Explainable Artificial Intelligence Paradigm in Cardiovascular Plaque Tissue Characterization and Classification for Risk Prediction

Atherosclerosis is the predominant cardiovascular disease caused by the blockage of blood vessels in various parts of the body which restricts the blood flow and can result in life-threatening conditions like stroke and heart attack. Several studies prove cardiovascular diseases are the major cause of fatality globally. Early detection of this precarious situation assumes importance in the proper administration of treatment methodologies. Artificial intelligence-based models can be effectively used for the segmentation, characterization, and classification of medical images to support clinicians in predicting the severity of blockages with improved accuracy and transparency. Conventional artificial neural network frameworks are perceived as ‘black-boxes’ as they are difficult to analyze and interpret by medical practitioners. Hence there is a growing interest in developing explainable ‘white-box’ techniques that are fully transparent and easily interpretable for medical and biological image processing to complement the existing approaches. This work covers a detailed literature review on how explainable artificial intelligence architectures enhance plaque segmentation, characterization, and classification. Also discusses the different modalities used for image acquisition, details of how model accuracy is calculated, limitations of the study, and future directions.