MI^2GAN: Generative Adversarial Network for Medical Image Domain Adaptation using Mutual Information Constraint

TL;DR

MI$^2$GAN is a novel generative adversarial network designed for medical image domain adaptation, which preserves image content during translation by maximizing mutual information between disentangled features, improving model generalization.

Contribution

The paper introduces MI$^2$GAN, a new GAN model that maintains image content during domain translation by disentangling features and maximizing mutual information, addressing limitations of existing methods.

Findings

Improves image translation quality in medical domain adaptation.

Enhances generalization performance of deep learning models like U-Net.

Effective in polyp segmentation and optic disc/cup segmentation tasks.

Abstract

Domain shift between medical images from multicentres is still an open question for the community, which degrades the generalization performance of deep learning models. Generative adversarial network (GAN), which synthesize plausible images, is one of the potential solutions to address the problem. However, the existing GAN-based approaches are prone to fail at preserving image-objects in image-to-image (I2I) translation, which reduces their practicality on domain adaptation tasks. In this paper, we propose a novel GAN (namely MI$^2$GAN) to maintain image-contents during cross-domain I2I translation. Particularly, we disentangle the content features from domain information for both the source and translated images, and then maximize the mutual information between the disentangled content features to preserve the image-objects. The proposed MI$^2$GAN is evaluated on two tasks---polyp segmentation using colonoscopic images and the segmentation of optic disc and cup in fundus images. The experimental results demonstrate that the proposed MI$^2$GAN can not only generate elegant translated images, but also significantly improve the generalization performance of widely used deep learning networks (e.g., U-Net).