Plug-and-Play Logit Fusion for Heterogeneous Pathology Foundation Models

TL;DR

This paper introduces LogitProd, a lightweight, sample-adaptive model fusion method for pathology foundation models that improves performance across diverse tasks without retraining encoders.

Contribution

The paper proposes a novel logit-based fusion strategy that combines heterogeneous models efficiently, with theoretical guarantees and extensive empirical validation.

Findings

LogitProd ranks first on 20 out of 22 benchmarks.

It improves average performance by ~3% over the best single expert.

Achieves multi-expert gains with ~12× lower training cost.

Abstract

Pathology foundation models (FMs) have become central to computational histopathology, offering strong transfer performance across a wide range of diagnostic and prognostic tasks. The rapid proliferation of pathology foundation models creates a model-selection bottleneck: no single model is uniformly best, yet exhaustively adapting and validating many candidates for each downstream endpoint is prohibitively expensive. We address this challenge with a lightweight and novel model fusion strategy, LogitProd, which treats independently trained FM-based predictors as fixed experts and learns sample-adaptive fusion weights over their slide-level outputs. The fusion operates purely on logits, requiring no encoder retraining and no feature-space alignment across heterogeneous backbones. We further provide a theoretical analysis showing that the optimal weighted product fusion is guaranteed to perform at least as well as the best individual expert under the training objective. We systematically evaluate LogitProd on \textbf{22} benchmarks spanning WSI-level classification, tile-level classification, gene mutation prediction, and discrete-time survival modeling. LogitProd ranks first on 20/22 tasks and improves the average performance across all tasks by ~3% over the strongest single expert. LogitProd enables practitioners to upgrade heterogeneous FM-based pipelines in a plug-and-play manner, achieving multi-expert gains with $\sim$12$\times$ lower training cost than feature-fusion alternatives.