Reliability and Influence on Decision Making of fully-automated vs. semi-automated Software Packages for Procedural Planning in TAVI

Figures

• 

  Figure 1. Bland-Altman plots of aortic valve annular area and perimeter and coronary distances to the aortic valve annular plane. Bland-Altman plots of annular area in mm² (mean difference −13.306 mm², upper limit 62.631 mm², lower limit −89.243 mm²), annular perimeter in mm (mean difference −1.638 mm, upper limit 4.987 mm, lower limit −8.263 mm), left coronary artery (LCA; mean difference 1.438 mm, upper limit 6.148 mm, lower limit −3.272 mm) and right coronary artery (RCA; mean difference 0.694 mm, upper limit 5.284 mm, lower limit −3.896 mm) distances to the annular plane measured by a 3mensio reference user and the fully-automated HN software. The thick dashed line indicates the mean difference and the upper and lower limits of agreement. The overlaid red line indicates the non-linear fit shaded with the corresponding 95% confidence interval.

• 

  Figure 2. Distribution plots of aortic valve annular area and perimeter and coronary distances to the aortic valve annular plane. Distribution plots of measured CT-morphological structures either with the semi-automated 3mensio or with the fully-automated HeartNavigator3 software package. Each plot depicts a distinct measurements of a morphological structure - shown are boxplots overlaid with violin distribution plots.

• 

  Figure 3. Agreement of 3mensio measurements. To assess the agreement of the semi-automated 3mensio measurements aortic valve annular area [mm³] and perimeter [mm] as well as coronary distances to the aortic valve annular plane [mm] were measured by three independent observers (operator A–C) on the same patient, respectively. Each dot represents an individual measurement.

• 

  Figure 4. Retrospective TAVI prosthesis sizing simulation. The waffle plot is showing each sizing decision and the difference to the actual 3mensio based sizing is color-coded. In 73 of 100 cases no change (light-green) of the sizing strategy was observed. In 9 cases the HeartNavigator measurements were not in the valid sizing range of the prosthesis which was actually used during implantation (i.e. annulus area was too big for a sapien 329 mm prosthesis) - depicted as dark-green squares. In 13 cases HeartNavigator based sizing would have led to a bigger (light-red) and in 5 cases to a smaller prosthesis (dark-red).