Artificial intelligence using a deep learning versus expert computed tomography human reading in calcium score and coronary artery calcium data and reporting system classification

This study evaluated the accuracy of artificial intelligence (AI) using Coreline AVIEW software versus expert human CT reading for coronary artery calcium (CAC) scoring and classification. The study involved 100 non-contrast calcium score images, processed both by AI and human level-3 CT readers, and compared their results using the Pearson correlation index. The CAC-DRS classification system was applied to determine any reclassification causes. The results showed a significant correlation between AI and human readings (R = 0.996), but 14% of patients had their CAC-DRS category reclassified, mainly due to AI underestimations and overestimations in certain areas. Despite these discrepancies, the AI software demonstrated excellent correlation with human readings and occasionally detected calcium missed by humans. The study concludes that while AI shows strong potential, further algorithm optimization is needed for better sensitivity and specificity, especially for minimal calcium volumes, to enhance its clinical utility in CAC scoring.