Automated mitral valve vortex ring extraction from 4D-flow MRI

Abstract

Purpose: To present and validate a method for automated extraction and analysis of the temporal evolution of the mitral valve (MV) vortex ring from MR 4D-flow data.

Methods: The proposed algorithm uses the divergence-free part of the velocity vector field for Q criterion-based identification and tracking of MV vortex ring core and region within the left ventricle (LV). The 4D-flow data of 20 subjects (10 healthy controls, 10 patients with ischemic heart disease) were used to validate the algorithm against visual analysis as well as to assess the method's sensitivity to manual LV segmentation. Quantitative MV vortex ring parameters were analyzed with respect to both their differences between healthy subjects and patients and their correlation with transmitral peak velocities.

Results: The algorithm successfully extracted MV vortex rings throughout the entire cardiac cycle, which agreed substantially with visual analysis (Cohen's kappa = 0.77). Furthermore, vortex cores and regions were robustly detected even if a static end-diastolic LV segmentation mask was applied to all frames (Dice coefficients 0.82 ± 0.08 and 0.94 ± 0.02 for core and region, respectively). Early diastolic MV vortex ring vorticity, kinetic energy and circularity index differed significantly between healthy controls and patients. In contrast to vortex shape parameters, vorticity and kinetic energy correlated strongly with transmitral peak velocities.

Conclusion: An automated method for temporal MV vortex ring extraction demonstrating robustness with respect to LV segmentation strategies is introduced. Quantitative vortex parameter analysis indicates importance of the MV vortex ring for LV diastolic (dys)function.

Keywords: 4D-flow; Q criterion; blood flow; left ventricle; mitral valve vortex ring; vortex parameter.

Conflict of interest statement

Gert Reiter is employed by Siemens Healthcare Diagnostics GmbH.

© 2020 The Authors. Magnetic Resonance in Medicine published by Wiley Periodicals LLC on behalf of International Society for Magnetic Resonance in Medicine.

Figures

Figure 1

Visual analysis of the presence/absence of an MV vortex ring in one sample frame of a healthy subject. The 3D streamlines (upper panel) and multiplanar reconstructed 2D vector fields (lower panel) are color‐coded by velocity. An MV vortex ring was identified if swirling flow was present in at least four of six regions in the vicinity of the mitral valve (white arrows) in cardiac long‐axis views

Figure 2

Pipeline of the proposed automated MV vortex ring extraction algorithm

Figure 3

Maximum Q throughout the cardiac cycle for identification of vortex candidates. The locations of all Q maxima (A) allow calculation of a Q threshold in each frame that will reveal the strongest vortices in the LV. The maximum Q time‐curve (B) allows identification of diastolic to early systolic “vortex frames” (blue). Abbreviations: LA, left atrium; LVOT, left ventricular outflow tract

Figure 4

Possible MV vortex ring shapes defining the stop criteria of the predictor‐corrector algorithm. The predictor‐corrector method is applied to each vortex candidate independently, elongating the vortex core first in the forward and then in the backward direction. Stop criteria for forward elongation are arriving back at the start point, thereby forming a closed vortex ring, getting stuck or reaching a voxel where Q ≤ 0. In the latter two cases, the vortex core is elongated in the backward direction until a closed vortex ring is formed (by fusing with the forward core line or arriving at the start point) or the core line gets stuck or reaches a voxel with Q ≤ 0. If no closed core line is formed, the forward and backward core lines are merged and tested for having a U‐shape. If there are two vortex candidates and neither of them yields a torus‐shape or U‐shape, the option that together they form a bracket‐shape is assessed

Figure 5

Reference condition and different segmentation approaches for MV vortex ring extraction and analysis experiments. In the reference condition (A), manual segmentation was performed on short‐axis images throughout the entire cardiac cycle. The slice distance was increased (B) and the segmentation mask at end‐diastole was applied to all cardiac frames (C) to simulate segmentation speed‐up. The segmentation masks of all frames were eroded (D) and dilated (E) to simulate segmentation variability

Figure 6

Temporal evolution of early and late diastolic MV vortex rings of a healthy subject. In each frame, the MV vortex ring is represented with Q isosurfaces that are color‐coded by MV vortex ring mean vorticity. For better orientation, the vortex rings are overlaid on a phase‐contrast MRA volume (notice the descending aorta on the right). The viewer position is at the apex looking upward. The diagram at the lower right shows the maximum forward velocity measured at the level of the mitral annulus

Figure 7

Error plots of early and late diastolic peak values of the MV vortex ring parameters volume, maximum vorticity, mean vorticity, absolute kinetic energy, and relative kinetic energy for different segmentation masks. Average errors are marked by ×, and a statistically significant difference to the reference condition is indicated by *