Modality Bank: Learn multi-modality images across data centers without sharing medical data

TL;DR

ModalityBank enables privacy-preserving multi-modality medical image synthesis across data centers by learning domain-specific parameters, allowing modality completion and improved downstream task performance without sharing raw data.

Contribution

The paper introduces a decentralized learning architecture that synthesizes multi-modality images across data centers without sharing sensitive data, enhancing medical image analysis.

Findings

Effective modality synthesis across data centers.

Improved downstream task performance with synthesized data.

Ability to complete missing modalities across centers.

Abstract

Multi-modality images have been widely used and provide comprehensive information for medical image analysis. However, acquiring all modalities among all institutes is costly and often impossible in clinical settings. To leverage more comprehensive multi-modality information, we propose a privacy secured decentralized multi-modality adaptive learning architecture named ModalityBank. Our method could learn a set of effective domain-specific modulation parameters plugged into a common domain-agnostic network. We demonstrate by switching different sets of configurations, the generator could output high-quality images for a specific modality. Our method could also complete the missing modalities across all data centers, thus could be used for modality completion purposes. The downstream task trained from the synthesized multi-modality samples could achieve higher performance than learning from one real data center and achieve close-to-real performance compare with all real images.