Deep learning for automated segmentation of pelvic muscles, fat, and bone from CT studies for body composition assessment

• Published in: Skeletal radiology Vol. 49; no. 3; pp. 387 - 395
• Main Authors: Hemke, Robert, Buckless, Colleen G., Tsao, Andrew, Wang, Benjamin, Torriani, Martin
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.03.2020; Springer; Springer Nature B.V
• Subjects: Acetabulum; Adipose tissue; Adipose Tissue - diagnostic imaging; Adipose tissues; Artificial neural networks; Automation; Body Composition; Bone composition; Computed tomography; Contrast Media; CT imaging; Data augmentation; Datasets; Datasets as Topic; Deep learning; Equalization; Female; Histograms; Humans; Image processing; Image segmentation; Imaging; Male; Medical imaging equipment; Medicine; Medicine & Public Health; Middle Aged; Muscle, Skeletal - diagnostic imaging; Muscles; Neural networks; Neural Networks, Computer; Nuclear Medicine; Orthopedics; Pathology; Pelvis; Pelvis - diagnostic imaging; Physiological aspects; Radiology; Retrospective Studies; Scientific Article; Tomography, X-Ray Computed; Training; Deep learning; Muscle; Segmentation; Computed tomography; Pelvis; Body composition
• ISSN: 0364-2348; 1432-2161; 1432-2161
• DOI: 10.1007/s00256-019-03289-8

Objective To develop a deep convolutional neural network (CNN) to automatically segment an axial CT image of the pelvis for body composition measures. We hypothesized that a deep CNN approach would achieve high accuracy when compared to manual segmentations as the reference standard. Materials and methods We manually segmented 200 axial CT images at the supra-acetabular level in 200 subjects, labeling background, subcutaneous adipose tissue (SAT), muscle, inter-muscular adipose tissue (IMAT), bone, and miscellaneous intra-pelvic content. The dataset was randomly divided into training (180/200) and test (20/200) datasets. Data augmentation was utilized to enlarge the training dataset and all images underwent preprocessing with histogram equalization. Our model was trained for 50 epochs using the U-Net architecture with batch size of 8, learning rate of 0.0001, Adadelta optimizer and a dropout of 0.20. The Dice (F1) score was used to assess similarity between the manual segmentations and the CNN predicted segmentations. Results The CNN model with data augmentation of N  = 3000 achieved accurate segmentation of body composition for all classes. The Dice scores were as follows: background (1.00), miscellaneous intra-pelvic content (0.98), SAT (0.97), muscle (0.95), IMAT (0.91), and bone (0.92). Mean time to automatically segment one CT image was 0.07 s (GPU) and 2.51 s (CPU). Conclusions Our CNN-based model enables accurate automated segmentation of multiple tissues on pelvic CT images, with promising implications for body composition studies.