Division of Artificial Intelligence in Medicine, Department of Medicine, Imaging, and Biomedical Sciences, Cedars-Sinai Medical Center, Los Angeles, California.
Department of Cardiac Sciences, University of Calgary and Libin Cardiovascular Institute, Calgary, Alberta, Canada.
J Nucl Med. 2022 Nov;63(11):1768-1774. doi: 10.2967/jnumed.121.263686. Epub 2022 May 5.
Artificial intelligence may improve accuracy of myocardial perfusion imaging (MPI) but will likely be implemented as an aid to physician interpretation rather than an autonomous tool. Deep learning (DL) has high standalone diagnostic accuracy for obstructive coronary artery disease (CAD), but its influence on physician interpretation is unknown. We assessed whether access to explainable DL predictions improves physician interpretation of MPI. We selected a representative cohort of patients who underwent MPI with reference invasive coronary angiography. Obstructive CAD, defined as stenosis ≥50% in the left main artery or ≥70% in other coronary segments, was present in half of the patients. We used an explainable DL model (CAD-DL), which was previously developed in a separate population from different sites. Three physicians interpreted studies first with clinical history, stress, and quantitative perfusion, then with all the data plus the DL results. Diagnostic accuracy was assessed using area under the receiver-operating-characteristic curve (AUC). In total, 240 patients with a median age of 65 y (interquartile range 58-73) were included. The diagnostic accuracy of physician interpretation with CAD-DL (AUC 0.779) was significantly higher than that of physician interpretation without CAD-DL (AUC 0.747, = 0.003) and stress total perfusion deficit (AUC 0.718, < 0.001). With matched specificity, CAD-DL had higher sensitivity when operating autonomously compared with readers without DL results ( < 0.001), but not compared with readers interpreting with DL results ( = 0.122). All readers had numerically higher accuracy with CAD-DL, with AUC improvement 0.02-0.05, and interpretation with DL resulted in overall net reclassification improvement of 17.2% (95% CI 9.2%-24.4%, < 0.001). Explainable DL predictions lead to meaningful improvements in physician interpretation; however, the improvement varied across the readers, reflecting the acceptance of this new technology. This technique could be implemented as an aid to physician diagnosis, improving the diagnostic accuracy of MPI.
人工智能可能会提高心肌灌注成像（MPI）的准确性，但可能会被用作医生解释的辅助工具，而不是自主工具。深度学习（DL）对阻塞性冠状动脉疾病（CAD）具有很高的独立诊断准确性，但它对医生解释的影响尚不清楚。我们评估了是否可以访问可解释的 DL 预测来改善 MPI 的医生解释。
我们选择了一组具有代表性的接受 MPI 检查的患者，这些患者与参考性冠状动脉造影检查结果一致。有一半的患者存在阻塞性 CAD，定义为左主干动脉狭窄≥50%或其他冠状动脉节段狭窄≥70%。我们使用了一种可解释的 DL 模型（CAD-DL），该模型是在来自不同地点的独立人群中开发的。三位医生首先使用临床病史、应激和定量灌注来解释研究，然后使用所有数据加上 DL 结果。使用受试者工作特征曲线下面积（AUC）评估诊断准确性。
共纳入 240 名中位年龄为 65 岁（四分位距 58-73）的患者。有 CAD-DL 的医生解释的诊断准确性（AUC 0.779）明显高于没有 CAD-DL 的医生解释的诊断准确性（AUC 0.747，<0.001）和应激总灌注缺损（AUC 0.718，<0.001）。当自主运行时，CAD-DL 的敏感性高于没有 DL 结果的读者（<0.001），但与解释带有 DL 结果的读者相比没有差异（=0.122）。所有读者的准确性均有数值上的提高，AUC 提高 0.02-0.05，使用 DL 进行解释导致整体净重新分类改善 17.2%（95%CI 9.2%-24.4%，<0.001）。
可解释的 DL 预测可显著改善医生的解释；然而，这种改善因读者而异，反映出对这项新技术的接受程度。该技术可作为医生诊断的辅助工具，提高 MPI 的诊断准确性。
Zotero 插件