Segment aorta and localize landmarks simultaneously on noncontrast CT using a multitask learning framework for patients without severe vascular disease

• Published in: Computers in biology and medicine 2023-06, Vol.160, p.107002-107002, Article 107002
• Main Authors: Yang, Jinrong, Li, Xiang, Cheng, Jie-Zhi, Xue, Zhong, Shi, Feng, Ji, Yuqing, Wang, Xuechun, Yang, Fan
• Format: Article
• Language: English
• Subjects: Algorithms; Aorta; Aorta - diagnostic imaging; Aorta segmentation; Aortic Diseases; Body measurements; Cancer screening; Chest; Coders; Computed tomography; Coronary vessels; Decoders; Deep learning; Early Detection of Cancer; Health risks; Humans; Hypertension; Image contrast; Image Processing, Computer-Assisted - methods; Image segmentation; Interfaces; Landmark localization; Localization; Lung cancer; Lung Neoplasms; Medical imaging; Medical screening; Metric space; Morphology; Multitask learning; Noncontrast CT; Simultaneous discrimination learning; Sinuses; Thorax; Tomography, X-Ray Computed - methods; Vascular diseases; Veins & arteries
• ISSN: 0010-4825; 1879-0534
• DOI: 10.1016/j.compbiomed.2023.107002

Non-contrast chest CT is widely used for lung cancer screening, and its images carry potential information of the thoracic aorta. The morphological assessment of the thoracic aorta may have potential value in the presymptomatic detection of thoracic aortic-related diseases and the risk prediction of future adverse events. However, due to low vasculature contrast in such images, visual assessment of aortic morphology is challenging and highly depends on physicians’ experience. The main objective of this study is to propose a novel multi-task framework based on deep learning for simultaneous aortic segmentation and localization of key landmarks on unenhanced chest CT. The secondary objective is to use the algorithm to measure quantitative features of thoracic aorta morphology. The proposed network is composed of two subnets to carry out segmentation and landmark detection, respectively. The segmentation subnet aims to demarcate the aortic sinuses of the Valsalva, aortic trunk and aortic branches, whereas the detection subnet is devised to locate five landmarks on the aorta to facilitate morphology measures. The networks share a common encoder and run decoders in parallel, taking full advantage of the synergy of the segmentation and landmark detection tasks. Furthermore, the volume of interest (VOI) module and the squeeze-and-excitation (SE) block with attention mechanisms are incorporated to further boost the capability of feature learning. Benefiting from the multitask framework, we achieved a mean Dice score of 0.95, average symmetric surface distance of 0.53 mm, Hausdorff distance of 2.13 mm for aortic segmentation, and mean square error (MSE) of 3.23 mm for landmark localization in 40 testing cases. We proposed a multitask learning framework which can perform segmentation of the thoracic aorta and localization of landmarks simultaneously and achieved good results. It can support quantitative measurement of aortic morphology for further analysis of aortic diseases, such as hypertension. In this paper, an automatic multitask learning framework (M-SL) is proposed for simultaneous aorta segmentation and landmark localization in NCCT (noncontrast CT). Our main contributions are summarized as follows:•A multitask learning framework is proposed for simultaneous aorta segmentation and landmark localization.•This parallel and end-to-end learning strategy can take advantage of the synergy of segmentation and landmark detection.•The volume of interest (VOI) module