Preliminary Feasibility of Aortic Wall Strain using 4D Whole Thoracic Automated Cardiac MC (AutoCMR)

Background: Clinical guidelines for invasive interventions for aortic pathologies are based on diameter acquired from CT, MRI, or TTE. However, aortic wall strain has been shown to be prognostic of risk in abdominal aortic aneurysm independent of maximum diameter. Current methods of two-dimensional phase-contrast gradient-echo sequences and CTA-gated strain have limited spatial resolution and temporal resolution respectively. As such, we propose leveraging an isotropic 4D whole thoracic automated cardiac MRI (AutoCMR) for high temporal and spatial resolution imaging of the aorta. Thus, we created an automated, temporally varying measurement of the diameters that can be used for global and regional analysis of the aorta during a full cardiac cycle.

Methods: We developed an automated pipeline extracting the diameters of the maximally inscribed sphere (MIS) along the centerline of the aorta using AutoCMR automatic aortic segmentations. Thirty frames across the cardiac cycle of 120 patients with a variety of cardiac pathologies were passed through a Python script containing scriptable 3D Slicer VMTK extension modules to automatically extract the centerline of the aorta and calculate the diameter of the maximally inscribed sphere (DMIS) at all points along the centerline (Figure 1D and Figure 1E). The average DMIS across every zone of the aorta is calculated, and the maximum and minimum values were utilized to calculate diameter strain (i.e. [DMIS, max - DMIS, min] / DMIS, min). Additionally, for regional analysis across the aorta, all 30 frames for select cases were annotated with the start of the aorta, end of the aorta, and boundaries between Zones 0 and 1, Zones 1 and 2, and Zones 2 and 3 (Figure 1B and 1C).

Results: In Figure 2, we demonstrate the distributions of diameter strain across various cardiac pathologies with ANOVA test results (p = 0.008341) indicating potential diagnostic discriminatory power of this metric. Additionally, regional diameter strain measurements for a case of Marfan’s syndrome with a stable aortic root aneurysm (Zone 0: 0.1058, Zone 1: 0.0944, Zone 2: 0.0677, Zone 3: 0.07, and Zone 4: 0.089) and chronic systolic heart failure (Zone 0: 0.1762, Zone 1: 0.1041, Zone 2: 0.1054, Zone 3: 0.0802, Zone 4: 0.1375) were visualized in Figure 3 under the assumption that normal strain is 10-15%.

Conclusion: Based on these initial feasibility experiments, we were able to assess spatiotemporal features of the aorta over the cardiac cycle. However, further testing is required evaluate validity of this metric with the goal of creating a non-invasive pipeline for estimating aortic strain at both global and regional levels.

A) Schematic of the zones of the aorta B) Manual annotations of zonal boundaries between Zone 0 and Zone 1 (orange); Zone 1 and Zone 2 (yellow); and Zone 2 and Zone 3 (green) that correspond with the termination of the brachiocephalic, left-common-carotid, and left-subclavian vessels branching off the aorta in a single frame C) Visualization of the landmarks of a single frame of the aorta in 3D Slicer. D) Visualization of the maximally inscribed sphere (MIS) centered along the centerline of the aorta progressing from the beginning to the end of a single automatically segmented frame of the aorta. E) Visualization of the spatiotemporal data capturing the motion of the different zones of the aorta across the 30 frames of the cardiac cycle.
Distributions of diameter strain across 120 total patients with a variety of different cardiac pathologies.
A) Example visualization of an aorta heatmap providing regional strain measurements for a patient with Marfan’s syndrome and a stable aortic root aneurysm. B) Example aorta heatmap of a patient with chronic systolic heart failure.