Author: Kelly Nicol
Background: Morphological criteria that result in inducible ischemia with anomalous aortic origin of coronary artery (AAOCA) have been postulated but not confirmed. Virtual AAOCA three-dimensional models (3DM) could elucidate the mechanism of flow alteration.There is significant heterogeneity in origin, ostia, angulation, intramurality and interarterial course of the AAOCA. Our objective is to test feasibility creating 3DM mesh of various AAOCA’s that can help understand the heterogeneity of this rare coronary anomaly and potentially help with clinical risk stratification. Methods: Data from CT angiograms (anisotropic voxels, 0.6-0.9 mm) were transferred to Seg3D2 to be resampled and to isolate the AAOCA. Median filtering was used to remove noise while preserving edge integrity, followed by a four component Otsu thresholding for segmentation and labeling of the AAOCA. Extraneous portions were manually deleted before application of a connected component filter. Final label maps (isotropic voxels, 0.2 mm) were transformed into mesh-based models using marching cubes. Meshlab was used for visualization of the coronary ostia and course. Clinical data was obtained from medical records. Preliminary Results: Eleven 3DM & mesh models (right AAOCA-9 and left AAOCA-5) were successfully created using our method, visualization of the ostia and proximal coronary course was possible in these models. Model creation took approximately 20 minutes each. Origin above the sinus of valsalva was noted in four, a narrow elliptical ostia was seen in nine, narrowing of the proximal lumen compared to distal was seen in four. Identification of intramural course was difficult using these 3DM. None of these patients had coronary related symptoms or stress test suggestive of inducible ischemia. None of the subjects underwent surgery for AAOCA. Conclusions: High resolution AAOCA 3DM can help identify heterogeneity of proximal coronary arteries. Significant heterogeneity in our sample is noted on all morphological features that are considered to be at high risk for inducible ischemia. Interestingly defining intramural course was challenging on this small sample. As this is a rare lesion, creating a multi-institutional deidentified DICOM image database that could be used to create 3DM can potentially aid in visualization, machine learning, and big data analysis. Given the shortcomings to define risk using traditional imaging techniques, AAOCA 3DM may help predict at-risk lesions for sudden cardiac death.
Co Author/Co-Investigator Names/Professional Title: 1. Santosh C. Uppu, MD 2. Shubhika Srivastava, MBBS 3. Anthony Costa, Phd