RPS 1210 - Imaging of body composition

Author Block: V. K. Venugopal¹, V. Rengan², S. Ingole¹; ¹New Delhi/IN, ²Chennai/IN
Purpose: To assess the validity and reliability of an automated sarcopenia estimation approach using a deep-learning based ensemble psoas segmentation and Hounsfield Unit Average Calculation (HUAC) model in comparison to traditional manual measurements.
Methods or Background: This study retrospectively analyzed 149 CT scans, comparing sarcopenia assessments between manual HUAC measurements and those derived from an automated TransUNet-based system. The AI model combined convolutional neural networks with Transformer blocks to enhance feature extraction and contextual understanding of muscle tissue, crucial for precise sarcopenia evaluation. The HUAC was calculated by measuring the area and mean Hounsfield Units (HU) of the left and right psoas muscles at the L3 vertebra level. Statistical analysis included mean, standard deviation, correlation, paired t-tests, Bland-Altman plots, and advanced validation metrics such as Intersection over Union (IoU) and Dice coefficient to evaluate the model's segmentation accuracy.
Results or Findings: The AI model produced a mean HUAC of 19.66, slightly higher than the 18.03 from manual assessments, with corresponding standard deviations of 4.27 and 4.54, respectively. The correlation coefficient of 0.78 indicated strong agreement between the two methods. The model achieved an IoU of 90% and a Dice coefficient of 0.90, demonstrating high precision in muscle segmentation. The systematic bias observed (mean difference of -1.63 HUAC) highlights areas for further calibration of the AI model.
Conclusion: The integration of AI in sarcopenia assessment through HUAC calculations offers a promising alternative to manual measurements, providing speed, reproducibility, and precision. Despite some variance, the AI method aligns closely with traditional approaches, suggesting that with further refinement, it could become a standard tool in clinical settings.
Limitations: Small sample set
Funding for this study: Nil
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: IRB Waiver

Author Block: N. Kaştan, I. Erdem Toslak, S. Bakırcı, A. Yavuz; Antalya/TR
Purpose: Ankylosing spondylitis (AS) is a chronic inflammatory disease affecting the axial skeleton. Inflammatory cytokines like TNFalpha and interleukins increase in AS, along with adipose tissue and muscle catabolism. Sarcopenia in AS correlates with higher inflammation, greater disease activity, and reduced muscle performance. We hypothesize that the sarcopenia index and quantitative measures of muscle and fatty tissues may relate to the response to biological agent treatment
Methods or Background: This retrospective study involved 62 adults diagnosed with AS who underwent CT at the L2 vertebra level before any treatment and received biological agent therapy. CT measurements included visceral and subcutaneous abdominal adipose tissue cross-sectional area (VAT cm², SAT cm²), total abdominal muscle area (TAMA), psoas muscle volume (PsoA), sarcopenia index (SMI), visceral and subcutaneous abdominal adipose tissue attenuation (VAT HU, SAT HU), and psoas muscle attenuation (Pso HU). BASDAI score changes were assessed at the first post-treatment visit to evaluate disease activity. Comparisons and correlations were performed between SMI, adipose and muscle tissue measurements, and clinical parameters.
Results or Findings: TAMA, PsoA, and SMI values were significantly higher in patients with full recovery compared to those with partial recovery (p<0.05). However, VAT and SAT (cm²), VAT and SAT (HU), and Pso HU did not significantly affect recovery (p>0.05). This study demonstrates that SMI, TAMA, and psoas muscle volume can serve as prognostic markers for biological treatment response in AS patients.
Conclusion: Our study is an example of opportunistic-quantitative imaging methods and it has been shown that SMI value, TAMA and psoas muscle volume value can be used as prognostic markers in response to biological treatment in AS patients.
Limitations: Retrospective design and small sample size were the limitations of our study.
Funding for this study: No external funding.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Local IRB

Author Block: L. D. Grünewald, V. Koch, S. Mahmoudi, J-E. Scholtz, S. Martin, C. Booz, I. Yel, T. Vogl; Frankfurt/DE
Purpose: To evaluate the impact of intravenous contrast agents on osteoporosis assessment via routine CT in arterial and venous phases and identify mitigation strategies using dual-energy CT (DECT).
Methods or Background: 288 patients (154 men, 134 women; median age 62 years) who underwent abdominal DECT scans in non-contrast, late-arterial, and portal venous phases between January 2018 and December 2023 were retrospectively analyzed. Trabecular HU values were measured in all phases, including 90kV and 150kV DECT series, using automatic segmentation. T-scores were calculated to classify patients as osteoporotic, osteopenic, or normal. Changes in HU values, T-scores, and classifications due to contrast were compared to non-contrast images, with effects quantified using Cohen’s d.
Results or Findings: Median trabecular HU at L1 was 147 (IQR 116–185). Contrast in late arterial and portal venous phases increased HU values by +14.4 (+11.2%) and +25.7 (+20.7%), respectively. Using 150kV DECT reduced these changes to -20.5 (-12.2%) for arterial and -23.15 (-12.6%) for venous phases. Cohen’s d was lowest for normal arterial phase (+0.55) and highest for 90kV arterial phase (+1.9). Based on T-scores, 120 patients were classified as healthy, 108 as osteopenic, and 60 as osteoporotic. The lowest number of reclassifications occurred in arterial (n=92) and venous (n=104) phases. For arterial phase, 44 patients shifted from osteoporosis to osteopenia; for venous phase, 52 shifted similarly. High-kV acquisition reduced these reclassifications (n=24 arterial, n=32 venous) but increased shifts from healthy to osteopenia.
Conclusion: Intravenous contrast significantly affects HU-based osteoporosis assessment, leading to reclassifications, especially from osteopenia to healthy. Using 150kV DECT can partially reduce these reclassifications, though it may incorrectly shift healthy cases toward osteopenia.
Limitations: Modifying kV settings is not immediately possible without dedicated equipment
Funding for this study: No funding was received for this study
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Consent waived due to retrospective nature

Author Block: H. Ha¹, H. Lim², M. Park¹; ¹Anyang-Si/KR, ²Seoul/KR
Purpose: This study aimed to assess the diagnostic performance of osteoporosis prediction by the combination of autosegmentation of the proximal femur and machine learning analysis with a reference standard of dual-energy X-ray absorptiometry (DXA)
Methods or Background: Abdomen-pelvic CT scans were retrospectively analyzed from 1,122 patients who received both DXA and abdomen-pelvic computed tomography (APCT) scan from January 2018 to December 2020. The study cohort consisted of a training cohort and a temporal validation cohort. The left proximal femur was automatically segmented, and a prediction model was built by machine-learning analysis using a random forest (RF) analysis and 854 PyRadiomics features. The technical success rate of autosegmentation, diagnostic test, area under the receiver operator characteristics curve (AUC), and precision recall curve (AUC-PR) analysis were used to analyze the training and validation cohorts.
Results or Findings: The osteoporosis prevalence of the training and validation cohorts was 24.5%, and 10.3%, respectively. The technical success rate of autosegmentation of the proximal femur was 99.7%. In the diagnostic test, the training and validation cohorts showed 78.4% vs. 63.3% sensitivity, 89.4% vs. 98.1% specificity. The prediction performance to identify osteoporosis within the groups used for training and validation cohort was high and the AUC and AUC-PR to forecast the occurrence of osteoporosis within the training and validation cohorts were 90.8% [95% confidence interval (CI), 88.4–93.2%] vs. 78.0% (95% CI, 76.0–79.9%) and 94.6% (95% CI, 89.3–99.8%) vs. 88.8% (95% CI, 86.2–91.5%), respectively.
Conclusion: The osteoporosis prediction model using autosegmentation of proximal femur and machine-learning analysis with PyRadiomics features on APCT showed excellent diagnostic feasibility and technical success.
Limitations: The limitation of this study was that there was an imbalance in the sex ratio of osteoporosis patients and that this was a single-center study.
Funding for this study: This work was supported by the Central Medical Service (CMS) Research Fund. The specific grant number was not assigned by the company or funder (Central Medical Service Company, Ltd., Seoul, Korea).
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study was approved by the institutional review board of Hallym University Sacred Heart Hospital (No. HALLYM 2020-12-015), and the need for informed consent was waived due to the nature of the retrospective analysis.

Author Block: G. Vignati, M. Moscatelli, R. Fabrizio, R. Pascuzzo, C. Foschini, F. Doniselli, D. Aquino, F. Mazzi, L. M. Sconfienza; Milan/IT
Purpose: To analyze radiomics features extracted from thigh fat fraction (FF) maps in order to identify common patterns in neuromuscular disorders across different patients.
Methods or Background: Radiomics features of the classes “shape”, “first-order”, and “gray-level co-occurrence matrix” (GLCM) were extracted from the thigh FF maps of all patients for each of the 13 VOIs using PyRadiomics, with a fixed bin size of 32.
A preliminary feature selection step was necessary due to the large number of features extracted (n=1305) relative to the limited number of patients (n=25).
After feature selection, the sparse K-means clustering algorithm was applied: it is a clustering approach that identifies relevant features while performing clustering.
Finally, Uniform Manifold Approximation and Projection (UMAP) was used to visualize the selected features and statistical analyses were done with R (version 4.3.1) using the caret, sparcl, and umap packages.
Results or Findings: The sparse K-means algorithm identified two clusters of 14 and 11 patients, respectively, based on 60 selected radiomic features from 8 muscles.
Clinical diagnosis of patients affected by neuromuscolar disorders is compared with cluster assignment and distinctive features were observed between genetic and inflammatory/autoimmune myopathies.
Conclusion: This study successfully utilized radiomics features from thigh fat fraction (FF) maps to identify distinct patterns in neuromuscolar disorders across different patients suggesting that radiomic analysis could be a valuable tool for understanding and classifying muscle disorders in clinical settings.
Limitations: Limitations of this study are the small sample size (only 25 patients), the large number of initial radiomic features (n=1305) and the feature selection process, which could introduce bias. The study also relied on manually segmented regions of interest (VOIs), which may introduce variability in the analysis
Funding for this study: None
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information: None

Author Block: C. S. Mathy, L. V. Gast, T. Gerhalter, M. Türk, T. Bäuerle, A. Doerfler, M. Uder, A. M. Nagel, R. Schröder; Erlangen/DE
Purpose: To characterize patterns of skeletal muscle changes in myofibrillar myopathy (MFM), a group of rare neuromuscular diseases with desmin-positive aggregates and myofibrillar degeneration, using multiparametric MRI.
Methods or Background: Less affected lower leg of nine patients with genetically confirmed MFM due to FLNC (n=5), DES (n=2) and LDB3 (n=2) mutations (50.9±8.6 years, 6m, 3f) and 10 healthy controls (50.0±11.0 years, 6m, 4f) were examined. 1H-MRI at 3 T included T1-weighted and T2-weigthed STIR for qualitative assessment of fatty replacement/edema, Dixon-type sequence for proton-density fat fraction (PDFF) quantification and diffusion-tensor imaging (DTI) for characterization of (micro-)structural changes. 39K/23Na-MRI acquisition-weighted Stack-of-Stars sequences at 7 T allowed after partial-volume and relaxation correction quantification of apparent tissue sodium/potassium concentrations (aTSC/aTPC).
Results or Findings: Muscular fatty replacement and edema-like alterations were highly variant intermuscular and interindividual. 35/63 of elevated muscle compartments of patients with MFM were highly fatty replaced (PDFF >50%). Calculated apparent diffusion coefficients (ADCs) from DTI were reduced in gastrocnemius lateralis (GL), peroneus (PER) and extensor digitorum longus (EDL) muscles (p = 0.04 – 0.03) when excluding highly fatty replaced muscles, but simulations showed that this behavior could primarily be attributed to increasing PDFF. Fat-corrected aTSC were increased in all muscle regions (mean all muscles: 55.6±16.3 mM vs 23.2±5.5 mM, p <0.001), aTPC decreased in all regions but GL and PER (mean all muscles: 75.4±13.3 mM vs 108.9±9.9 mM, p <0.001).
Conclusion: Alterations of 39K/23Na ion homeostasis that go beyond changes caused by fatty-replacement in irreversible disease stages could be proved in patients with MFM. This could form the basis for a novel biomarker for determining early disease extent and disease response to therapies. (Micro-)structural changes were indistinguishable from mere fatty replacement changes.
Limitations: Low number of participants (MFM prevalence <1:100.000).
Funding for this study: C.S.M. and T.B. were founded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – 493624887 (Clinician Scientist Program NOTICE). Funding by the DFG is gratefully acknowledged (project 500888779 / RU5534 MR biosignatures at UHF).
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Approved by local ethic comittee of Friedrich-Alexander University Erlangen-Nuremberg

Author Block: L. Moore; Dublin/IE
Purpose: To identify and characterize the most influential publications relating to the musculoskeletal system and radiology. The number of citations an article receives is reflective of its impact in the scientific community.
Methods or Background: The top 100 most cited articles were identified using the Web of Science database. Data pertaining to the year of publication, publishing journal, journal impact factor, authorship, country of origin and institution were collected.
Results or Findings: The number of citations per article for the top 100 list ranged from 149 to 709 (median 208; mean 240). The average number of citations per year, per article, ranged from 5 to 60 (median 12, mean 26). The United States was the most common country of origin (n=74). The Journal with the greatest number of articles was Radiology (n=34). The University of California contributed the most articles (n=11).
Conclusion: This study presents a detailed analysis of the top 100 most-cited articles published in musculoskeletal radiology. It provides clinicians and researchers with insight into the current influential research papers in this field and the characteristics of those studies. It also highlights research trends and areas that may benefit from further research.
Limitations: The use of citation count is a source of bias; the more time that has passed since the publication of an article, the more likely it is to be cited over time. In order to mitigate this source of bias, the average citation count per year was also used. Some articles may have been inadvertently excluded as a result of search criteria used. In addition, using journal IF from one particular year (2024) does not allow for temporal fluctuations in IF. Furthermore, the potential bias of self-citation was not accounted for in this study.
Funding for this study: None.
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information: This article does not require ethics committee approval.