AI-based automated body composition analysis from whole-body MRI and (PET-) CT: an intermodal comparison from same-day examinations.
7 affiliations
- 1Department of Radiology, German Cancer Research Centre (DKFZ), Im Neuenheimer Feld 280, Heidelberg, 69120 , Germany. [email protected].
- 2Department of Radiology, German Cancer Research Centre (DKFZ), Im Neuenheimer Feld 280, Heidelberg, 69120 , Germany.
- 3Faculty of Medicine and University Hospital Cologne, Institute for Diagnostic and Interventional Radiology, Cologne, Germany.
- 4German-Speaking Myeloma Multicenter Group (GMMG), Heidelberg University Hospital, Heidelberg, 69120, Germany.
- 5Department of Internal Medicine V, Hematology, Oncology and Rheumatology, Medical Faculty Heidelberg, Heidelberg Myeloma Center, Heidelberg University Hospital, Heidelberg, 69120, Germany.
- 6Clinical Cooperation Unit Nuclear Medicine, German Cancer Research Centre (DKFZ), Heidelberg, Germany.
- 7Department of Medical Oncology, National Center for Tumor Diseases (NCT), Heidelberg University Hospital, Heidelberg, 69120, Germany.
Abstract
Background
To determine intermodal agreement of AI-derived volumetric body composition metrics between [18 F]FDG-PET/CT-derived low-dose CT and whole-body MRI (wb-MRI).Methods
We retrospectively analyzed same-day whole-body low-dose CT, obtained from [¹⁸F]FDG-PET/CT, and wb-MRI from consecutive multiple myeloma patients from the prospective GMMG-HD7- and GMMG-HD8-/DSMM-XIX-trials. An automated deep learning pipeline based on the TotalSegmentator model, and its MRI variant were used to segment skeletal muscle (SM), subcutaneous- (SAT) and abdominal visceral adipose tissue (VAT) from CT and MRI, respectively. Volumetric measures and single-slice estimates were derived. Agreement between CT- and MRI-based metrics was assessed using Pearson r, intraclass correlation coefficients (ICC), and Bland-Altman analyses.Results
Forty-two multiple myeloma patients with same-day PET/CT and wb-MRI were included. For AI-based volumetric body composition metrics, CT and MRI demonstrated high agreement (ICC > 0.95) across all SM, SAT and VAT. However, Bland-Altmann plots revealed systematic modality-related differences between MRI and CT: both SM (bias: -2582 ml, p < 0.0001) and VAT (bias: -211 ml, p < 0.0001) were significantly lower when extracted from MRI compared to CT, while for SAT, the difference was not statistically significant (bias: -263 ml, p = 0.1629). For AI-based single slice analysis, agreement remained high for SAT and VAT (ICC > 0.97), while being lower for SM (ICC 0.87). Bland-Altman plots showed that SM was systematically and significantly lower when obtained from MRI than from CT (bias: -18.4cm², p < 0.0001).Conclusions
Automated body composition measures from (PET-)CT and wb-MRI are highly concordant but show consistent intermodal biases that should be considered when pooling multimodal data or applying modality-agnostic thresholds.Clinical trial number
Not applicable.
Related
- Phase-adaptive automated whole-liver quantification of signal intensity-derived liver indices from gadoxetic acid-enhanced MRI
- Comparison of two automated PGMI-based software tools for breast positioning quality assessment in DBT screening
- Method Comparison of Three Automated D-Dimer Assays.
- Effects of Sleeve Gastrectomy and Subsequent Whole-Body Vibration Training on Resting Metabolic Rate and Body Composition: A Pilot Randomized Controlled Trial
- Unsupervised anomaly detection for longitudinal comparison in whole-body PET/CT images
BibTeX
@article{holzschuh2026aibased,
title = {{AI-based automated body composition analysis from whole-body MRI and (PET-) CT: an intermodal comparison from same-day examinations.}},
author = {Julius C Holzschuh and Lukas T Rotkopf and Christian Neelsen and Fabian Bauer and Kevin S Zhang and Ekaterina Menis and Marina Hajiyianni and Elias Karl Mai and Niels Weinhold and Lukas John and Christos Sachpekidis and Antonia Dimitrakopoulou-Strauss and Marc-Steffen Raab and Hartmut Goldschmidt and Johann Jende and Stefan Delorme and Heinz-Peter Schlemmer and Markus Wennmann},
journal = {Cancer imaging : the official publication of the International Cancer Imaging Society},
year = {2026},
month = oct,
doi = {10.1186/s40644-026-01134-5},
url = {https://europepmc.org/article/MED/42928255},
note = {CC BY},
}
RIS
TY - JOUR
TI - AI-based automated body composition analysis from whole-body MRI and (PET-) CT: an intermodal comparison from same-day examinations.
AU - Julius C Holzschuh
AU - Lukas T Rotkopf
AU - Christian Neelsen
AU - Fabian Bauer
AU - Kevin S Zhang
AU - Ekaterina Menis
AU - Marina Hajiyianni
AU - Elias Karl Mai
AU - Niels Weinhold
AU - Lukas John
AU - Christos Sachpekidis
AU - Antonia Dimitrakopoulou-Strauss
AU - Marc-Steffen Raab
AU - Hartmut Goldschmidt
AU - Johann Jende
AU - Stefan Delorme
AU - Heinz-Peter Schlemmer
AU - Markus Wennmann
PY - 2026
DA - 2026/10/24
JO - Cancer imaging : the official publication of the International Cancer Imaging Society
DO - 10.1186/s40644-026-01134-5
UR - https://europepmc.org/article/MED/42928255
AB - Background To determine intermodal agreement of AI-derived volumetric body composition metrics between [18 F]FDG-PET/CT-derived low-dose CT and whole-body MRI (wb-MRI).Methods We retrospectively analyzed same-day whole-body low-dose CT, obtained from [¹⁸F]FDG-PET/CT, and wb-MRI from consecutive multiple myeloma patients from the prospective GMMG-HD7- and GMMG-HD8-/DSMM-XIX-trials. An automated deep learning pipeline based on the TotalSegmentator model, and its MRI variant were used to segment skeletal muscle (SM), subcutaneous- (SAT) and abdominal visceral adipose tissue (VAT) from CT and MRI, respectively. Volumetric measures and single-slice estimates were derived. Agreement between CT- and MRI-based metrics was assessed using Pearson r, intraclass correlation coefficients (ICC), and Bland-Altman analyses.Results Forty-two multiple myeloma patients with same-day PET/CT and wb-MRI were included. For AI-based volumetric body composition metrics, CT and MRI demonstrated high agreement (ICC > 0.95) across all SM, SAT and VAT. However, Bland-Altmann plots revealed systematic modality-related differences between MRI and CT: both SM (bias: -2582 ml, p < 0.0001) and VAT (bias: -211 ml, p < 0.0001) were significantly lower when extracted from MRI compared to CT, while for SAT, the difference was not statistically significant (bias: -263 ml, p = 0.1629). For AI-based single slice analysis, agreement remained high for SAT and VAT (ICC > 0.97), while being lower for SM (ICC 0.87). Bland-Altman plots showed that SM was systematically and significantly lower when obtained from MRI than from CT (bias: -18.4cm², p < 0.0001).Conclusions Automated body composition measures from (PET-)CT and wb-MRI are highly concordant but show consistent intermodal biases that should be considered when pooling multimodal data or applying modality-agnostic thresholds.Clinical trial number Not applicable.
SN - 1740-5025
SN - 1470-7330
ER -