{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"institution":[{"name":"medRxiv"}],"indexed":{"date-parts":[[2026,3,26]],"date-time":"2026-03-26T20:20:21Z","timestamp":1774556421573,"version":"3.50.1"},"posted":{"date-parts":[[2025,1,20]]},"group-title":"Radiology and Imaging","reference-count":30,"publisher":"openRxiv","license":[{"start":{"date-parts":[[2025,1,20]],"date-time":"2025-01-20T00:00:00Z","timestamp":1737331200000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by-nc-nd\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"accepted":{"date-parts":[[2025,1,20]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                <jats:sec>\n                  <jats:title>Purpose<\/jats:title>\n                  <jats:p>Given the high prevalence of vertebral fractures post-radiotherapy in patients with metastatic spine disease, accurate and rapid muscle segmentation could support efforts to quantify muscular changes due to disease or treatment and enable biomechanical modeling for assessments of vertebral loading to improve personalized evaluation of vertebral fracture risk. This study presents a deep-learning approach for segmenting the complete volume of the trunk muscles from clinical CT images trained using sparsely annotated data.<\/jats:p>\n                <\/jats:sec>\n                <jats:sec>\n                  <jats:title>Materials and Methods<\/jats:title>\n                  <jats:p>we extracted 2,009 axial CT images at the midpoint of each vertebral level (T4 to L4) from clinical CT of 148 cancer patients. The key extensor and flexor muscles (up to 8 muscles per side) were manually contoured and labeled per image in the thoracic and lumbar regions. We first trained a 2D nnU-Net deep-learning model on these labels to segment key extensor and flexor muscles. Using these sparse annotations per spine, we trained the model to segment each muscle\u2019s entire 3D volume<\/jats:p>\n                <\/jats:sec>\n                <jats:sec>\n                  <jats:title>Results<\/jats:title>\n                  <jats:p>The proposed method achieved comparable performance to manual segmentations, as assessed by expert radiologists, with a mean Dice score above 0.769. Significantly, the model drastically reduced segmentation time, from 4.3-6.5 hours for manual segmentation of 14 single axial CT images to approximately 1 minute for segmenting the complete thoracic-abdominal 3D volume.<\/jats:p>\n                <\/jats:sec>\n                <jats:sec>\n                  <jats:title>Conclusion<\/jats:title>\n                  <jats:p>The approach demonstrates high potential for automating 3D muscle segmentation, significantly reducing the manual intervention required for generating musculoskeletal models, and could be instrumental in enhancing clinical decision-making and patient care in radiation oncology.<\/jats:p>\n                <\/jats:sec>\n                <jats:sec>\n                  <jats:title>Summary<\/jats:title>\n                  <jats:p>A deep learning 2D nnU-Net model, trained on a sparse set of 2D muscle annotations, successfully segmented the entire volume of 20 thoracolumbar muscles from cancer patients\u2019 clinical CT data. The model showed a remarkable increase in segmentation efficacy and generalizability, achieving comparable performance to manual segmentations in delineating each muscle anatomy.<\/jats:p>\n                <\/jats:sec>\n                <jats:sec>\n                  <jats:title>Key Points<\/jats:title>\n                  <jats:list list-type=\"simple\">\n                    <jats:list-item>\n                      <jats:p>\u25aa A deep learning model (2D nnU-Net), developed using a sparse set of single axial CT-slice at each mid-per vertebral level, containing manual image annotation of 20 thoracic and lumbar muscles, achieved comparable performance to manual segmentations, as assessed by expert radiologists, with a mean Dice score above 0.769.<\/jats:p>\n                    <\/jats:list-item>\n                    <jats:list-item>\n                      <jats:p>\u25aa The model drastically reduced segmentation time, from 4.3-6.5 hours for manual segmentation of 14 single axial CT images to approximately 1 minute for segmenting the complete thoracic-abdominal 3D volume.<\/jats:p>\n                    <\/jats:list-item>\n                    <jats:list-item>\n                      <jats:p>\u25aa Radiologist assessment based on a Likert scale (0-5) for clinical acceptability of the muscle anatomical segmentation showed strong model performance for a representative sample of clinical CT data (a (mean(SD) of 4.66 (0.73)) and external data (4.66 (0.73).<\/jats:p>\n                    <\/jats:list-item>\n                  <\/jats:list>\n                <\/jats:sec>","DOI":"10.1101\/2025.01.13.25319967","type":"posted-content","created":{"date-parts":[[2025,1,21]],"date-time":"2025-01-21T01:15:19Z","timestamp":1737422119000},"source":"Crossref","is-referenced-by-count":0,"title":["Automated Segmentation of Trunk Musculature with a Deep CNN Trained from Sparse Annotations in Radiation Therapy Patients with Metastatic Spine Disease"],"prefix":"10.64898","author":[{"given":"Vy","family":"Hong","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-4193-9578","authenticated-orcid":false,"given":"Steve","family":"Pieper","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Joanna","family":"James","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5171-7445","authenticated-orcid":false,"given":"Dennis E","family":"Anderson","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9982-3194","authenticated-orcid":false,"given":"Csaba","family":"Pinter","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Yi Shuen","family":"Chang","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8565-2141","authenticated-orcid":false,"given":"Bulent","family":"Aslan","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8916-396X","authenticated-orcid":false,"given":"David","family":"Kozono","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0009-0004-5001-933X","authenticated-orcid":false,"given":"Patrick F","family":"Doyle","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8037-8836","authenticated-orcid":false,"given":"Sarah","family":"Caplan","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7543-1284","authenticated-orcid":false,"given":"Heejoo","family":"Kang","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Tracy","family":"Balboni","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1085-3205","authenticated-orcid":false,"given":"Alexander","family":"Spektor","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Mai Anh","family":"Huynh","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3624-0079","authenticated-orcid":false,"given":"Mario","family":"Keko","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7227-7058","authenticated-orcid":false,"given":"Ron","family":"Kikinis","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9014-1998","authenticated-orcid":false,"given":"David B","family":"Hackney","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-9198-8039","authenticated-orcid":false,"given":"Ron N","family":"Alkalay","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"54368","reference":[{"issue":"3","key":"2025012208402278000_2025.01.13.25319967v1.1","doi-asserted-by":"crossref","first-page":"314","DOI":"10.1093\/neuros\/nyx493","article-title":"Vertebral Compression Fracture After Spine Stereotactic Body Radiation Therapy: A Review of the Pathophysiology and Risk Factors","volume":"83","year":"2018","journal-title":"Neurosurgery"},{"key":"2025012208402278000_2025.01.13.25319967v1.2","doi-asserted-by":"crossref","first-page":"11","DOI":"10.1016\/j.jocn.2017.04.004","article-title":"Contemporary spinal oncology treatment paradigms and outcomes for metastatic tumors to the spine: A systematic review of breast, prostate, renal, and lung metastases","volume":"41","year":"2017","journal-title":"J Clin Neurosci"},{"key":"2025012208402278000_2025.01.13.25319967v1.3","doi-asserted-by":"publisher","DOI":"10.1016\/j.jbiomech.2004.11.030"},{"key":"2025012208402278000_2025.01.13.25319967v1.4","doi-asserted-by":"publisher","DOI":"10.1007\/s11914-016-0305-4"},{"key":"2025012208402278000_2025.01.13.25319967v1.5","doi-asserted-by":"publisher","DOI":"10.3389\/fbioe.2022.866970"},{"issue":"11","key":"2025012208402278000_2025.01.13.25319967v1.6","first-page":"5","article-title":"Instability and impending instability of the thoracolumbar spine in patients with spinal metastases: a systematic review","volume":"38","year":"2011","journal-title":"Int J Oncol"},{"issue":"8","key":"2025012208402278000_2025.01.13.25319967v1.7","doi-asserted-by":"crossref","first-page":"081003","DOI":"10.1115\/1.4030408","article-title":"Development and Validation of a Musculoskeletal Model of the Fully Articulated Thoracolumbar Spine and Rib Cage","volume":"137","year":"2015","journal-title":"J Biomech Eng"},{"key":"2025012208402278000_2025.01.13.25319967v1.8","doi-asserted-by":"publisher","DOI":"10.1007\/s00256-019-03289-8"},{"key":"2025012208402278000_2025.01.13.25319967v1.9","doi-asserted-by":"crossref","unstructured":"Edwards, K. , et al., Abdominal muscle segmentation from CT using a convolutional neural network. Proc SPIE Int Soc Opt Eng, 2020. 11317.","DOI":"10.1117\/12.2549406"},{"key":"2025012208402278000_2025.01.13.25319967v1.10","doi-asserted-by":"crossref","unstructured":"Ackermans, L. , et al., Deep Learning Automated Segmentation for Muscle and Adipose Tissue from Abdominal Computed Tomography in Polytrauma Patients. Sensors (Basel), 2021. 21(6).","DOI":"10.3390\/s21062083"},{"issue":"3","key":"2025012208402278000_2025.01.13.25319967v1.11","doi-asserted-by":"crossref","first-page":"1384","DOI":"10.21037\/qims-22-330","article-title":"A deep learning model based on the attention mechanism for automatic segmentation of abdominal muscle and fat for body composition assessment","volume":"13","year":"2023","journal-title":"Quant Imaging Med Surg"},{"key":"2025012208402278000_2025.01.13.25319967v1.12","doi-asserted-by":"crossref","first-page":"81","DOI":"10.1007\/978-981-13-1396-7_7","article-title":"Muscle Segmentation for Orthopedic Interventions","volume":"1093","year":"2018","journal-title":"Adv Exp Med Biol"},{"key":"2025012208402278000_2025.01.13.25319967v1.13","doi-asserted-by":"crossref","first-page":"83","DOI":"10.3991\/ijoe.v18i10.31531","article-title":"A Brief Survey on Weakly Supervised Semantic Segmentation","volume":"18","year":"2022","journal-title":"Int. J. Online Biomed. Eng"},{"key":"2025012208402278000_2025.01.13.25319967v1.14","doi-asserted-by":"crossref","unstructured":"Cai, H ., et al. 3D Medical Image Segmentation with Sparse Annotation via Cross-Teaching between 3D and 2D Networks. in International Conference on Medical Image Computing and Computer-Assisted Intervention. 2023.","DOI":"10.1007\/978-3-031-43898-1_59"},{"key":"2025012208402278000_2025.01.13.25319967v1.15","doi-asserted-by":"crossref","unstructured":"\u00c7i\u00e7ek, \u00d6 ., et al. 3D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotation. in International Conference on Medical Image Computing and Computer-Assisted Intervention. 2016.","DOI":"10.1007\/978-3-319-46723-8_49"},{"key":"2025012208402278000_2025.01.13.25319967v1.16","doi-asserted-by":"crossref","first-page":"111710","DOI":"10.1016\/j.jbiomech.2023.111710","article-title":"Dependence of trunk muscle size and position on age, height, and weight in a multi-ethnic cohort of middle-aged and older men and women","volume":"157","year":"2023","journal-title":"J Biomech"},{"key":"2025012208402278000_2025.01.13.25319967v1.17","doi-asserted-by":"publisher","DOI":"10.1109\/42.952724"},{"key":"2025012208402278000_2025.01.13.25319967v1.18","doi-asserted-by":"crossref","unstructured":"Boerner, T. , et al., ACCESS: Advancing Innovation: NSF\u2019s Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support. 2023. 173\u2013176.","DOI":"10.1145\/3569951.3597559"},{"key":"2025012208402278000_2025.01.13.25319967v1.19","doi-asserted-by":"crossref","unstructured":"Hancock, D. , et al., Jetstream2: Accelerating cloud computing via Jetstream. 2021. 1\u20138.","DOI":"10.1145\/3437359.3465565"},{"key":"2025012208402278000_2025.01.13.25319967v1.20","unstructured":"(ACRIN), L.S.S.g.L.a.t.A.C.o.R.I.N., National Lung Screening Trial (NLST) N.C.I. (NCI), Editor. 2009."},{"key":"2025012208402278000_2025.01.13.25319967v1.21","doi-asserted-by":"publisher","DOI":"10.1016\/0031-9384(94)00276-B"},{"key":"2025012208402278000_2025.01.13.25319967v1.22","doi-asserted-by":"crossref","first-page":"812159","DOI":"10.3389\/fonc.2022.812159","article-title":"Establishment and Validation of Pre-Therapy Cervical Vertebrae Muscle Quantification as a Prognostic Marker of Sarcopenia in Patients With Head and Neck Cancer","volume":"12","year":"2022","journal-title":"Front Oncol"},{"key":"2025012208402278000_2025.01.13.25319967v1.23","doi-asserted-by":"crossref","first-page":"28","DOI":"10.1016\/j.oraloncology.2016.09.006","article-title":"Feasibility of using head and neck CT imaging to assess skeletal muscle mass in head and neck cancer patients","volume":"62","year":"2016","journal-title":"Oral Oncol"},{"issue":"Suppl 3","key":"2025012208402278000_2025.01.13.25319967v1.24","first-page":"S30","article-title":"Clinical evaluation of automated segmentation for body composition analysis on abdominal L3 CT slices in polytrauma patients","volume":"53","year":"2022","journal-title":"Injury"},{"key":"2025012208402278000_2025.01.13.25319967v1.25","doi-asserted-by":"crossref","first-page":"142","DOI":"10.1016\/j.ejrad.2018.10.031","article-title":"Intra- and interobserver variability in skeletal muscle measurements using computed tomography images","volume":"109","year":"2018","journal-title":"Eur J Radiol"},{"key":"2025012208402278000_2025.01.13.25319967v1.26","doi-asserted-by":"crossref","unstructured":"Mortellaro, S. , et al., Quantitative and Qualitative Radiological Assessment of Sarcopenia and Cachexia in Cancer Patients: A Systematic Review. J Pers Med, 2024. 14(3).","DOI":"10.3390\/jpm14030243"},{"key":"2025012208402278000_2025.01.13.25319967v1.27","doi-asserted-by":"publisher","DOI":"10.1038\/s41592-020-01008-z"},{"key":"2025012208402278000_2025.01.13.25319967v1.28","doi-asserted-by":"publisher","DOI":"10.1148\/ryai.230024"},{"issue":"9","key":"2025012208402278000_2025.01.13.25319967v1.29","doi-asserted-by":"crossref","first-page":"569","DOI":"10.1016\/j.clon.2020.03.011","article-title":"Falls and Cancer","volume":"32","year":"2020","journal-title":"Clin Oncol (R Coll Radiol"},{"issue":"1","key":"2025012208402278000_2025.01.13.25319967v1.30","first-page":"S139","article-title":"Deep Learning for Multi-Tissue Segmentation and Fully Automatic Personalized Biomechanical Models from BACPAC Clinical Lumbar Spine MRI","volume":"24","year":"2022","journal-title":"Pain Medicine"}],"container-title":[],"original-title":[],"link":[{"URL":"https:\/\/syndication.highwire.org\/content\/doi\/10.1101\/2025.01.13.25319967","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,1,15]],"date-time":"2026-01-15T20:25:13Z","timestamp":1768508713000},"score":1,"resource":{"primary":{"URL":"http:\/\/medrxiv.org\/lookup\/doi\/10.1101\/2025.01.13.25319967"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,1,20]]},"references-count":30,"URL":"https:\/\/doi.org\/10.1101\/2025.01.13.25319967","relation":{},"subject":[],"published":{"date-parts":[[2025,1,20]]},"subtype":"preprint"}}