Artificial Intelligence in ANOCA

Overview

Angina pectoris is diagnosed in \>180.000 people in the Netherlands each year. Diagnosis in angina pectoris focuses on epicardial coronary stenosis, the identification of which may lead to guideline-directed medical therapy or revascularization. However, no such stenosis is identified in 40-70% of patients. This condition, angina with no obstructed coronary artery (ANOCA), is more prevalent in women and is related to poor quality of life, high medical expenses, and a higher incidence of adverse events. The origin of ANOCA can be evaluated during invasive coronary angiography by coronary function testing (CFT) to identify coronary vasomotor disorders. This relates to vasospasm of the coronary artery and microcirculation, or to impaired microvascular vasodilation. For the diagnosis of vasospasm, CFT needs to result in electrocardiographic signs of myocardial ischemia as part of the diagnostic criteria. This is a critical point in the diagnosis of vasospasm, as these signs can be subtle and can vary, and are therefore prone to misinterpretation. Apart from this caveat, the diagnosis approach therefore currently requires an invasive procedure for the diagnosis. This limits the broad application and hampers early identification and treatment of ANOCA. During CFT, a coronary guide wire is routinely advanced in the coronary artery which also allows obtaining an intracoronary ECG by attaching a sterile alligator clamp to a standard electrocardiogram lead. This allows continuous recording of intracoronary ECG throughout CFT on the same monitor as the routine ECG. This technique can increase sensitivity for myocardial ischemia during CFT. Further, Holter ECG monitoring allows the identification of ischemic changes in the ECG in the outpatient setting. Evidence is lacking on the patterns of myocardial ischemia that occur during spontaneous angina pectoris symptoms in ANOCA patients, and on the sensitivity of Holter ECG for this purpose. Finally, the interpretation of ischemic patterns on ECG tracings can be cumbersome, especially when changes are subtle or change from beat to beat. The use of deep learning techniques allows to automate the interpretation of ECG traces and may improve the standardized diagnosis in ANOCA.