Journal of the American Heart Association · 2026
Artificial Intelligence-Driven Angiographic Quantification of Fractional Flow Reserve: Proof-of-Concept Validation Study.
СтатьяВалидационное исследованиеМногоцентровое исследованиеПоддержка исследования: вне правительства СШАНазвание 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.
Lee Seung Hun, Gim Dong Hyun, Hwang Doyeon, Kim Hyun Kuk, Cho Hyun, Kim Sung Eun, Cho Sung Woo, Choi Ki Hong, Park Taek Kyu, Yang Jeong Hoon, Song Young Bin, Doh Joon-Hyung, Nam Chang-Wook, Hahn Joo-Yong, Koo Bon-Kwon, Choi Seung-Hyuk, Gwon Hyeon-Cheol, Yoon Hyuck-Jun, Lee Joo Myoung
Аннотация
BACKGROUND: Angiography-based fractional flow reserve (FFR) techniques offer wire-free alternatives but often require manual segmentation and 3-dimensional reconstruction. Although an artificial intelligence-driven approach automates these steps, validation remains limited. This study investigated the diagnostic performance of an artificial intelligence-driven angiography-based FFR (medipixel FFR [MPFFR]), compared with quantitative flow ratio (QFR) in predicting functionally significant coronary artery stenosis defined by FFR ≤0.80.
METHODS: A total of 599 vessels from 452 patients who underwent clinically indicated FFR measurement were prospectively enrolled from 5 university hospitals in Korea. MPFFR used automated processes in frame selection, artificial intelligence contour detection, corresponding points matching, 3-dimensional reconstruction, and analytical hemodynamic modeling. The primary end point was diagnostic accuracy for detecting FFR ≤0.80. Secondary end points were target vessel failure (composite of cardiac death, target-vessel myocardial infarction, and target-vessel revascularization) at 2 years.
RESULTS: Mean analysis time of MPFFR was 12.5±1.7 seconds and manual correction was needed in 32 vessels (5.3%). MPFFR showed similar diagnostic performance with QFR (correlation with FFR; MPFFR versus QFR: R=0.885 versus R=0.860, P for comparison=0.011; area under the curve to predict FFR ≤0.80; 0.949 versus 0.953, P for comparison=0.631). At a median follow-up of 2 years (interquartile range, 1.6-2.6 years), patients with MPFFR ≤0.80 had higher risk of target vessel failure than those with MPFFR >0.80 (4.5% versus 0.8%; adjusted hazard ratio, 5.94 [95% CI, 1.27-27.91]; P=0.024). C-index to predict target vessel failure was comparable between MPFFR and QFR (0.770 versus 0.753, P for comparison=0.469).
CONCLUSIONS: In this multicenter registry, an artificial intelligence-driven angiography-based FFR demonstrated comparable diagnostic accuracy with QFR in identifying functionally significant stenosis, and similar prognostic ability with QFR in terms of target vessel failure at 2 years.
REGISTRATION: Multicenter QFR Registry; Unique Identifier: NCT03791788.
METHODS: A total of 599 vessels from 452 patients who underwent clinically indicated FFR measurement were prospectively enrolled from 5 university hospitals in Korea. MPFFR used automated processes in frame selection, artificial intelligence contour detection, corresponding points matching, 3-dimensional reconstruction, and analytical hemodynamic modeling. The primary end point was diagnostic accuracy for detecting FFR ≤0.80. Secondary end points were target vessel failure (composite of cardiac death, target-vessel myocardial infarction, and target-vessel revascularization) at 2 years.
RESULTS: Mean analysis time of MPFFR was 12.5±1.7 seconds and manual correction was needed in 32 vessels (5.3%). MPFFR showed similar diagnostic performance with QFR (correlation with FFR; MPFFR versus QFR: R=0.885 versus R=0.860, P for comparison=0.011; area under the curve to predict FFR ≤0.80; 0.949 versus 0.953, P for comparison=0.631). At a median follow-up of 2 years (interquartile range, 1.6-2.6 years), patients with MPFFR ≤0.80 had higher risk of target vessel failure than those with MPFFR >0.80 (4.5% versus 0.8%; adjusted hazard ratio, 5.94 [95% CI, 1.27-27.91]; P=0.024). C-index to predict target vessel failure was comparable between MPFFR and QFR (0.770 versus 0.753, P for comparison=0.469).
CONCLUSIONS: In this multicenter registry, an artificial intelligence-driven angiography-based FFR demonstrated comparable diagnostic accuracy with QFR in identifying functionally significant stenosis, and similar prognostic ability with QFR in terms of target vessel failure at 2 years.
REGISTRATION: Multicenter QFR Registry; Unique Identifier: NCT03791788.