NCA in Ultrasound

This master's thesis at TUM in cooperation with Helmholtz explores Neural Cellular Automata (NCA) as a lightweight, bio-inspired approach to estimate tissue displacement from ultrasound image sequences. The model learns local update rules that propagate spatially — achieving displacement estimation with significantly fewer parameters than conventional deep learning methods.

Future work targets sonification of displacement values to give doctors real-time audio feedback during biopsies, translating spatial displacement patterns into sound for intuitive guidance.