Determination of lipid-rich plaques by artificial intelligence-enabled quantitative computed tomography using near-infrared spectroscopy as reference

• DOI: 10.1016/j.atherosclerosis.2023.117363
• Publication Date: 2023-10-29T08:52:11Z
• Authors: Hiroyuki Omori; Hitoshi Matsuo; Shinichiro Fujimoto; Yoshihiro Sobue; Yui Nozaki; Gaku Nakazawa; Kuniaki Takahashi; Kazuhiro Osawa; Ryo Okubo; Umihiko Kaneko; Hideyuki Sato; Takashi Kajiya; Toru Miyoshi; Keishi Ichikawa; Mitsunori Abe; Toshiro Kitagawa; Hiroki Ikenaga; Mike Saji; Nobuo Iguchi; Takeshi Ijichi; Hiroshi Mikamo; Akira Kurata; Masao Moroi; Raisuke Iijima; Shant Malkasian; Tami Crabtree; James K. Min; James P. Earls; Rine Nakanishi
• Keywords: Spectroscopy, Near-Infrared; Artificial Intelligence; Computed Tomography Angiography; Predictive Value of Tests; Humans; Coronary Artery Disease; Tomography, X-Ray Computed; Coronary Angiography; Coronary Vessels; Lipids; Plaque, Atherosclerotic; Ultrasonography, Interventional; 3. Good health

Artificial intelligence quantitative CT (AI-QCT) determines coronary plaque morphology with high efficiency and accuracy. Yet, its performance to quantify lipid-rich remains unclear. This study investigated the of AI-QCT for detection low-density noncalcified (LD-NCP) using near-infrared spectroscopy-intravascular ultrasound (NIRS-IVUS).The INVICTUS Registry is a multi-center registry enrolling patients undergoing clinically indicated angiography IVUS, NIRS-IVUS, or optical coherence tomography. We assessed various Hounsfield unit (HU) volume thresholds LD-NCP maxLCBI4mm ≥ 400 as reference standard correlation vessel area, lumen burden, lesion length between IVUS.This included 133 atherosclerotic plaques from 47 who underwent NIRS-IVUS The area under curve LD-NCP<30HU was 0.97 (95% confidence interval [CI]: 0.93-1.00] an optimal threshold 2.30 mm3. Accuracy, sensitivity, specificity were 94% CI: 88-96%], 93% 76-98%), 88-98%), respectively, <30 HU 2.3 mm3, versus 42%, 100%, 27% >0 mm3 (p < 0.001 accuracy specificity). strongly correlated IVUS measurements; (r2 = 0.87), burden 0.78) 0.88), respectively.AI-QCT demonstrated excellent diagnostic in detecting significant standard. Additionally, derived respective measurements.