Pixel-Level Deep Segmentation: Artificial Intelligence Quantifies Muscle on Computed Tomography for Body Morphometric Analysis

Abstract

Pretreatment risk stratification is key for personalized medicine. While many physicians rely on an “eyeball test” to assess whether patients will tolerate major surgery or chemotherapy, “eyeballing” is inherently subjective and difficult to quantify. The concept of morphometric age derived from cross-sectional imaging has been found to correlate well with outcomes such as length of stay, morbidity, and mortality. However, the determination of the morphometric age is time intensive and requires highly trained experts. In this study, we propose a fully automated deep learning system for the segmentation of skeletal muscle cross-sectional area (CSA) on an axial computed tomography image taken at the third lumbar vertebra. We utilized a fully automated deep segmentation model derived from an extended implementation of a fully convolutional network with weight initialization of an ImageNet pre-trained model, followed by post processing to eliminate intramuscular fat for a more accurate analysis. This experiment was conducted by varying window level (WL), window width (WW), and bit resolutions in order to better understand the effects of the parameters on the model performance. Our best model, fine-tuned on 250 training images and ground truth labels, achieves 0.93 ± 0.02 Dice similarity coefficient (DSC) and 3.68 ± 2.29% difference between predicted and ground truth muscle CSA on 150 held-out test cases. Ultimately, the fully automated segmentation system can be embedded into the clinical environment to accelerate the quantification of muscle and expanded to volume analysis of 3D datasets.

Fully-Automated Analysis of Body Composition from CT in Cancer Patients Using Convolutional Neural Networks

“…The Dice similarity coefficient (DSC) is used to measure the difference between the automatic segmentation and the manual ground truth. Our results improve upon those of Lee et al [7], who had an average DSC of 0.93 for muscle, suggesting that the additional representational power of the U-Net and the more informative three-class training labels were effective at improving network accuracy. Additionally, our results improve upon those of Popuri et al [11] who achieved Jaccard indices of 0.904 for muscle, and 0.912 for fat (visceral and subcutaneous as a single class) which correspond to DSC values of 0.950 and 0.954.…”

Thoracic Skeletal Muscle Is Associated With Adverse Outcomes After Lobectomy for Lung Cancer

“…Specifically, reference values of expected thoracic muscle in men and women of different ages need to be established. Fortunately, muscle measurements can be performed in less than 10 minutes per patient with free and universally available software, and may be entirely automated with machine learning in the future [31]. Future work could assess the correlation of thoracic muscle measured on CT with evaluation of muscle strength and frailty, and whether the associations identified in this study are reproducible at other centers.…”

Artificial Intelligence in Musculoskeletal Imaging: A Paradigm Shift

“…Examples of opportunistic screening include bone mineral densitometry, visceral fat analysis, and sarcopenia assessment on CT scans obtained for colorectal cancer screening or other indications (14,62,(66)(67)(68)(69)(70)(71) (Fig. Examples of opportunistic screening include bone mineral densitometry, visceral fat analysis, and sarcopenia assessment on CT scans obtained for colorectal cancer screening or other indications (14,62,(66)(67)(68)(69)(70)(71) (Fig.…”

“…Opportunistic screening means the detection of abnormalities unrelated to the primary indication for the scan. Examples of opportunistic screening include bone mineral densitometry, visceral fat analysis, and sarcopenia assessment on CT scans obtained for colorectal cancer screening or other indications (14,62,(66)(67)(68)(69)(70)(71) (Fig. 6).…”