Skull stripping with purely synthetic data

TL;DR

This paper introduces PUMBA, a novel brain extraction method trained solely on synthetic data, achieving comparable accuracy across diverse modalities, species, and pathological cases without using real images or labels.

Contribution

The work presents a completely synthetic data training approach for skull stripping, demonstrating its effectiveness and generalizability across multiple scenarios.

Findings

Achieves comparable accuracy to real-data methods

Works across multi-modal and multi-species cases

Handles pathological cases effectively

Abstract

While many skull stripping algorithms have been developed for multi-modal and multi-species cases, there is still a lack of a fundamentally generalizable approach. We present PUMBA(PUrely synthetic Multimodal/species invariant Brain extrAction), a strategy to train a model for brain extraction with no real brain images or labels. Our results show that even without any real images or anatomical priors, the model achieves comparable accuracy in multi-modal, multi-species and pathological cases. This work presents a new direction of research for any generalizable medical image segmentation task.