International journal of cardiology. Heart & vasculature · 2026
Automated segmentation and clinical characterization of layered plaque in intracoronary optical coherence tomography.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
van der Waerden Ruben G A, van der Zande Joske L, Saitta Simone, Volleberg Rick H J A, Mol Jan-Quinten, Cancian Pierandrea, Luttikholt Thijs J, Gu Xiaojin, Heil Leah, van Leeuwen Maarten, Meuwissen Martijn, Protopopov Alexey V, Laanmets Peep, Krestyaninov Oleg V, Dennert Robert, Oemrawsingh Rohit M, van Kuijk Jan-Peter, Arkenbout Karin, van der Heijden Dirk J, Kedhi Elvin, Roleder Tomasz, Sánchez Clara I, van Ginneken Bram, Išgum Ivana, Thannhauser Jos, van Royen Niels
Аннотация
BACKGROUND: Layered plaque, representing organized thrombus overlying a previously disrupted plaque, is a marker of coronary destabilization and plaque vulnerability. Despite its clinical relevance, identifying layered plaque remains expert dependent with interobserver variability, limiting large-scale assessment. We developed and validated an artificial intelligence (AI) model for layered plaque segmentation, and assessed occurrence and associations with plaque vulnerability in non-culprit lesions.
METHODS: Optical coherence tomography (OCT) pullbacks from the PECTUS-obs and ORANGE datasets were used to develop an enhanced version of our validated multiclass segmentation model (OCT-AID), by adding layered plaque as an additional pixel-level class. Pullbacks were divided into a training set (n = 4121 frames, 493 with layered plaque) and an independent held-out test set (n = 433 frames, 71 with layered plaque). Using this model, prevalence of layered plaque and associations with plaque vulnerability features were studied in patients with non-culprit lesions (PECTUS-obs data; 414 patients, 488 lesions).
RESULTS: The model achieved a Dice coefficient of 0.68 ± 0.27 for layered plaque segmentation, with pullback-level sensitivity of 96.2% and specificity of 75.0%. At least one layered plaque was present in 366/414 patients (88.4%). Presence of layered plaque was higher in STEMI patients (p = 0.018). Lesions containing layered plaques showed higher prevalence of lipid-rich plaque (76.4% vs. 60.9%, p = 0.008), thin-cap fibroatheroma (29.2% vs. 15.6%, p = 0.016), macrophage accumulation (25.0% vs 9.4%, p = 0.010), and cholesterol clefts (0.0% vs. 8.3%; p = 0.009).
CONCLUSIONS: Automated OCT-based identification of layered plaque using AI is feasible. In non-culprit coronary lesions, layered plaque was highly prevalent and associated with established markers of plaque vulnerability.
METHODS: Optical coherence tomography (OCT) pullbacks from the PECTUS-obs and ORANGE datasets were used to develop an enhanced version of our validated multiclass segmentation model (OCT-AID), by adding layered plaque as an additional pixel-level class. Pullbacks were divided into a training set (n = 4121 frames, 493 with layered plaque) and an independent held-out test set (n = 433 frames, 71 with layered plaque). Using this model, prevalence of layered plaque and associations with plaque vulnerability features were studied in patients with non-culprit lesions (PECTUS-obs data; 414 patients, 488 lesions).
RESULTS: The model achieved a Dice coefficient of 0.68 ± 0.27 for layered plaque segmentation, with pullback-level sensitivity of 96.2% and specificity of 75.0%. At least one layered plaque was present in 366/414 patients (88.4%). Presence of layered plaque was higher in STEMI patients (p = 0.018). Lesions containing layered plaques showed higher prevalence of lipid-rich plaque (76.4% vs. 60.9%, p = 0.008), thin-cap fibroatheroma (29.2% vs. 15.6%, p = 0.016), macrophage accumulation (25.0% vs 9.4%, p = 0.010), and cholesterol clefts (0.0% vs. 8.3%; p = 0.009).
CONCLUSIONS: Automated OCT-based identification of layered plaque using AI is feasible. In non-culprit coronary lesions, layered plaque was highly prevalent and associated with established markers of plaque vulnerability.