly principle inspired the algorithm’s creators.
The fantastical scanners from Star Trek move one step closer: algorithms are adopting a droplet’s geometry.