Multi-Scale Convolutional-Stack Aggregation for Robust White Matter Hyperintensities Segmentation

TL;DR

This paper introduces a multi-scale aggregation model with specialized networks for small lesion segmentation, achieving state-of-the-art results in white matter hyperintensity segmentation and generalizing well across datasets.

Contribution

The paper presents a novel multi-scale aggregation framework and a specialized Stack-Net architecture for improved segmentation of lesions of varying sizes.

Findings

Outperforms state-of-the-art on MICCAI WMH Challenge Dataset

Achieves first place on the challenge's hidden test set

Effective cross-center generalization on MS lesion dataset

Abstract

Segmentation of both large and small white matter hyperintensities/lesions in brain MR images is a challenging task which has drawn much attention in recent years. We propose a multi-scale aggregation model framework to deal with volume-varied lesions. Firstly, we present a specifically-designed network for small lesion segmentation called Stack-Net, in which multiple convolutional layers are connected, aiming to preserve rich local spatial information of small lesions before the sub-sampling layer. Secondly, we aggregate multi-scale Stack-Nets with different receptive fields to learn multi-scale contextual information of both large and small lesions. Our model is evaluated on recent MICCAI WMH Challenge Dataset and outperforms the state-of-the-art on lesion recall and lesion F1-score under 5-fold cross validation. In addition, we further test our pre-trained models on a Multiple Sclerosis lesion dataset with 30 subjects under cross-center evaluation. Results show that the aggregation model is effective in learning multi-scale spatial information.It claimed the first place on the hidden test set after independent evaluation by the challenge organizer. In addition, we further test our pre-trained models on a Multiple Sclerosis lesion dataset with 30 subjects under cross-center evaluation. Results show that the aggregation model is effective in learning multi-scale spatial information.