A Machine Learning Algorithm to Estimate Sarcopenia on Abdominal CT

Rationale and Objectives

To assess whether a fully-automated deep learning system can accurately detect and analyze truncal musculature at multiple lumbar vertebral levels and muscle groupings on abdominal CT for potential use in the detection of central sarcopenia.

Materials and Methods

A computer system for automated segmentation of truncal musculature groups was designed and created. Abdominal CT scans of 102 sequential patients (mean age 68 years, range 59–81 years; 53 women, 49 men) conducted between January 2015 and February 2015 were assembled as a data set. Truncal musculature was manually segmented on axial CT images at multiple lumbar vertebral levels as reference standard data, divided into training and testing subsets, and analyzed by the system. Dice similarity coefficients were calculated to evaluate system performance. IRB approval was obtained, with waiver of informed consent in this retrospective study.

Results

System performance as gauged by the Dice coefficients, for detecting the total abdominal muscle cross-section at the level of the third and fourth lumbar vertebrae, were, respectively, 0.953 ± 0.015 and 0.953 ± 0.011 for the training set, and 0.938 ± 0.028 and 0.940 ± 0.026 for the testing set. Dice coefficients for detecting total psoas muscle cross-section at the level of the third and fourth lumbar vertebrae, were, respectively, 0.942 ± 0.040 and 0.951 ± 0.037 for the training set, and 0.939 ± 0.028 and 0.946 ± 0.032 for the testing set.

Conclusion

This system fully-automatically and accurately segments multiple muscle groups at all lumbar spine levels on abdominal CT for detection of sarcopenia.

Introduction

Sarcopenia, initially defined as the loss of muscle mass by Rosenberg, has evolved as a syndrome to include other muscle function measurements such as strength and performance (1, 2). In 2016, the CDC established an ICD-10 code for sarcopenia, making it a recognized medical condition (3). This crucial step is allowing clinicians to code for sarcopenia, a risk factor associated with physical decline, decreased quality of life, and increased mortality (4). Computed tomography (CT), the gold standard for muscle mass and body composition measurement, is the imaging tool of choice for opportunistic sarcopenia diagnosis. With over 70 million CT scans conducted every year in the US alone (5), scans are readily available for sarcopenia screening justifying the need for a fully automated deep learning system.

Central sarcopenia is a risk factor for mortality in multiple disease processes, including cancer (6, 7). Varying criteria for determination of central sarcopenia on CT have been developed and linked to patient outcomes in clinical series. These criteria have in common that they attempt to gauge the mass of truncal muscle groups as a presumed measure of patient physiologic reserve (8). The common metric association of these series is that worsening patient outcomes are correlated with decreased cross-sectional area of tested truncal muscle groups.

The most common muscle groups tested in these series include the total axial cross section of truncal abdominal musculature also known as skeletal muscle area (SMA) normalized for patient height (skeletal muscle index – SMI or MI), the total psoas muscle cross-section (total psoas area – TPA) normalized for patient height (total psoas index – TPI), and the paraspinous muscle group, as measured on axial CT images. The SMI has been used as an imaging biomarker for mortality in lung cancer (L1 and L3), cirrhosis (L3), and trauma patients (L3); blood transfusion and hospitalization in proximal femoral fracture patients (L4); and therapy monitoring in pancreatic cancer patients (L3) (9, 10, 11, 12, 13, 14, 15). The TPA muscle group has been previously proposed for risk assessment in elderly trauma (L3 level), liver transplant (L4), and in pancreatic (L3), esophageal (L3), and endometrial cancer (L3) groups (16, 17, 18, 19, 20). The paraspinous muscle group has been discussed as a central sarcopenia determinant metric for survival and surgical outcomes in patients with colorectal cancer (L4), liver transplant (T12), and general and vascular surgery patients (T12) (15,21,22).

The standard for segmentation used in the determination of sarcopenia on CT is manual tracing of muscle group margins on axial sections, replicated over many patients. Preliminary work has been performed for automated segmentation of the total abdominal muscle cross section on a manually extracted CT image at the L3 level (23). The previously discussed studies conducted on patients with varying medical conditions, showed that different muscle groups assessed at distinct lumbar spine levels are linked to different outcomes. As such, a system which would automatically detect vertebral levels and segment multiple combinations of muscle groups at those levels would allow the radiologist to provide directed information to multiple specialties according to their own metric standards, for use in surgical planning and pretreatment risk assessment.

The purpose of this paper is to assess whether a fully-automated deep learning system can be created to opportunistically detect and analyze truncal musculature at set vertebral levels and for varied muscle groupings, for potential use in the detection of central sarcopenia using Computed Tomography (CT) scans.