Efficient KernelSHAP Explanations for Patch-based 3D Medical Image Segmentation

TL;DR

This paper introduces an efficient KernelSHAP framework for 3D medical image segmentation that reduces computation and enhances clinical interpretability by using region-specific feature abstractions and patch caching.

Contribution

The authors develop a region-restricted, cache-accelerated KernelSHAP method tailored for volumetric CT segmentation, improving efficiency and interpretability over existing techniques.

Findings

Caching reduces redundant computation by 15-30%.

Organ-aware supervoxels improve clinical interpretability.

Regular supervoxels maximize perturbation metrics but lack anatomical relevance.

Abstract

Perturbation-based explainability methods such as KernelSHAP provide model-agnostic attributions but are typically impractical for patch-based 3D medical image segmentation due to the large number of coalition evaluations and the high cost of sliding-window inference. We present an efficient KernelSHAP framework for volumetric CT segmentation that restricts computation to a user-defined region of interest and its receptive-field support, and accelerates inference via patch logit caching, reusing baseline predictions for unaffected patches while preserving nnU-Net's fusion scheme. To enable clinically meaningful attributions, we compare three automatically generated feature abstractions within the receptive-field crop: whole-organ units, regular FCC supervoxels, and hybrid organ-aware supervoxels, and we study multiple aggregation/value functions targeting stabilizing evidence (TP/Dice/Soft Dice) or false-positive behavior. Experiments on whole-body CT segmentations show that caching substantially reduces redundant computation (with computational savings ranging from 15% to 30%) and that faithfulness and interpretability exhibit clear trade-offs: regular supervoxels often maximize perturbation-based metrics but lack anatomical alignment, whereas organ-aware units yield more clinically interpretable explanations and are particularly effective for highlighting false-positive drivers under normalized metrics.