Interpretable Vertebral Fracture Diagnosis

TL;DR

This paper investigates how neural networks diagnose vertebral fractures in CT images, aiming to improve interpretability and trust by identifying and analyzing the concepts used by the models.

Contribution

It introduces a framework for associating neural network neurons with clinically relevant concepts, enhancing explainability in vertebral fracture diagnosis.

Findings

Identified concepts correlated with correct diagnoses

Analyzed concepts leading to false positives

Proposed methods improve interpretability of neural networks

Abstract

Do black-box neural network models learn clinically relevant features for fracture diagnosis? The answer not only establishes reliability quenches scientific curiosity but also leads to explainable and verbose findings that can assist the radiologists in the final and increase trust. This work identifies the concepts networks use for vertebral fracture diagnosis in CT images. This is achieved by associating concepts to neurons highly correlated with a specific diagnosis in the dataset. The concepts are either associated with neurons by radiologists pre-hoc or are visualized during a specific prediction and left for the user's interpretation. We evaluate which concepts lead to correct diagnosis and which concepts lead to false positives. The proposed frameworks and analysis pave the way for reliable and explainable vertebral fracture diagnosis.