Feasibility and Reproducibility of Semi-Automated Magnetic Resonance Angiography Measurements of the Thoracic Aorta using Commercial Software

Figures

• 

  Fig. 1. Curved MPR of T2 prep balanced steady state free precession MRA acquired with iNAV and VD-CASPR: A) Failure of segmentation due to blood pool inhomogeneity (asterisks). B) Failure of segmentation due to fluid in the anterior and posterior superior aortic recess (arrows). C) Inversion recovery gradient echo MRA demonstrates improved blood pool uniformity and signal suppression in the fluid-filled pericardial recess (arrows). iNAV image-based navigator, VD-CASPR variable density Cartesian sampling with spiral-like order, MRA magnetic resonance angiography, MPR multi-planar reformation, iNAV image navigator

• 

  Fig. 2. Patient recruitment flow chart for the study with images from the five excluded patients: 1) movement during iNAV CE-MRA; 2,3) bioprosthetic artifact (*) at the sinus of Valsalva; 4) hybrid repair with TEVAR and residual type B aortic dissection (D); and 5) bioprosthetic artifact (*) and residual type B aortic dissection (D). 1One subject was imaged at both sites- MRA data was used only from the first encounter at site 1. 2Subject was imaged at both sites, and therefore excluded at both locations. iNAV image-based navigator, CE-MRA contrast-enhanced magnetic resonance angiography, TEVAR thoracic endovascular aortic repair, MRA magnetic resonance angiography

• 

  Fig. 3. Software layout for semi-automated segmentation of the thoracic aorta by center line method in a subject from site 1. The thoracic aorta boundaries are outlined by markers A (sinus of Valsalva) and B (thoracoabdominal junction). Top right panel: VRT with centerline pathway displayed as the green line. Bottom right panel: Axial overview of the 3D volume with the individual centerline points represented by blue dots. VRT volume rendered technique, 3D three-dimensional

• 

  Fig. 4. A) Semi-automated segmentation of the thoracic aorta by center line method in a site 2 subject with thoracic aortic aneurysm and permanent pacemaker. B) Image-based navigator MRA images are free from substantial local field inhomogeneity artifacts compared to C) balance steady-state free precession. MRA magnetic resonance angiography

• 

  Fig. 5. Aortic diameter measurements using the double oblique MPR method. Here, the crosshairs are adjusted on the sagittal plane (bottom right) and coronal plane (bottom left) to obtain ascending aortic diameter on the double oblique image (top left). 3D VRT is displayed in the top right panel. MPR multiplanar reformation, VRT volume rendered technique

• 

  Fig. 6. Semi-automated generation of aortic inner-to-inner luminal contours at the sinus of Valsalva (1), sinotubular junction (2), and ascending aorta (3) via centerline method shown with volume rendered extraction. Demarcation of the right, left, and non-coronary cusps insertions (R, L, N) has been added manually

• 

  Fig. 7. Semi-quantitative grading of contour failure relative to the vessel circumference: 1–25% failure (A); 26–50% failure (B); 51–75% failure (C); and (D) >75% failure. The dashed segments in red outline the appropriate inner-to-inner vessel wall contours

• 

  Fig. 8. Bland–Altman plots of the intraobserver analysis for CSAM1 and DO MPR performed by physician 1 and physician 2 in the aortic disease cohort from site 2 (N=31 subjects contributing 3 measurements each). CSAM1 centerline semi-automated measurements, DO double oblique, MPR multiplanar reformation

• 

  Fig. 9. Bland–Altman plots of the intraobserver analysis for centerline semi-automated measurements recorded by fellow 2 (CSAM2), and interobserver analysis between fellow 1 (CSAM1) and CSAM2 in the aortic disease cohort (N = 41 subjects contributing 3 measurements each)

• 

  Fig. 10. Bland–Altman plots of the interobserver analysis (all subjects, n=68) between centerline semi-automated measurements (CSAM1) and DO MPR performed by physician 1 and physician 2. DO double oblique, MPR multiplanar reformation

• 

  Fig. 11. Bland–Altman plots of the interobserver analysis (aortic disease cohort, n = 41) between centerline semi-automated measurements (CSAM1) and DO MPR performed by physician 1 and physician 2. DO double oblique, MPR multiplanar reformation

• 

  Fig. 12. Bland–Altman plots of the interobserver analysis (normal cohort, n = 27) between centerline semi-automated measurements (CSAM1) and DO MPR performed by physician 1 and physician 2. DO double oblique, MPR multiplanar reformation