Explainability Through Human-Centric Design for XAI in Lung Cancer Detection

TL;DR

This work addresses the opacity of deep learning in chest X-ray lung cancer detection by evaluating existing XAI approaches and introducing XpertXAI, an expert-driven concept bottleneck model. By curating clinically meaningful concepts from radiology reports and enforcing them as supervision through a two-model CBM, XpertXAI achieves superior concept and label-prediction performance and produces explanations that align more closely with radiologist reasoning than post-hoc methods or unsupervised CBMs. The approach demonstrates that integrating domain expertise into the explainability design yields clinically actionable insights and better generalizability to multi-pathology tasks, with potential to scale to other imaging modalities. The findings underscore the value of human-centric XAI design for trustworthy AI in high-stakes medical diagnostics and suggest future work extending to CT/MRI and broader expert validation.

Abstract

Deep learning models have shown promise in lung pathology detection from chest X-rays, but widespread clinical adoption remains limited due to opaque model decision-making. In prior work, we introduced ClinicXAI, a human-centric, expert-guided concept bottleneck model (CBM) designed for interpretable lung cancer diagnosis. We now extend that approach and present XpertXAI, a generalizable expert-driven model that preserves human-interpretable clinical concepts while scaling to detect multiple lung pathologies. Using a high-performing InceptionV3-based classifier and a public dataset of chest X-rays with radiology reports, we compare XpertXAI against leading post-hoc explainability methods and an unsupervised CBM, XCBs. We assess explanations through comparison with expert radiologist annotations and medical ground truth. Although XpertXAI is trained for multiple pathologies, our expert validation focuses on lung cancer. We find that existing techniques frequently fail to produce clinically meaningful explanations, omitting key diagnostic features and disagreeing with radiologist judgments. XpertXAI not only outperforms these baselines in predictive accuracy but also delivers concept-level explanations that better align with expert reasoning. While our focus remains on explainability in lung cancer detection, this work illustrates how human-centric model design can be effectively extended to broader diagnostic contexts - offering a scalable path toward clinically meaningful explainable AI in medical diagnostics.

Figures (6)

• Figure 1: For LIME, SHAP and Grad-CAM, (a) shows mean pixel overlap between techniques on the MIMIC-CXR test set. (b) shows mean medical ground truth captured on the VinDr-CXR test set.
• Figure 2: Analysis by an expert radiologist of explanations generated for a subset of 40 cancerous and 20 healthy chest X-rays. The expert was asked to score explanations between 0 and 3 (see legend). XpertXAI refers to the DT architecture.
• Figure 3: Explanations generated by each XAI technique for a cancerous chest X-ray. (a) shows the ground truth hilar mass. (b) shows LIME (most important = intense green). (c) shows SHAP (most important = green). (d) shows Grad-CAM (most important = red). (e) shows XCBs (concepts with the 5 highest absolute values, positive (+) or negative (-)). (f) shows the radiology report generated by CXR-LLaVA. (g) shows XpertXAI (DT architecture), with the top 2 scoring concepts.
• Figure 4: Example of a cancerous radiology report from the MIMIC-CXR dataset. Clinical concepts extracted ('Nodule') are highlighted by bounding box. Note the negative mention in the final paragraph is not extracted.
• Figure 5: The pipeline for XpertXAI: We take a chest X-ray as input, which is fed into a trained concept prediction model, producing prediction scores for a pre-set list of clinical concepts. These scores are then input to a trained label prediction model, which outputs the binary classification label. Explanations supporting model decisions are shown as the two highest scoring concepts.