Kruger et al. investigates the use of neural network-based approaches for sectioning the aortic root from pre-interventional CT scans in planning for transcatheter aortic valve implantation. A cascade of convolutional neural networks were developed to extract geometric measures of a patient’s aortic root by reducing the image to the aortic root, valve, and left ventricular outflow tract and then segmenting the aortic valve, aorta, and aortic annulus. Researchers found that by using 90 extracted CT scans for training and the test results from 36 patients, the segmentation of the aorta and valve achieved an F1 score of 0.94, which is consistent with published results of other approaches. The area-derived annulus diameter was measured with a mean error below 2 mm between automized and annotated measurements. The authors conclude that neural network-based approaches are accurate and efficient with little training data but could highly benefit with more diverse datasets.