Investigating the morphological dynamics of patient-specific myocardial bridge models throughout the cardiac cycle

Myocardial bridging (MB) is a coronary artery anomaly characterized by an intramyocardial course of a coronary segment, resulting in dynamic luminal compression during systole. Although traditionally considered benign, increasing evidence has associated MB with altered coronary hemodynamics and associated increased risk of atherosclerotic plaque formation, potentially culminating in myocardial ischemia or infarction. This study aimed to establish a personalized framework for the morphological characterization of MB segments by (i) assessing the impact of operator-dependent variability on the reconstruction of patient-specific MB geometries, and (ii) conducting a detailed morphological quantification of the reconstructed arterial models. To do that, the left anterior descending coronary artery of three ostensibly healthy individuals was imaged using dynamic computed tomography (CT) across multiple phases of the cardiac cycle, sampled at 10% intervals (P1: 40–80%; P2 and P3: 10–100%). Patients P1 and P2 each presented a single proximal MB segment, whereas patient P3 exhibited two intramyocardial segments, one located proximally and the other distally. For each subject, the CT datasets were segmented to reconstruct patient-specific MB geometries at every cardiac phase by an inexperienced initial user (Train) and an expert operator (Expert). A detailed centerline-based comparison between Train and Expert reconstructions was conducted for each patient at each phase of the cardiac cycle, focusing on centerlines’ distance and luminal cross-sectional areas. Morphological characterization was then performed on Expert models through a centerline-based analysis, from which key geometrical parameters known to influence coronary hemodynamics were computed, including luminal cross-sectional area, local curvature and torsion, and shape index. The results demonstrate a marked dependence of the reconstruction accuracy on the operator’s technical skill and experience. Resulting Train vs. Expert centerlines’ distance ranged between 0.1 and 0.5 mm, corresponding to 5% and 13% of the median vessel diameter, respectively. Overall, Train models overestimated the vessel area across all three models (median error: 16%–198%). However, the myocardial bridge remained visible in most of the reconstructed phases, indicating that the anatomical feature could still be reliably identified despite global overestimation. It is worth noting that the last model reconstructed appears more consistent with the Expert one, likely benefiting (median error: 8%–28%). As for the morphological analysis, the MB segments exhibited a remarkable and consistent reduction in luminal cross-sectional area, with a maximum decrease of 70% relative to the physiological reference value. Notably, the lumen reduction persisted across all analyzed cardiac phases, with the absence of diastolic recovery. In contrast, a negligible impact of the MB emerged on the other geometric quantities. This study demonstrated that patient-specific coronary reconstruction can robustly capture the anatomical and morphological features of MBs, despite operator-dependent variability. Luminal area emerged as the most sensitive morphological parameter with MB segment showing persistent luminal narrowing over the cardiac cycle. These results confirm the reliability of image-based reconstruction as a valuable tool for morphological, biomechanical, and clinical investigation of coronary anomalies.