Causal Scoring Medical Image Explanations: A Case Study On Ex-vivo Kidney Stone Images

TL;DR

This paper introduces a quantitative method called Causal Explanation Score (CaES) to measure the causal relationship between image features and classifier outputs, demonstrating improved causal measurement with model-generated masks over human annotations in kidney stone images.

Contribution

The paper proposes CaES, a novel quantitative metric for assessing causal relationships in explainable AI, validated on ex-vivo kidney stone images.

Findings

CaES improves causal relationship measurement with model-generated masks.

Model-based masks outperform human annotations in causal assessment.

The method enhances understanding of model explanations in medical imaging.

Abstract

On the promise that if human users know the cause of an output, it would enable them to grasp the process responsible for the output, and hence provide understanding, many explainable methods have been proposed to indicate the cause for the output of a model based on its input. Nonetheless, little has been reported on quantitative measurements of such causal relationships between the inputs, the explanations, and the outputs of a model, leaving the assessment to the user, independent of his level of expertise in the subject. To address this situation, we explore a technique for measuring the causal relationship between the features from the area of the object of interest in the images of a class and the output of a classifier. Our experiments indicate improvement in the causal relationships measured when the area of the object of interest per class is indicated by a mask from an explainable method than when it is indicated by human annotators. Hence the chosen name of Causal Explanation Score (CaES)