Evaluation of paraspinal muscle quality using Hounsfield unit in simple elliptical regions of interest: correlation with magnetic resonance imaging-based intramuscular fat infiltration in spine surgery patients
- 1Nagoya University
Abstract
Study designRetrospective cohort study.
ObjectivesTo correlate Hounsfield unit (HU) values, using elliptical regions of interest (ROI), that can be easily defined in routine clinical practice with magnetic resonance imaging (MRI) T2-hyperintense area fraction, as a surrogate for paraspinal muscle fat infiltration and to establish specific HU screening thresholds that may be applied with standard picture archiving and communication system (PACS).
MethodsWe included 136 patients (71 men; 61.0 {+/-} 15.4 years) who underwent preoperative computed tomography (CT) and MRI within an 8-week period. Elliptical ROI HU values were measured at L2/3 and L4/5 for erector spinae, multifidus, and psoas major. MRI T2-hyperintense area fraction (Otsu thresholding) served as the fat infiltration reference. Linear mixed-effects (LME) models were used to assess the HU-T2 association and level-specific receiver operating characteristic (ROC) analyses (lower HU value side; n=136 per muscle-level) to identify thresholds for [≥]30% and [≥]50% infiltration criteria.
ResultsIntraclass coefficients = 0.709 (HU) and 0.857 (T2 fraction); Goutallier weighted kappa = 0.579. In the overall LME, {beta} was -0.880 HU per 1% T2-fraction increase (95% confidence interval -0.935 to -0.825; marginal R{superscript 2} =0.502); the association was steeper in multifidus ({beta} = -1.020) than in erector spinae ({beta} = -0.753). Psoas major (R = -0.226) was excluded from ROC analyses. Difference between L2/3 and L4/5 HU cutoffs was [~]20 HU. The [≥]50% criterion revealed higher discrimination.
ConclusionsElliptical ROI-based HU measurements may reliably screen paraspinal muscle fat infiltration in erector spinae and multifidus using standard PACS. Specific thresholds may allow practical preoperative evaluation without additional costs or radiation.
Related
- Evaluation of Intraperitoneal Instillation Combined with Port-Site Infiltration of 0.5% Bupivacaine for Postoperative Pain Management in Laparoscopic Cholecystectomy: A Prospective Randomized Controlled Trial
- Hounsfield Units–Anchored Trabecular Bone Attenuation on Low-Dose Chest CT Scanning: Opportunities and Gaps
- Association Between Hounsfield Units and Vertebral Body Re-collapse 3 Months After Posterior Fixation for Thoracolumbar Burst Fractures in Working-Age Patients: A Retrospective Register-Based Cohort Study
- Can Deltoid Hounsfield Unit/Supraspinatus Hounsfield Unit Predict Postoperative Rotator Cuff Retear?
- Underestimated value of Hounsfield unit in differentiating benign and malignant lymph nodes via EBUS-TBNA.
BibTeX
@misc{segi2026evaluation,
title = {{Evaluation of paraspinal muscle quality using Hounsfield unit in simple elliptical regions of interest: correlation with magnetic resonance imaging-based intramuscular fat infiltration in spine surgery patients}},
author = {N. Segi and Y. Okada and Y. Takeichi and S. Ito and J. Ouchida and Y. Nagatani and Y. Kagami and H. Tachi and K. Ohshima and K. Ogura and S. Imagama and H. Nakashima},
howpublished = {medRxiv},
year = {2026},
month = aug,
doi = {10.64898/2026.08.31.26361574},
url = {https://www.medrxiv.org/content/10.64898/2026.08.31.26361574v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Evaluation of paraspinal muscle quality using Hounsfield unit in simple elliptical regions of interest: correlation with magnetic resonance imaging-based intramuscular fat infiltration in spine surgery patients
AU - N. Segi
AU - Y. Okada
AU - Y. Takeichi
AU - S. Ito
AU - J. Ouchida
AU - Y. Nagatani
AU - Y. Kagami
AU - H. Tachi
AU - K. Ohshima
AU - K. Ogura
AU - S. Imagama
AU - H. Nakashima
PY - 2026
DA - 2026/08/31
PB - medRxiv
DO - 10.64898/2026.08.31.26361574
UR - https://www.medrxiv.org/content/10.64898/2026.08.31.26361574v1
AB - Study designRetrospective cohort study. ObjectivesTo correlate Hounsfield unit (HU) values, using elliptical regions of interest (ROI), that can be easily defined in routine clinical practice with magnetic resonance imaging (MRI) T2-hyperintense area fraction, as a surrogate for paraspinal muscle fat infiltration and to establish specific HU screening thresholds that may be applied with standard picture archiving and communication system (PACS). MethodsWe included 136 patients (71 men; 61.0 {+/-} 15.4 years) who underwent preoperative computed tomography (CT) and MRI within an 8-week period. Elliptical ROI HU values were measured at L2/3 and L4/5 for erector spinae, multifidus, and psoas major. MRI T2-hyperintense area fraction (Otsu thresholding) served as the fat infiltration reference. Linear mixed-effects (LME) models were used to assess the HU-T2 association and level-specific receiver operating characteristic (ROC) analyses (lower HU value side; n=136 per muscle-level) to identify thresholds for [≥]30% and [≥]50% infiltration criteria. ResultsIntraclass coefficients = 0.709 (HU) and 0.857 (T2 fraction); Goutallier weighted kappa = 0.579. In the overall LME, {beta} was -0.880 HU per 1% T2-fraction increase (95% confidence interval -0.935 to -0.825; marginal R{superscript 2} =0.502); the association was steeper in multifidus ({beta} = -1.020) than in erector spinae ({beta} = -0.753). Psoas major (R = -0.226) was excluded from ROC analyses. Difference between L2/3 and L4/5 HU cutoffs was [~]20 HU. The [≥]50% criterion revealed higher discrimination. ConclusionsElliptical ROI-based HU measurements may reliably screen paraspinal muscle fat infiltration in erector spinae and multifidus using standard PACS. Specific thresholds may allow practical preoperative evaluation without additional costs or radiation.
ER -