Imaging 4D morphology and dynamics of mitral annulus in humans using cardiac cine MR feature tracking

Figures

• 

  Figure 1. Radially rotational long-axis cine CMR. Localization of the 18 imaging slices in a short-axis view (red lines represent cutting planes) (A) and selected radial slices in 3D reconstructed view (B). Two mitral annulus points in each of the 18 acquired radial slices were tracked (red squares in (B)).

• 

  Figure 2. Mitral annulus dynamics obtained by feature tracking. In (A,B), the MA velocity and displacement were plotted as a function of annular region and cardiac cycle for all radially rotational slices and cardiac phases. In (C,D), the spatially averaged velocity and displacement curves were plotted along the cardiac cycle for 18, 9, 6, and 3 slices, respectively. For the 3-slice version, the clinical routine 2-, 3- and 4-chamber views were chosen. Data were obtained from a 39-year-old male normal control. Sm: peak systolic velocity; Am: peak late diastolic velocity; Em: peak early diastolic velocity; MAPSE: mitral annular plane systolic excursion.

• 

  Figure 3. Comparison of mitral annulus dynamics based on long-axis cine feature tracking. Correlation and difference between results from 18 radially rotational slices and 3 clinical routine long-axis views are shown in scatter plot (left) and Bland Altman plots of difference against average (right) for all dynamics parameters Sm, Em, Am, and MAPSE. Normal controls indicated by circles and patients by squares.

• 

  Figure 4. Mitral annulus morphology obtained by feature tracking and geometry reconstruction. 3D reconstructed annulus (A), and temporal changes of the quantitative parameters 3D area (B), perimeter (C), height (D), AP (E) and IC diameter (F) along the cardiac cycle for 18, 9, 6, and 3 slices, respectively. For 3-slice version, the clinical routine 2-, 3- and 4-chamber views were chosen. Data were obtained from a 49-yr-old female normal control.

• 

  Figure 5. Mitral annulus morphology obtained by feature tracking and geometry reconstruction with different numbers of radially rotational long-axis slices. Reconstructed geometries of the mitral annulus are shown for (A) 18, (B) 9, (C) 6, and (D) 3 imaging slices at both end-diastole (top) and end-systole (bottom). Data were obtained from a 49-yr-old female normal control.

• 

  Figure 6. Comparison of mitral annulus morphology based on long-axis cine feature tracking and geometry reconstruction. Correlation and difference between results from 18 and 6 radially rotational long-axis slices are shown in scatter plot (left) and Bland Altman plots of difference against average (right) for all morphological parameters 3D area, perimeter, diameter, and height, respectively. Normal controls indicated by circles and patients by squares.

• 

  Figure 7. Semi-automated feature tracking algorithm on cine CMR data. (A,B) User selected mitral annulus (MA) point in the end diastolic phase of one long-axis cine-CMR series as mask (red square); (C,D) Program auto-selected search region in the target frame of the same series (blue square); (E) Tracking produced correlation coefficient map with the location of highest correlation coefficient corresponding to the best match; (F) Updated mask with located MA point in the target frame; (G) Tracking process iteratively executed for all frames and cine-CMR series; (H) Reconstructed MA ring with spatial transformation and interpolation. Details see text.

• 

  Figure 8. Semi-automatic feature tracking of the mitral annulus. Six series out of 18 radially rotational slices are displayed here for both end diastole (top) and end systole (bottom). Slice 4, 10 and 16 corresponded to routine 2-, 4-, and 3-chamber views, respectively, in this case.