Beyond Occlusion: In Search for Near Real-Time Explainability of CNN-Based Prostate Cancer Classification

TL;DR

The paper tackles the slowdown of occlusion-based explanations in CNN-driven prostate cancer detection on Whole Slide Images by benchmarking fast, single-pass explainability methods (CAM, GradCAM++, HiResCAM, Composite-LRP) across speed, faithfulness, localization, and clinical usefulness. It finds CAM-based approaches, especially HiResCAM, offer near-real-time explanations with equal or better fidelity than occlusion, supported by domain expert input. Implementing HiResCAM reduced the total explainability time for 87 WSIs from about 3.3 days to around 2 hours, illustrating practical impact. The proposed evaluation framework serves as a template for benchmarking explainability methods in digital pathology and related CNN applications.

Abstract

Deep neural networks are starting to show their worth in critical applications such as assisted cancer diagnosis. However, for their outputs to get accepted in practice, the results they provide should be explainable in a way easily understood by pathologists. A well-known and widely used explanation technique is occlusion, which, however, can take a long time to compute, thus slowing the development and interaction with pathologists. In this work, we set out to find a faster replacement for occlusion in a successful system for detecting prostate cancer. Since there is no established framework for comparing the performance of various explanation methods, we first identified suitable comparison criteria and selected corresponding metrics. Based on the results, we were able to choose a different explanation method, which cut the previously required explanation time at least by a factor of 10, without any negative impact on the quality of outputs. This speedup enables rapid iteration in model development and debugging and brings us closer to adopting AI-assisted prostate cancer detection in clinical settings. We propose that our approach to finding the replacement for occlusion can be used to evaluate candidate methods in other related applications.

Figures (4)

• Figure 1: Faithfulness according to the ROAD method.
• Figure 2: Agreement with pathologist annotations.
• Figure 3: Agreement with occlusion-produced maps.
• Figure 4: Annotated saliency map produced by HiResCAM.