Automatic Calcium Scoring in Low-Dose Chest CT Using Deep Neural Networks With Dilated Convolutions

• Published in: IEEE transactions on medical imaging Vol. 37; no. 2; pp. 615 - 625
• Main Authors: Lessmann, Nikolas, van Ginneken, Bram, Zreik, Majd, de Jong, Pim A., de Vos, Bob D., Viergever, Max A., Isgum, Ivana
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.02.2018; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Aged; Algorithms; Aorta; Aortic valve; Aortic Valve - diagnostic imaging; Aortic Valve - pathology; Aortic Valve Stenosis - diagnostic imaging; Arteries; Arteriosclerosis; Artificial neural networks; Atherosclerosis; Calcification; Calcification (ectopic); Calcinosis - diagnostic imaging; Calcium; Calcium scoring; Cancer; Cancer screening; Cardiovascular diseases; Chest; Computed tomography; convolutional neural networks; Coronary artery; Coronary Artery Disease - diagnostic imaging; Coronary vessels; Coronary Vessels - diagnostic imaging; Deep learning; dilated convolutions; Health risks; Heart diseases; Humans; Image detection; Image reconstruction; low-dose chest CT; Lung cancer; lung cancer screening; Lung Neoplasms - diagnostic imaging; Lungs; Medical imaging; Medical screening; Middle Aged; Mitral valve; Neural networks; Neural Networks (Computer); Radiography, Thoracic - methods; Risk assessment; Smoking; Thorax; Tomography; Tomography, X-Ray Computed - methods; Valves; Veins & arteries
• ISSN: 0278-0062; 1558-254X; 1558-254X
• DOI: 10.1109/TMI.2017.2769839

Heavy smokers undergoing screening with low-dose chest CT are affected by cardiovascular disease as much as by lung cancer. Low-dose chest CT scans acquired in screening enable quantification of atherosclerotic calcifications and thus enable identification of subjects at increased cardiovascular risk. This paper presents a method for automatic detection of coronary artery, thoracic aorta, and cardiac valve calcifications in low-dose chest CT using two consecutive convolutional neural networks. The first network identifies and labels potential calcifications according to their anatomical location and the second network identifies true calcifications among the detected candidates. This method was trained and evaluated on a set of 1744 CT scans from the National Lung Screening Trial. To determine whether any reconstruction or only images reconstructed with soft tissue filters can be used for calcification detection, we evaluated the method on soft and medium/sharp filter reconstructions separately. On soft filter reconstructions, the method achieved F 1 scores of 0.89, 0.89, 0.67, and 0.55 for coronary artery, thoracic aorta, aortic valve, and mitral valve calcifications, respectively. On sharp filter reconstructions, the F 1 scores were 0.84, 0.81, 0.64, and 0.66, respectively. Linearly weighted kappa coefficients for risk category assignment based on per subject coronary artery calcium were 0.91 and 0.90 for soft and sharp filter reconstructions, respectively. These results demonstrate that the presented method enables reliable automatic cardiovascular risk assessment in all low-dose chest CT scans acquired for lung cancer screening.