Automatic Quantitative Coronary Analysis Based on Deep Learning

As a core technique to quantitatively assess the stenosis severity of coronary arteries, quantitative analysis (QCA) is urgently supposed become more automated and intelligent, especially for regions lacking expertise technology. The existing QCA methods highly depend on manual operation, which time-consuming subject personal experience. This study innovatively proposes fully automatic workflow based artificial intelligence (AI-QCA), can quickly accurately make assessment severity. whole AI-QCA mainly consists three parts: boundary-aware segmentation angiogram (CAG) images, AI-enabled artery tree construction, diameter fitting detection. Experiments show that precision, recall, F1 score segmentation, evaluated 1322 CAGs, are 0.866, 0.897, 0.879, respectively. Furthermore, RMSE between assessed by served senior experts, 249 0.064, Pearson coefficient 0.765. Meanwhile, operation time be reduced from tens minutes several seconds AI-QCA. conclusion, proposed able quantify parameters as significant intelligent diagnosis treatment disease.