ARHNet: Adaptive Region Harmonization for Lesion-aware Augmentation to Improve Segmentation Performance

TL;DR

ARHNet introduces an adaptive region harmonization framework that enhances lesion-aware augmentation by reducing intensity disparities and artifacts, thereby significantly improving brain lesion segmentation accuracy in MRI scans.

Contribution

The paper presents a novel ARHNet framework with an adaptive region harmonization module that dynamically aligns foreground and background features for realistic synthetic image generation.

Findings

ARHNet outperforms existing harmonization methods on ATLAS 2.0 dataset.

Synthetic images generated by ARHNet improve segmentation performance.

The method effectively reduces intensity disparities and boundary artifacts.

Abstract

Accurately segmenting brain lesions in MRI scans is critical for providing patients with prognoses and neurological monitoring. However, the performance of CNN-based segmentation methods is constrained by the limited training set size. Advanced data augmentation is an effective strategy to improve the model's robustness. However, they often introduce intensity disparities between foreground and background areas and boundary artifacts, which weakens the effectiveness of such strategies. In this paper, we propose a foreground harmonization framework (ARHNet) to tackle intensity disparities and make synthetic images look more realistic. In particular, we propose an Adaptive Region Harmonization (ARH) module to dynamically align foreground feature maps to the background with an attention mechanism. We demonstrate the efficacy of our method in improving the segmentation performance using real and synthetic images. Experimental results on the ATLAS 2.0 dataset show that ARHNet outperforms other methods for image harmonization tasks, and boosts the down-stream segmentation performance. Our code is publicly available at https://github.com/King-HAW/ARHNet.