人工智能自动可解释冠状动脉造影解读辅助筛选CABG候选者
本研究开发了一种自动化计算机视觉模型,用于从冠状动脉造影视频中识别符合指南的CABG候选者。模型分析了349例患者的4472个造影视频片段,检测狭窄病变并判断多支病变是否符合CABG标准,与造影报告相比准确率74%、阳性预测值61%、阴性预测值87%。模型误判正常20例(6%),推荐PCI而实际建议CABG 31例,主要因左主干或回旋支病变识别不足。
要点
- 模型可自动检测左主干、前降支、回旋支和右冠状动脉狭窄病变。
- 模型判断多支病变符合CABG标准的准确率74%、阳性预测值61%、阴性预测值87%。
- 误判原因包括选错最大对比帧(6%误判正常)和左主干/回旋支病变识别不足(31例误推PCI)。
核心结论
- 自动化模型未来改进后可辅助造影时视觉判读,提高质量控制和指南导向血运重建策略。
研究设计与样本
- 回顾性研究,筛选2018至2023年首次CABG患者的造影视频与报告,以临床确定的正常、单支和双支病变为参照,比较模型与造影报告识别CABG指征的效能。
- 共分析349例患者的4472个造影视频片段,识别682个病变。
- 平均分析时间51.1秒/研究,或4.6秒/视频片段。
主要发现
- 模型检测左主干、前降支、回旋支和右冠状动脉狭窄病变,判断多支病变是否符合CABG标准,与造影报告相比准确率74%、阳性预测值61%、阴性预测值87%。
- 模型错误预测造影正常20例(6%),因选择错误的最大对比帧。
- 模型推荐PCI而实际建议CABG 31例,因左主干或回旋支病变识别不足。
关键数值与推荐
- 参照美国心脏协会/美国心脏病学会2021指南判断CABG指征。
- 模型判断多支病变符合CABG标准与造影报告相比:准确率74%、阳性预测值61%、阴性预测值87%。
临床怎么用
- 模型可作为造影时视觉判读的辅助工具,帮助识别符合指南的CABG候选者,但需注意左主干和回旋支病变的识别局限性。
局限性
- 模型误判正常占6%,因选错最大对比帧;推荐PCI而实际需CABG 31例,因左主干或回旋支病变识别不足。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
OBJECTIVES: Visual estimation of coronary artery stenosis on angiography is subject to human error. Although machine learning may facilitate more accurate interpretation, clinical utility has been limited by lack of human interpretable models. We developed an automated computer vision model to identify candidates for coronary artery bypass (CABG) from coronary angiograms. METHODS: Medical records for primary CABG between 2018 and 2023 were screened for coronary angiogram video with angiographic and operative reports. A clinically determined reference group of angiographically normal, single-, and double-vessel disease was compared with automated angiogram reports to identify patients with indications for CABG per American Heart Association/American College of Cardiology 2021 guidelines. RESULTS: A total of 4472 angiographic video clips from 349 patients were analyzed identifying 682 lesions. Mean analysis was 51.1 seconds/study, or 4.6 seconds/video clip. Detection algorithm results were compared with original reports/images in cases where the model differed in recommendation. The model detected stenotic lesions in the left main, left anterior descending, circumflex, and right coronary arteries and calculated whether multivessel disease met criteria for CABG (accuracy: 74%, positive predicted value: 61%, and negative predictive value: 87%) compared with lesions documented in angiogram reports. Incorrect angiographically normal prediction occurred in 20 cases (6%) due to the selection of an incorrect maximum contrast frame. Prediction of percutaneous intervention when bypass was recommended (n = 31) was due to underrecognition of a left main or circumflex lesion. CONCLUSIONS: With future improvements, automated identification of CABG candidates could augment visual reading at the time of angiography to improve quality control and guideline-directed revascularization strategies.
引用
- Liu TX, McCarthy PM, Manerikar A, et al. Automated and explainable coronary angiogram interpretation for selection of coronary artery bypass grafting candidates using artificial intelligence. J Thorac Cardiovasc Surg. 2025. PMID: 40818785. doi:10.1016/j.jtcvs.2025.08.005
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。