Astudillo et al. developed a deep learning model to automatically predict aortic annulus perimeter and area from CT images to assist device sizing in transcatheter aortic valve implantation (TAVI) planning. Using 355 training and 118 test scans, the model combined two convolutional neural networks with a postprocessing step to generate measurements and prosthesis size recommendations. Accuracy was compared to interoperator variability in manual measurements of the 118 scans, showing that the differences between automatic predictions and manual measures were similar to those between two independent human observers. The paired difference in area was 3.3 ± 16.8 mm2 for the model and 1.3 ± 21.1 mm2 for the interobserver comparison; perimeter differences were 0.6 ± 1.7 mm2 and 0.2 ± 2.5 mm2 respectively. Further, prosthesis size predictions aligned closely with operator selections, and the pipeline required under one second per case. This study demonstrates that automated annulus sizing with deep learning achieves accuracy comparable to expert observers while markedly reducing analysis time.