Federative ischemic stroke segmentation as alternative to overcome domain-shift multi-institution challenges

TL;DR

This paper presents a federated learning framework for ischemic stroke lesion segmentation in DWI images, enabling multi-institutional collaboration without sharing sensitive data, and demonstrating strong generalization across diverse centers.

Contribution

It introduces a federated approach that leverages deep center-independent representations for stroke lesion segmentation, overcoming domain-shift and data scarcity issues.

Findings

Achieved a DSC of 0.71 across 14 centers

Demonstrated strong generalization to out-of-distribution centers

Outperformed centralized and other federated models

Abstract

Stroke is the second leading cause of death and the third leading cause of disability worldwide. Clinical guidelines establish diffusion resonance imaging (DWI, ADC) as the standard for localizing, characterizing, and measuring infarct volume, enabling treatment support and prognosis. Nonetheless, such lesion analysis is highly variable due to different patient demographics, scanner vendors, and expert annotations. Computational support approaches have been key to helping with the localization and segmentation of lesions. However, these strategies are dedicated solutions that learn patterns from only one institution, lacking the variability to generalize geometrical lesions shape models. Even worse, many clinical centers lack sufficient labeled samples to adjust these dedicated solutions. This work developed a collaborative framework for segmenting ischemic stroke lesions in DWI sequences by sharing knowledge from deep center-independent representations. From 14 emulated healthcare centers with 2031 studies, the FedAvg model achieved a general DSC of $0.71 \pm 0.24$, AVD of $5.29 \pm 22.74$, ALD of $2.16 \pm 3.60$ and LF1 of $0.70 \pm 0.26$ over all centers, outperforming both the centralized and other federated rules. Interestingly, the model demonstrated strong generalization properties, showing uniform performance across different lesion categories and reliable performance in out-of-distribution centers (with DSC of $0.64 \pm 0.29$ and AVD of $4.44 \pm 8.74$ without any additional training).