Comparative analysis of body composition using torso CT from PET/CT with bioelectrical impedance and muscle strength in healthy adults

Dong Yun Lee, Jungsu S. Oh, Jeong Won Kim, Jae Seung Kim, Minyoung Oh, Yong Il Kim, Duk Han Ko, Sung‐Jin Bae, Hong‐Kyu Kim, Jin‐Sook RyuScientific 2024 · DOI

Abstract

The role of torso computed tomography (CT) in evaluating body composition has been unexplored. This study assessed the potential of low-dose torso CT from positron emission tomography (PET)/CT for analyzing body composition and its relation to muscle strength. We retrospectively recruited 384 healthy Korean adults (231 men, 153 women) who underwent torso <sup>18</sup>F-FDG PET/CT, bioelectrical impedance analysis (BIA), and muscle strength tests (handgrip strength [HGS] and knee extension strength [KES]). CT images were segmented into three compartments: torso volumetric, abdominal volumetric, and abdominal areal. Muscle amounts from each compartment were indexed to height (m<sup>2</sup>). BIA and HGS served as reference standards, with correlation coefficients (r) calculated. Torso muscle volumetric index (TorsoMVI) had the strongest correlations with BIA-derived values (r = 0.80 for men; r = 0.73 for women), surpassing those from the abdominal compartments. TorsoMVI was also correlated significantly with HGS (r = 0.39, p < 0.01) and differentiated between normal and possible sarcopenia in men (n = 225, 5960 ± 785 cm<sup>3</sup>/m<sup>2</sup> vs. n = 6, 5210 ± 487 cm<sup>3</sup>/m<sup>2</sup>, p = 0.02). In women, KES correlated more strongly with muscle parameters than HGS. Despite gender-specific variations, torso CT-derived parameters show promise for evaluating body composition and sarcopenia.

Introduction

Body composition, specifically the ratio of muscle to fat, is a critical biomarker for aging, metabolic health, and surgical outcomes. While Computed Tomography (CT) is the "gold standard" for measuring muscle mass, routine scans are rarely performed solely for this purpose due to radiation concerns. Consequently, researchers often employ "opportunistic" screening, analyzing CT scans taken for other reasons, such as PET/CT for cancer screening. However, most current methods rely on a single 2D slice at the third lumbar vertebra (L3). This study investigates whether a 3D volumetric analysis of the entire torso provides a more accurate representation of body composition and muscle strength than the standard 2D abdominal slice.

Methodology

The researchers conducted a retrospective study of 384 healthy Korean adults (231 men, 153 women). Participants underwent three types of evaluation:

  1. Torso PET/CT: Low-dose CT scans were segmented using AI-driven 3D U-Nets to automatically identify tissue types (muscle, visceral fat, subcutaneous fat, etc.).
  2. Bioelectrical Impedance Analysis (BIA): Used as a reference standard for skeletal muscle index (SMI) and fat mass.
  3. Muscle Strength Tests: Handgrip strength (HGS) for upper limb and knee extension strength (KES) for lower limb.

The study compared three measurement compartments: Torso Volumetric (base of skull to upper thigh), Abdominal Volumetric (12th rib to iliac crest), and Abdominal Areal (traditional single-slice at L3).

Results and Discussion

Correlation with Clinical Standards

The Torso Muscle Volumetric Index (TorsoMVI) showed significantly higher correlations with BIA-derived muscle mass ($r=0.80$ for men) than abdominal measurements. This suggests that whole-torso data is a more reliable proxy for total body composition than localized abdominal slices.

Correlation between abdominal muscle volumetric index and skeletal muscle index for men and women.

As shown in the figure above, the abdominal volumetric index correlates moderately with the skeletal muscle index ($r=0.66$ for men, $r=0.64$ for women). However, the study found that expanding the analysis to the full torso significantly improved these correlation coefficients, reaching $0.80$ and $0.73$ respectively.

Sarcopenia Screening and Gender Differences

A key finding was the divergence in strength correlations between genders. In men, HGS correlated most strongly with torso muscle parameters, while in women, KES was the more significant indicator. Furthermore, TorsoMVI was able to statistically differentiate between "normal" individuals and those with "possible sarcopenia" in the male subgroup, a distinction that traditional abdominal measurements failed to make.

Distribution of muscle strength and mass categories among the study participants (Men n=231, Women n=153).

The distribution of the study population (visualized above) highlights that while most participants were healthy, the volumetric torso parameters provided the sensitivity needed to identify the small subset of individuals with abnormal muscle strength or mass, particularly in men.

Conclusion

Volumetric analysis of the entire torso using low-dose CT scans provides a more robust assessment of body composition and muscle strength than traditional single-slice abdominal measurements. By leveraging AI-driven segmentation, the Torso Muscle Volumetric Index (TorsoMVI) emerges as a superior indicator for identifying early signs of sarcopenia. As total-body PET/CT scanners become more common, these volumetric AI tools could be integrated into routine clinical workflows to provide a holistic health check alongside primary diagnoses.

Summary updated

Abstract

The role of torso computed tomography (CT) in evaluating body composition has been unexplored. This study assessed the potential of low-dose torso CT from positron emission tomography (PET)/CT for analyzing body composition and its relation to muscle strength. We retrospectively recruited 384 healthy Korean adults (231 men, 153 women) who underwent torso <sup>18</sup>F-FDG PET/CT, bioelectrical impedance analysis (BIA), and muscle strength tests (handgrip strength [HGS] and knee extension strength [KES]). CT images were segmented into three compartments: torso volumetric, abdominal volumetric, and abdominal areal. Muscle amounts from each compartment were indexed to height (m<sup>2</sup>). BIA and HGS served as reference standards, with correlation coefficients (r) calculated. Torso muscle volumetric index (TorsoMVI) had the strongest correlations with BIA-derived values (r = 0.80 for men; r = 0.73 for women), surpassing those from the abdominal compartments. TorsoMVI was also correlated significantly with HGS (r = 0.39, p < 0.01) and differentiated between normal and possible sarcopenia in men (n = 225, 5960 ± 785 cm<sup>3</sup>/m<sup>2</sup> vs. n = 6, 5210 ± 487 cm<sup>3</sup>/m<sup>2</sup>, p = 0.02). In women, KES correlated more strongly with muscle parameters than HGS. Despite gender-specific variations, torso CT-derived parameters show promise for evaluating body composition and sarcopenia.
BibTeX
@inproceedings{lee2024comparative,
  author = {Dong Yun Lee and Jungsu S. Oh and Jeong Won Kim and Jae Seung Kim and Minyoung Oh and Yong Il Kim and Duk Han Ko and Sung‐Jin Bae and Hong‐Kyu Kim and Jin‐Sook Ryu},
  title = {{Comparative analysis of body composition using torso CT from PET/CT with bioelectrical impedance and muscle strength in healthy adults}},
  booktitle = {Scientific},
  year = {2024},
  doi = {10.1038/s41598-024-71878-2},
}
Figure