Investigating Explainable AI for Enhanced Atherosclerosis Detection and Decision Transparency

EasyChair Preprint no. 11570

Abstract

Atherosclerosis, a cardiovascular disease, is commonly diagnosed through non-invasive imaging methods like Coronary CT Angiography (CCTA). Deep learning algorithms have demonstrated remarkable potential in assisting with the classification of CCTA images. However, the inherent opacity in the decision-making processes of black-box AI models presents a significant challenge in the medical field. Healthcare professionals require clear insights into the rationale behind AI system recommendations. In this research, we investigate the advantages of Explainable AI (XAI) algorithms in the context of a previously established deep learning model for atherosclerosis classification from CCTA images. Our study not only highlights the capability of XAI in elucidating the model's decision-making process but also showcases its potential in identifying misclassified cases, thereby contributing to the refinement and enhancement of the model's performance.

Keyphrases: Atherosclerosis, Coronary CT angiography, deep learning, Explainable Artificial Intelligence, GradCam., XAI

@Booklet{EasyChair:11570,
  author = {Amel Laidi and Mohammed Ammar},
  title = {Investigating Explainable AI for Enhanced Atherosclerosis Detection and Decision Transparency},
  howpublished = {EasyChair Preprint no. 11570},

  year = {EasyChair, 2023}}