Federated Distillation for Whole Slide Image via Gaussian-Mixture Feature Alignment and Curriculum Integration

TL;DR

FedHD introduces a federated distillation framework for whole slide image analysis that aligns features via Gaussian mixtures and uses curriculum strategies to improve collaborative pathology modeling across institutions.

Contribution

The paper presents FedHD, a novel federated distillation approach that handles heterogeneity in digital pathology, using Gaussian-mixture feature alignment and curriculum integration for improved WSI analysis.

Findings

FedHD outperforms state-of-the-art federated and distillation methods on multiple datasets.

The framework effectively handles heterogeneity across institutions in digital pathology.

Synthetic feature alignment enhances model performance without sharing raw data.

Abstract

Federated learning (FL) offers a promising framework for collaborative digital pathology by enabling model training across institutions. However, real-world deployments face heterogeneity arising from diverse multiple instance learning (MIL) architectures and heterogeneous feature extractors across institutions. We propose FedHD, a novel FL framework that performs local Gaussian-mixture feature alignment tailored for WSI analysis. Instead of exchanging model parameters, each client independently distills semantically rich synthetic feature representations aligned with the distribution of real WSIs. To preserve diagnostic diversity, FedHD adopts a one-to-one distillation strategy, generating a synthetic counterpart for each real slide to avoid over-compression. During federation, a curriculum-based integration strategy progressively incorporates cross-site synthetic features into local training once performance plateaus. Furthermore, an optional interpretation module reconstructs pseudo-patches from synthetic embeddings, enhancing transparency. FedHD is architecture-agnostic, privacy-preserving, and supports personalized yet collaborative training across diverse institutions. Experiments on TCGA-IDH, CAMELYON16, and CAMELYON17 show that FedHD consistently outperforms state-of-the-art federated and distillation baselines.