Adversarial normalization for multi domain image segmentation

TL;DR

This paper introduces an adversarial normalization method for multi-domain medical image segmentation, learning a shared normalizer across datasets to improve segmentation accuracy and realism.

Contribution

It proposes a novel adversarial normalization approach that jointly normalizes multiple datasets, enhancing segmentation performance and image realism.

Findings

Dice improvements of up to 59.6% over baseline

Effective normalization across infant and adult brain images

Enhanced segmentation accuracy on iSEG and MRBrainS datasets

Abstract

Image normalization is a critical step in medical imaging. This step is often done on a per-dataset basis, preventing current segmentation algorithms from the full potential of exploiting jointly normalized information across multiple datasets. To solve this problem, we propose an adversarial normalization approach for image segmentation which learns common normalizing functions across multiple datasets while retaining image realism. The adversarial training provides an optimal normalizer that improves both the segmentation accuracy and the discrimination of unrealistic normalizing functions. Our contribution therefore leverages common imaging information from multiple domains. The optimality of our common normalizer is evaluated by combining brain images from both infants and adults. Results on the challenging iSEG and MRBrainS datasets reveal the potential of our adversarial normalization approach for segmentation, with Dice improvements of up to 59.6% over the baseline.