Cross-Domain Distribution Alignment for Segmentation of Private Unannotated 3D Medical Images

TL;DR

This paper presents a novel source-free unsupervised domain adaptation method for 3D medical image segmentation, addressing privacy and annotation challenges by using pseudo-labels for model refinement, achieving state-of-the-art results.

Contribution

Introduces a new source-free UDA approach that estimates source distribution and employs pseudo-labels for improved 3D medical image segmentation.

Findings

Achieves state-of-the-art performance on real-world 3D medical datasets.

Effectively handles privacy constraints in medical image segmentation.

Enhances model accuracy through self-training with pseudo-labels.

Abstract

Manual annotation of 3D medical images for segmentation tasks is tedious and time-consuming. Moreover, data privacy limits the applicability of crowd sourcing to perform data annotation in medical domains. As a result, training deep neural networks for medical image segmentation can be challenging. We introduce a new source-free Unsupervised Domain Adaptation (UDA) method to address this problem. Our idea is based on estimating the internally learned distribution of a relevant source domain by a base model and then generating pseudo-labels that are used for enhancing the model refinement through self-training. We demonstrate that our approach leads to SOTA performance on a real-world 3D medical dataset.