AUTHOR=Yang Tingting , Zhu Guangyu , Cai Li , Yeo Joon Hock , Mao Yu , Yang Jian 

TITLE=A benchmark study of convolutional neural networks in fully automatic segmentation of aortic root

JOURNAL=Frontiers in Bioengineering and Biotechnology

VOLUME=Volume 11 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/bioengineering-and-biotechnology/articles/10.3389/fbioe.2023.1171868

DOI=10.3389/fbioe.2023.1171868

ISSN=2296-4185

ABSTRACT=Recent clinical studies have suggested that introducing 3D patient-specific aortic root models into the pre-operative assessment procedure of transcatheter aortic valve replacement (TAVR) would reduce the incident rate of peri-operative complications. Tradition manual segmentation is labor-intensive and low-efficient, which cannot meet the clinical demands of processing large data volumes. Recent developments in machine learning provided a viable way for accurate and efficient medical image segmentation for 3D patient-specific models automatically. This study quantitively evaluated the auto segmentation quality of four popular segmentation-dedicated convolutional neural network (CNN) architectures, including 3D UNet, VNet, 3D Res-UNet and SegResNet. All the CNNs were implemented in Pytorch platform, and chest CTA image sets of 98 anonymized patients were retrospectively selected from the database for training and testing of the CNNs. The results showed that the 3D Res-UNet has the best overall performance in aortic root segmentation. Despite all four CNNs having similar Dice similarity coefficient (DSC), Precision and Jaccard index on the segmentation of the aortic root, the segmentation efficiency of the 3D Res-UNet is significantly (P<0.001) higher than the rest CNNs. The average segmentation time on the testing set of the 3D Res-UNet is 0.10±0.04 s for each patient, which is 91.2%, 95.3% and 64.3% faster than 3D UNet, VNet and SegResNet, respectively. In addition, the Hausdorff distance (HD) of the segmentation results from 3D Res-UNet is 8.56±2.28, which is only 9.8% higher than that of the VNet, but 25.5% and 86.4% lower than that of the 3D UNet and SegResNet, respectively. The results from this study suggested that the 3D Res-UNet is a suitable candidate for accurate and fast automatic aortic root segmentation.