Comparison of Accuracy of Coronary Artery Calcium Score between Artificial Intelligence and Manual Method

Abstract

Agaston Score used as indicator of quantative coronary artery calcification is calculated manually by Human being. The accuracy of the values measured using artificial intelligence programs and the manual measurement method, which is an existing conventional method, was compared. The comparison of the average value of the results of 136 people using the initial version showed a statistically significant difference, and the results of 156 people using the upgraded version showed no statistically significant difference. In the case of initial version of the program, the accuracy of the division of the anatomical structure was poor, resulting in a difference in the result values. After that, in the case of the group using the improved version, the problem was improved, resulting in no statistically significant difference. It was possible to confirm accuracy and the possibility of development of deep learning based AI.