Fully automatic volume segmentation of infrarenal abdominal aortic aneurysm computed tomography images with deep learning approaches versus physician controlled manual segmentation

Abstract

Objective: Imaging software has become critical tools in the diagnosis and decision making for the treatment of abdominal aortic aneurysms (AAA). However, the interobserver reproducibility of the maximum cross-section diameter is poor. This study aimed to present and assess the quality of a new fully automated software (PRAEVAorta) that enables fast and robust detection of the aortic lumen and the infrarenal AAA characteristics including the presence of thrombus.

Methods: To evaluate the segmentation obtained with this new software, we performed a quantitative comparison with the results obtained from a semiautomatic segmentation manually corrected by a senior and a junior surgeon on a dataset of 100 preoperative computed tomography angiographies from patients with infrarenal AAAs (13,465 slices). The Dice similarity coefficient (DSC), Jaccard index, sensitivity, specificity, volumetric similarity (VS), Hausdorff distance, maximum aortic transverse diameter, and the duration of segmentation were calculated between the two methods and, for the semiautomatic software, also between the two observers.

Results: The analyses demonstrated an excellent correlation of the volumes, surfaces, and diameters measured with the fully automatic and manually corrected segmentation methods, with a Pearson's coefficient correlation of greater than 0.90 (P < .0001). Overall, a comparison between the fully automatic and manually corrected segmentation method by the senior surgeon revealed a mean Dice similarity coefficient of 0.95 ± 0.01, a Jaccard index of 0.91 ± 0.02, sensitivity of 0.94 ± 0.02, specificity of 0.97 ± 0.01, VS of 0.98 ± 0.01, and mean Hausdorff distance per slice of 4.61 ± 7.26 mm. The mean VS reached 0.95 ± 0.04 for the lumen and 0.91 ± 0.07 for the thrombus. For the fully automatic method, the segmentation time varied from 27 seconds to 4 minutes per patient vs 5 minutes to 80 minutes for the manually corrected methods (P < .0001).

Conclusions: By enabling a fast and fully automated detailed analysis of the anatomic characteristics of infrarenal AAAs, this software could have strong applications in daily clinical practice and clinical research.

Keywords: Abdominal aortic aneurysm; Artificial intelligence; Automatic segmentation; Deep learning; Endovascular aortic repair; Thrombus; Volume.