Artificial intelligence-supervised vectorcardiography for the diagnosis of a young adult with abnormal origin of the right coronary artery from aorta

Abstract

Coronary anomalies occur in 0.2% to 1.2% of the population, with the anomalous aortic origin of the coronary arteries accounting for one third of these cases. Clinical presentations can vary from asymptomatic to experiencing cardiac symptoms and sudden death, making diagnosis challenging. In this report, we present a novel artificial intelligence-supervised vectorcardiographic analysis and the subsequent successful surgical treatment of a young patient.

Case report

A 28-year-old male patient (height: 188 cm, weight: 121 kg) was admitted to our clinic with palpitations and exertional dyspnoea, which had become more prominent in recent months. At the onset of his symptoms, he lost 20 kgs through a personalised diet programme and had flu-like symptoms for a week. Three years ago, he was suspected of having Wallenberg (Lateral Medullary) Syndrome due to thrombosis of the posterior communicating artery, but he had no neurological complaints at the time of admission.

Myocardial perfusion scintigraphy performed at another cardiology institute revealed areas of heterogeneous perfusion in the inferolateral and apical segments, suspected to be secondary to possible myocarditis. Transthoracic echocardiography showed a left ventricular ejection fraction of 50%, along with mild dilation of the left and right ventricular cavities. Conventional angiography did not reveal any coronary stenosis, so the patient was treated symptomatically with analgesics and muscle relaxants. However, his symptoms persisted, prompting a 3D coronary angiography and contrast-enhanced computed tomography, which demonstrated the abnormal origin and intraarterial course of the right coronary artery, originating at the commissure between the right and left aortic cusps. The proximal segment of the right coronary artery was significantly compressed (Figure 1). The patient underwent surgery to reanastomose the right coronary ostial button to the right sinus, restoring its original anatomical position.

Figure 1. Contrast enhanced computed tomography of the patient. Red dotted lines demonstrate the abnormal origin and interarterial course of the right coronary artery. Blue asterisk marks the normal anatomical position of the right coronary artery. Ao: aorta, PA: pulmonary artery, RV: right ventricle.

Cardisiography, a combination of five-lead vectorcardiography and an artificial intelligence algorithm, was developed as a non-invasive diagnostic tool for detecting myocardial ischaemia at rest. ¹ It focuses on the spatial heterogeneity of cardiac muscle excitation, which indirectly indicates hypoxia at the cellular level. In our patient’s case, cardisiography revealed significant changes in the vectorcardiography, particularly concerning the QRS-T angle and the scatter values of the QRS and T vectors (Figure 2).

Figure 2. Cardisiography report of the patient demonstrating vectorcardiography analysis and cardiac loops.

Discussion

Although artificial intelligence-enabled electrocardiogram algorithms have advanced worldwide over the last decade, most networks are trained with traditional 12-lead electrocardiogram recordings, primarily targeting the detection of atrial fibrillation. ² On the other hand, the clinical value of vectorcardiography was demonstrated decades ago, but its clinical applicability remained limited. ³ Unlike traditional 2D electrocardiograms, 3D-loop recordings in cardisiography exponentially increase the number of parameters, recording 290 vectoral signals per beat. These parameters are well-documented in the literature and classified as normal or abnormal based on reference intervals. Vectorcardiography visualises the movement of heart vectors throughout the cardiac cycle as loops, with the QRS and T loops reflecting depolarisation and repolarisation, respectively. ⁴ Accordingly, the spatial QRS-T angle can be measured, a prominent variable in stratifying cardiac risk. ⁵ Several other parameters derived from vectorcardiography analysis are reported in the literature to diagnose various cardiac conditions such as myocardial ischaemia, arrhythmias, heart failure, and hypertrophic cardiomyopathy. ⁶

Our patient presented with typical chest pain suspected to be secondary to myocarditis. Transthoracic echocardiography demonstrated decreased ventricular function, but no definite underlying cause could be identified. The 12-lead traditional electrocardiography did not reveal any abnormalities in the ST and T segment analysis, and conventional coronary angiography, due to its non-stenotic nature, ruled out ischaemic aetiology. However, real-time vectorcardiography demonstrated significantly disturbed patterns of vector loops due to hypoxia at the cellular level.

In conclusion, artificial intelligence-supervised vectorcardiography enables quick and comprehensive data collection, recording vectoral loop patterns in just four minutes. The artificial intelligence component, trained with parameters from both healthy and sick individuals, can predict numerous parameters, aiding in the early detection of cardiac issues such as ischaemia. This makes artificial intelligence-supervised vectorcardiography a highly promising tool for the future, with the potential to screen and detect cardiac diseases before serious events like myocardial infarction or sudden death occur.