MS Lesion Segmentation: Revisiting Weighting Mechanisms for Federated Learning

TL;DR

This paper introduces a novel federated learning framework for MS lesion segmentation that employs re-weighting mechanisms to improve performance across diverse datasets, surpassing traditional methods and even centralized training.

Contribution

The paper proposes the first federated learning MS lesion segmentation framework with learnable and lesion-volume-based re-weighting mechanisms, enhancing accuracy across heterogeneous data sources.

Findings

Outperforms other FL methods significantly.

Can surpass centralized training performance.

Improves brain volume difference estimation after lesion inpainting.

Abstract

Federated learning (FL) has been widely employed for medical image analysis to facilitate multi-client collaborative learning without sharing raw data. Despite great success, FL's performance is limited for multiple sclerosis (MS) lesion segmentation tasks, due to variance in lesion characteristics imparted by different scanners and acquisition parameters. In this work, we propose the first FL MS lesion segmentation framework via two effective re-weighting mechanisms. Specifically, a learnable weight is assigned to each local node during the aggregation process, based on its segmentation performance. In addition, the segmentation loss function in each client is also re-weighted according to the lesion volume for the data during training. Comparison experiments on two FL MS segmentation scenarios using public and clinical datasets have demonstrated the effectiveness of the proposed method by outperforming other FL methods significantly. Furthermore, the segmentation performance of FL incorporating our proposed aggregation mechanism can exceed centralised training with all the raw data. The extensive evaluation also indicated the superiority of our method when estimating brain volume differences estimation after lesion inpainting.