Department of Automation, Tsinghua University, Beijing, China.
Shenzhen International Graduate School, Tsinghua University, Shenzhen, Guangdong, China.
BMJ Health Care Inform. 2024 Jun 3;31(1):e100942. doi: 10.1136/bmjhci-2023-100942.
Current approaches for initial coronary artery disease (CAD) assessment rely on pretest probability (PTP) based on risk factors and presentations, with limited performance. Infrared thermography (IRT), a non-contact technology that detects surface temperature, has shown potential in assessing atherosclerosis-related conditions, particularly when measured from body regions such as faces. We aim to assess the feasibility of using facial IRT temperature information with machine learning for the prediction of CAD.
Individuals referred for invasive coronary angiography or coronary CT angiography (CCTA) were enrolled. Facial IRT images captured before confirmatory CAD examinations were used to develop and validate a deep-learning IRT image model for detecting CAD. We compared the performance of the IRT image model with the guideline-recommended PTP model on the area under the curve (AUC). In addition, interpretable IRT tabular features were extracted from IRT images to further validate the predictive value of IRT information.
A total of 460 eligible participants (mean (SD) age, 58.4 (10.4) years; 126 (27.4%) female) were included. The IRT image model demonstrated outstanding performance (AUC 0.804, 95% CI 0.785 to 0.823) compared with the PTP models (AUC 0.713, 95% CI 0.691 to 0.734). A consistent level of superior performance (AUC 0.796, 95% CI 0.782 to 0.811), achieved with comprehensive interpretable IRT features, further validated the predictive value of IRT information. Notably, even with only traditional temperature features, a satisfactory performance (AUC 0.786, 95% CI 0.769 to 0.803) was still upheld.
In this prospective study, we demonstrated the feasibility of using non-contact facial IRT information for CAD prediction.
目前，初始冠状动脉疾病 (CAD) 的评估方法依赖于基于风险因素和表现的术前概率 (PTP)，但其性能有限。红外热成像 (IRT) 是一种检测表面温度的非接触技术，已显示出在评估动脉粥样硬化相关疾病方面的潜力，尤其是当从面部等身体部位进行测量时。我们旨在评估使用面部 IRT 温度信息和机器学习进行 CAD 预测的可行性。
纳入了因疑似 CAD 而接受有创冠状动脉造影或冠状动脉 CT 血管造影 (CCTA) 检查的个体。在进行确认性 CAD 检查之前，采集面部 IRT 图像，用于开发和验证用于检测 CAD 的深度学习 IRT 图像模型。我们比较了 IRT 图像模型与指南推荐的 PTP 模型在曲线下面积 (AUC) 上的性能。此外，还从 IRT 图像中提取可解释的 IRT 表格特征，以进一步验证 IRT 信息的预测价值。
共纳入 460 名符合条件的参与者（平均 (SD) 年龄为 58.4 (10.4) 岁；126 名 [27.4%] 为女性）。IRT 图像模型与 PTP 模型相比，具有出色的性能（AUC 为 0.804，95%CI 为 0.785 至 0.823）（AUC 为 0.713，95%CI 为 0.691 至 0.734）。使用全面的可解释 IRT 特征，进一步验证了 IRT 信息的预测价值，可实现一致的卓越性能（AUC 为 0.796，95%CI 为 0.782 至 0.811）。值得注意的是，即使仅使用传统的温度特征，仍能保持令人满意的性能（AUC 为 0.786，95%CI 为 0.769 至 0.803）。
在这项前瞻性研究中，我们证明了使用非接触式面部 IRT 信息进行 CAD 预测的可行性。
