Artificial Intelligence in Cardiovascular Imaging: Current Landscape, Clinical Impact, and Future Directions

DISCOVERIES (ISSN 2359-7232), 2025, volume 13

_{ORIGINAL ARTICLE}

_{FULL text (PDF)}

_{CITATION: Sudeep Edpuganti, Department of Medicine, Faculty of Medicine, Tbilisi State Medical University, 33 Vazha Pshvela Avenue, Tbilisi, Georgia; Tel. +995-595775182; Email: edpugantisudeep@gmail.com.}

_{Artificial Intelligence in Cardiovascular Imaging: Current Landscape, Clinical Impact, and Future Directions}

Sudeep Edpuganti ^1,*, Amna Shamim ¹, Vilina Hemant Gangolli ¹, Ranasinghe Arachchige Dona Kashmira Nawodi Weerasekara ¹, Amulya Yellamilli ²

• ¹ Department of Medicine, Faculty of Medicine, Tbilisi State Medical University, Tbilisi, Georgia
• ² Department of Medicine, School of Health Sciences, The University of Georgia, Tbilisi, Georgia

 * Corresponding author: Sudeep Edpuganti, Department of Medicine, Faculty of Medicine, Tbilisi State Medical University, 33 Vazha Pshvela Avenue, Tbilisi, Georgia; Tel. +995-595775182; Email: edpugantisudeep@gmail.com.

_Abstract

Cardiovascular (CV) imaging is rapidly transforming with the advent of artificial intelligence (AI), automating and augmenting diagnostic pipelines in echocardiography, computed tomography (CT), magnetic resonance imaging (MRI), and nuclear imaging. In this review, we summarize recent developments in convolutional neural networks for real-time echocardiographic interpretation, deep learning for coronary artery calcium scoring that achieves near-perfect agreement with manual methods, and AI-driven plaque quantification and stenosis detection on coronary CT angiography, which achieves an accuracy of ≥ 96%. FDA-approved platforms (e.g., Aidoc, HeartFlow, Caption Health) emphasize clinical translation, while automated segmentation and perfusion analysis in cardiac MRI produce Dice coefficients ≥ 0.93. We critically analyze persistent issues, algorithmic bias, explainability, data privacy, regulatory heterogeneity, and medico-legal liability. We also discuss risk-reduction tactics, such as federated learning and human-in-the-loop oversight. Reactive diagnostics will allow proactive, personalized treatment in the future, assuming we look ahead, thanks to multimodal AI, wearable sensors, and predictive analytics. For AI to fully optimize cardiovascular care, thorough validation, open algorithmic design, and interdisciplinary cooperation will be necessary.

_{Access FULL text of the manuscript here: FULL ARTICLE (PDF)}