Automated Muscle Segmentation from Clinical CT Using Bayesian U-Net for Personalized Musculoskeletal Modeling

Abstract

We propose a method for automatic segmentation of individual muscles from a clinical CT. The method uses Bayesian convolutional neural networks with the U-Net architecture, using Monte Carlo dropout that infers an uncertainty metric in addition to the segmentation label. We evaluated the performance of the proposed method using two data sets: 20 fully annotated CTs of the hip and thigh regions and 18 partially annotated CTs that are publicly available from The Cancer Imaging Archive (TCIA) database. The experiments showed a Dice coefficient (DC) of 0.891±0.016 (mean±std) and an average symmetric surface distance (ASD) of 0.994±0.230 mm over 19 muscles in the set of 20 CTs. These results were statistically significant improvements compared to the state-of-the-art hierarchical multi-atlas method which resulted in 0.845±0.031 DC and 1.556±0.444 mm ASD. We evaluated validity of the uncertainty metric in the multi-class organ segmentation problem and demonstrated a correlation between the pixels with high uncertainty and the segmentation failure. One application of the uncertainty metric in activelearning is demonstrated, and the proposed query pixel selection method considerably reduced the manual annotation cost for expanding the training data set. The proposed method allows an accurate patient-specific analysis of individual muscle shapes in a clinical routine. This would open up various applications including personalization of biomechanical simulation and quantitative evaluation of muscle atrophy.

Image-based biomechanical models of the musculoskeletal system

“…U-Net is a convolutional network architecture widely used for fast and precise segmentation of images. In a work by Hiasa et al [28], a Bayesian convolutional network with the U-Net architecture was implemented to automatically extract muscle segmentation from clinical CT scans. The authors evaluated the performances on two datasets using the Dice coefficient (DC) [29] and the average symmetric surface distance (ASD).…”

Recent Collaborative R&D in Medical Imaging in Asia

“…Musculoskeletal applications are becoming important in the super-aging society in Japan. CNN-based segmentation of vertebrae of X-ray video during swallowing is addressed by University of Tsukuba [32], while segmentation of individual muscles, bones and implants as well as metal artifact reduction from CT are addressed by Nara Institute of Science and Technology (NAIST) and Osaka University [33,34] whose particular advantage is prediction of segmentation accuracy using uncertainty estimated from Bayesian U-net [34]. Statistical shape models are used for mandibular segmentation at NAIST in collaboration with University of Tehran [35].…”

Coronary artery segmentation in angiographic videos utilizing spatial-temporal information

“…Many blood vessel extraction methods based on image segmentation have emerged driven by this motivation. Recently, with the development of deep learning, various deep neural network architectures have been proposed and applied in the medical image segmentation field [ 1 – 4 ]. Early deep learning–based approaches used the image patches and a sliding window block to traverse the image [ 5 ].…”