Enables 3D medical image segmentation pre-training using only mathematical formulas and implicit functions, requiring zero real-world data or expert annotations.
March 25, 2026
Original Paper
FDIF: Formula-Driven supervised Learning with Implicit Functions for 3D Medical Image Segmentation
arXiv · 2603.23199
The Takeaway
It bypasses the biggest bottleneck in medical AI—data privacy and annotation costs—by demonstrating that math-generated textures and geometries can match the performance of self-supervised learning on large-scale real datasets.
From the abstract
Deep learning-based 3D medical image segmentation methods relies on large-scale labeled datasets, yet acquiring such data is difficult due to privacy constraints and the high cost of expert annotation. Formula-Driven Supervised Learning (FDSL) offers an appealing alternative by generating training data and labels directly from mathematical formulas. However, existing voxel-based approaches are limited in geometric expressiveness and cannot synthesize realistic textures. We introduce Formula-Driv