[Paper Review] Artificial Intelligence in Image-based Cardiovascular Disease Analysis

Xin Wang, Hu, Mingcheng+2 more|arXiv (Cornell University)|Feb 4, 2024

Radiomics and Machine Learning in Medical Imaging6 citations

TL;DR

A comprehensive survey of how AI is used in image-based cardiovascular disease analysis, categorized by non-vessel and vessel heart structures, imaging modalities, public datasets, and future directions.

ABSTRACT

Recent advancements in Artificial Intelligence (AI) have significantly influenced the field of Cardiovascular Disease (CVD) analysis, particularly in image-based diagnostics. Our paper presents an extensive review of AI applications in image-based CVD analysis, offering insights into its current state and future potential. We systematically categorize the literature based on the primary anatomical structures related to CVD, dividing them into non-vessel structures (such as ventricles and atria) and vessel structures (including the aorta and coronary arteries). This categorization provides a structured approach to explore various imaging modalities like Computed tomography (CT) and Magnetic Resonance Imaging (MRI), which are commonly used in CVD research. Our review encompasses these modalities, giving a broad perspective on the diverse imaging techniques integrated with AI for CVD analysis. We conclude with an examination of the challenges and limitations inherent in current AI-based CVD analysis methods and suggest directions for future research to overcome these hurdles.

Motivation & Objective

• Categorize AI literature on CVD image analysis by anatomical structures (non-vessel vs vessel) and imaging modality.
• Review imaging modalities (MRI, CT, X-ray, ultrasound, nuclear imaging) and AI tasks (segmentation, classification, risk prediction, decision support).
• Survey public datasets and code repositories to enhance reproducibility and collaboration.
• Discuss current challenges, limitations, and future research directions in AI-based cardiovascular imaging.

Proposed method

• Systematically categorize studies based on anatomical heart structures (non-vessel and vessel) and analyze linked AI methods.
• Integrate discussion of imaging modalities with AI applications, highlighting structural and functional analyses.
• Summarize AI tasks in cardiac imaging such as segmentation, registration, feature extraction, and downstream quantification.
• Provide summaries of publicly available datasets and code repositories to support reproducibility.
• Identify challenges and propose future research directions, including multi-modal data integration.

Experimental results

Research questions

• RQ1 How is AI applied to non-vessel versus vessel cardiovascular structures in imaging data?
• RQ2 What imaging modalities and AI tasks dominate the current literature on image-based CVD analysis?
• RQ3 What public datasets and code repositories are accessible to researchers for reproducibility and benchmarking?
• RQ4 What are the main challenges and future directions for AI in image-based cardiovascular imaging?

Key findings

• AI advances have improved segmentation, disease classification, risk prediction, and clinical decision support across cardiac imaging modalities.
• There is a trend toward integrating imaging with population-based data and imaging genetics to enhance understanding of CVDs.
• The survey consolidates publicly available datasets and code repositories to promote reproducibility and collaboration.
• There are notable challenges and limitations in current AI-based cardiovascular imaging methods, with recommended future research directions.
• This work is the first comprehensive survey to address both structural and functional aspects of cardiac imaging in AI contexts.