Keywords:
Arteriosclerosis, Segmentation, Computer Applications-Detection, diagnosis, Computer Applications-3D, CT-Quantitative, CT-Angiography, CT, Cardiovascular system, Abdomen
Authors:
P. Moeskops, B. de Vos, W. B. Veldhuis, P. A. de Jong, I. Isgum, T. Leiner; Utrecht/NL
Methods and Materials
The feasibility of fully automated body composition measurement was studied using a dataset of 20 CT scans of the abdomen (in-plane resolution 0.63–0.75 mm, slice thickness 5.0 mm, slice increment 5.0 mm). Trained observers defined the reference standard by manual annotation of subcutaneous fat, visceral fat and psoas muscle in all slices that contain the psoas muscle. Images from 10 patients were used to train a dilated convolutional neural network with a receptive field of 131 × 131 voxels to distinguish between the three tissue classes. Voxels were assigned to the class with the highest probability. Data from the remaining 10 patients were used to evaluate the performance of the method. Segmentation performance was evaluated with Dice coefficients between the manual and automatic segmentations. Additionally, linear correlation coefficients (Pearson’s r) were computed between the manual and automatic segmentation volumes.