Automatic segmentation of multiple cardiovascular structures from cardiac computed tomography angiography images using deep learning

• Published in: PloS one 2020-05, Vol.15 (5), p.e0232573-e0232573
• Main Authors: Baskaran, Lohendran, Al'Aref, Subhi J, Maliakal, Gabriel, Lee, Benjamin C, Xu, Zhuoran, Choi, Jeong W, Lee, Sang-Eun, Sung, Ji Min, Lin, Fay Y, Dunham, Simon, Mosadegh, Bobak, Kim, Yong-Jin, Gottlieb, Ilan, Lee, Byoung Kwon, Chun, Eun Ju, Cademartiri, Filippo, Maffei, Erica, Marques, Hugo, Shin, Sanghoon, Choi, Jung Hyun, Chinnaiyan, Kavitha, Hadamitzky, Martin, Conte, Edoardo, Andreini, Daniele, Pontone, Gianluca, Budoff, Matthew J, Leipsic, Jonathon A, Raff, Gilbert L, Virmani, Renu, Samady, Habib, Stone, Peter H, Berman, Daniel S, Narula, Jagat, Bax, Jeroen J, Chang, Hyuk-Jae, Min, James K, Shaw, Leslee J
• Format: Article
• Language: English
• Subjects: Angiography; Aorta; Automation; Biology and Life Sciences; Cardiology; Cardiovascular disease; CAT scans; Computed tomography; Computer aided medical diagnosis; Computer and Information Sciences; Coronary angiography; Coronary vessels; Deep learning; Diagnostic imaging; Ground truth; Heart; Hospitals; Image processing; Image segmentation; Internal medicine; Machine learning; Medical imaging; Medical imaging equipment; Medical schools; Medicine; Medicine and Health Sciences; Methods; Pixels; Pulmonary arteries; Pulmonary artery; Research and Analysis Methods; Tomography; Ventricle
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0232573

To develop, demonstrate and evaluate an automated deep learning method for multiple cardiovascular structure segmentation. Segmentation of cardiovascular images is resource-intensive. We design an automated deep learning method for the segmentation of multiple structures from Coronary Computed Tomography Angiography (CCTA) images. Images from a multicenter registry of patients that underwent clinically-indicated CCTA were used. The proximal ascending and descending aorta (PAA, DA), superior and inferior vena cavae (SVC, IVC), pulmonary artery (PA), coronary sinus (CS), right ventricular wall (RVW) and left atrial wall (LAW) were annotated as ground truth. The U-net-derived deep learning model was trained, validated and tested in a 70:20:10 split. The dataset comprised 206 patients, with 5.130 billion pixels. Mean age was 59.9 ± 9.4 yrs., and was 42.7% female. An overall median Dice score of 0.820 (0.782, 0.843) was achieved. Median Dice scores for PAA, DA, SVC, IVC, PA, CS, RVW and LAW were 0.969 (0.979, 0.988), 0.953 (0.955, 0.983), 0.937 (0.934, 0.965), 0.903 (0.897, 0.948), 0.775 (0.724, 0.925), 0.720 (0.642, 0.809), 0.685 (0.631, 0.761) and 0.625 (0.596, 0.749) respectively. Apart from the CS, there were no significant differences in performance between sexes or age groups. An automated deep learning model demonstrated segmentation of multiple cardiovascular structures from CCTA images with reasonable overall accuracy when evaluated on a pixel level.