{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"institution":[{"name":"medRxiv"}],"indexed":{"date-parts":[[2026,8,31]],"date-time":"2026-08-31T18:22:08Z","timestamp":1788200528361,"version":"build-2803163510"},"posted":{"date-parts":[[2026,8,5]]},"group-title":"Radiology and Imaging","reference-count":27,"publisher":"openRxiv","license":[{"start":{"date-parts":[[2026,8,5]],"date-time":"2026-08-05T00:00:00Z","timestamp":1785888000000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"US National Institutes of Health and the US National Cancer Institute","award":["1R01CA235564"],"award-info":[{"award-number":["1R01CA235564"]}]},{"name":"US National Institutes of Health and the US National Cancer Institute","award":["P30CA016672"],"award-info":[{"award-number":["P30CA016672"]}]},{"name":"US National Institutes of Health and the US National Cancer Institute","award":["T32CA261856"],"award-info":[{"award-number":["T32CA261856"]}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"accepted":{"date-parts":[[2026,8,27]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                <jats:sec>\n                  <jats:title>Purpose<\/jats:title>\n                  <jats:p>To build a 3D U-Net model, BioDeformUNet, to predict the liver\u2019s deformation vector field (DVF) in near real-time, for efficient intra-procedural evaluation of the minimal ablative margin (MAM).<\/jats:p>\n                <\/jats:sec>\n                <jats:sec>\n                  <jats:title>Materials and Methods<\/jats:title>\n                  <jats:p>\n                    This retrospective study included 170 contrast-enhanced computed tomography (CECT) image pairs from 157 patients who underwent liver ablation treatment between 2020-2024. Each data instance included one pre-ablation CECT (pre-CECT) and one post-ablation CECT (post-CECT). BioDeformUNet was trained under the guidance of DVFs generated by a biomechanical model-based deformable image registration (DIR) algorithm using a loss function that focused on large liver deformations. Data were split patient-wise into training (92-93 patients), validation (23-24 patients), and testing sets (42 patients). We compared our method\u2019s performance with two deep learning-based DIR methods: VoxelMorph and VFA. Evaluation metrics included: target registration error (TRE), Dice similarity coefficient (DSC), Minimum Ablation Margin (MAM), and inference time. For BioDeformUNet, we additionally evaluated the accuracy of the deformed tumor center-of-mass mapping by comparing the predicted tumor center location with that generated by Morfeus. A mapping error less than 3.0 mm (corresponding to the voxel size) was considered accurate. We used the Wilcoxon signed-rank test to assess the significancy of each test result. Our code is available at\n                    <jats:ext-link xmlns:xlink=\"http:\/\/www.w3.org\/1999\/xlink\" ext-link-type=\"uri\" xlink:href=\"https:\/\/github.com\/XinyueZhang831\/BioDeformUNET\">https:\/\/github.com\/XinyueZhang831\/BioDeformUNET<\/jats:ext-link>\n                    .\n                  <\/jats:p>\n                <\/jats:sec>\n                <jats:sec>\n                  <jats:title>Results<\/jats:title>\n                  <jats:p>The TRE of BioDeformUNet was not significantly different from Morfeus (3.31\u00b11.17 BioDeformUNet; 3.23\u00b11.06 Morfeus; p-value=0.41). The BioDeformUNet DVF magnitude was within 3.0 mm of Morfeus DVF for an average of 91.9 \u00b1 17.6% (mean \u00b1 SD) of the voxels. Tumor mapping errors greater than 3.0 mm occurred in only 8 cases. The inference time of BioDeformUNet was 0.6s per image pair, 0.2s for VoxelMorph, 0.3s for VFA, and 20.2s for Morfeus.<\/jats:p>\n                <\/jats:sec>\n                <jats:sec>\n                  <jats:title>Conclusion<\/jats:title>\n                  <jats:p>BioDeformUNet achieved a similar performance to the biomechanical model-based algorithm but required fewer computational operations, resulting in a 34\u00d7 speedup in DVF computation.<\/jats:p>\n                <\/jats:sec>","DOI":"10.64898\/2026.08.03.26359612","type":"posted-content","created":{"date-parts":[[2026,8,5]],"date-time":"2026-08-05T18:25:20Z","timestamp":1785954320000},"source":"Crossref","is-referenced-by-count":0,"title":["BioDeformUNet: A Deep Learning Model for Biomechanically Informed Liver Image Registration"],"prefix":"10.64898","author":[{"given":"Xinyue","family":"Zhang","sequence":"first","affiliation":[{"name":"The University of Texas MD Anderson Cancer Center, Rice University, 1515 Holcombe Blvd, Houston, TX 77030"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Caleb S.","family":"O\u2019Connor","sequence":"additional","affiliation":[{"name":"The University of Texas MD Anderson Cancer Center, Rice University, 1515 Holcombe Blvd, Houston, TX 77030"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0009-3015-3570","authenticated-orcid":false,"given":"Austin","family":"Castelo","sequence":"additional","affiliation":[{"name":"The University of Texas MD Anderson Cancer Center, Rice University, 1515 Holcombe Blvd, Houston, TX 77030"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"McKell E.","family":"Woodland","sequence":"additional","affiliation":[{"name":"The University of Texas MD Anderson Cancer Center, Rice University, 1515 Holcombe Blvd, Houston, TX 77030"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Bilel","family":"Daoud","sequence":"additional","affiliation":[{"name":"The University of Texas MD Anderson Cancer Center, Rice University, 1515 Holcombe Blvd, Houston, TX 77030"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Iwan","family":"Paolucci","sequence":"additional","affiliation":[{"name":"The University of Texas MD Anderson Cancer Center, Rice University, 1515 Holcombe Blvd, Houston, TX 77030"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jessica","family":"Albuquerque","sequence":"additional","affiliation":[{"name":"The University of Texas MD Anderson Cancer Center, Rice University, 1515 Holcombe Blvd, Houston, TX 77030"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Mais","family":"Altaie","sequence":"additional","affiliation":[{"name":"The University of Texas MD Anderson Cancer Center, Rice University, 1515 Holcombe Blvd, Houston, TX 77030"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Noreen","family":"Siddiqi","sequence":"additional","affiliation":[{"name":"The University of Texas MD Anderson Cancer Center, Rice University, 1515 Holcombe Blvd, Houston, TX 77030"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ankit B.","family":"Patel","sequence":"additional","affiliation":[{"name":"Baylor College of Medicine, Rice University"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Bruno C.","family":"Odisio","sequence":"additional","affiliation":[{"name":"The University of Texas MD Anderson Cancer Center, Rice University, 1515 Holcombe Blvd, Houston, TX 77030"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Kristy K.","family":"Brock","sequence":"additional","affiliation":[{"name":"The University of Texas MD Anderson Cancer Center, Rice University, 1515 Holcombe Blvd, Houston, TX 77030"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"54368","reference":[{"key":"2026083110350717000_2026.08.03.26359612v2.1","unstructured":"Herszenyi L , Tulassay Z . Epidemiology of gastrointestinal and liver tumors. Eur Rev Med Pharmacol Sci. 2010;14(4)."},{"issue":"5","key":"2026083110350717000_2026.08.03.26359612v2.2","doi-asserted-by":"crossref","first-page":"442","DOI":"10.1016\/S2468-1253(25)00024-X","article-title":"Software-based versus visual assessment of the minimal ablative margin in patients with liver tumours undergoing percutaneous thermal ablation (COVER-ALL): a randomised phase 2 trial","volume":"10","year":"2025","journal-title":"Lancet Gastroenterol Hepatol"},{"key":"2026083110350717000_2026.08.03.26359612v2.3","unstructured":"National Comprehensive Cancer Network. NCCN Clinical Practice Guidelines in Oncology: Colon Cancer (version 3.2024). National Comprehensive Cancer Network. February 27, 2025."},{"key":"2026083110350717000_2026.08.03.26359612v2.4","doi-asserted-by":"publisher","DOI":"10.1016\/S1470-2045(24)00660-0"},{"key":"2026083110350717000_2026.08.03.26359612v2.5","doi-asserted-by":"publisher","DOI":"10.1016\/j.jhep.2021.11.018"},{"issue":"4","key":"2026083110350717000_2026.08.03.26359612v2.6","doi-asserted-by":"crossref","first-page":"192","DOI":"10.1053\/j.tvir.2013.08.002","article-title":"Tumor ablation: common modalities and general practices","volume":"16","year":"2013","journal-title":"Tech Vasc Interv Radiol"},{"key":"2026083110350717000_2026.08.03.26359612v2.7","doi-asserted-by":"crossref","unstructured":"Lin YM , Paolucci I , Albuquerque Marques Silva J , et al. Ablative margin quantification using deformable versus rigid image registration in colorectal liver metastasis thermal ablation: a retrospective single-center study. Eur Radiol. Published online 2024:1\u201310.","DOI":"10.1007\/s00330-024-10632-8"},{"key":"2026083110350717000_2026.08.03.26359612v2.8","doi-asserted-by":"publisher","DOI":"10.1109\/TMI.2002.808367"},{"issue":"2","key":"2026083110350717000_2026.08.03.26359612v2.9","doi-asserted-by":"crossref","first-page":"e221373","DOI":"10.1148\/radiol.221373","article-title":"Ablative margins of colorectal liver metastases using deformable CT image registration and autosegmentation","volume":"307","year":"2023","journal-title":"Radiology"},{"issue":"10","key":"2026083110350717000_2026.08.03.26359612v2.10","doi-asserted-by":"crossref","first-page":"6226","DOI":"10.1002\/mp.15147","article-title":"A novel use of biomechanical model-based deformable image registration (DIR) for assessing colorectal liver metastases ablation outcomes","volume":"48","year":"2021","journal-title":"Med Phys"},{"key":"2026083110350717000_2026.08.03.26359612v2.11","doi-asserted-by":"publisher","DOI":"10.1118\/1.1915012"},{"issue":"11","key":"2026083110350717000_2026.08.03.26359612v2.12","doi-asserted-by":"crossref","first-page":"1365","DOI":"10.1007\/s11709-022-0882-5","article-title":"Bridging finite element and deep learning: High-resolution stress distribution prediction in structural components","volume":"16","year":"2022","journal-title":"Frontiers of Structural and Civil Engineering"},{"key":"2026083110350717000_2026.08.03.26359612v2.13","doi-asserted-by":"publisher","DOI":"10.1109\/TMI.2019.2897538"},{"key":"2026083110350717000_2026.08.03.26359612v2.14","doi-asserted-by":"crossref","unstructured":"Chen J , Frey EC , He Y , Segars WP , Li Y , Du Y . Transmorph: Transformer for unsupervised medical image registration. Med Image Anal. 2022;82:102615.","DOI":"10.1016\/j.media.2022.102615"},{"issue":"6","key":"2026083110350717000_2026.08.03.26359612v2.15","first-page":"64001","article-title":"Vector field attention for deformable image registration","volume":"11","year":"2024","journal-title":"Journal of Medical Imaging"},{"key":"2026083110350717000_2026.08.03.26359612v2.16","doi-asserted-by":"publisher","DOI":"10.1002\/mp.12307"},{"issue":"1","key":"2026083110350717000_2026.08.03.26359612v2.17","doi-asserted-by":"crossref","first-page":"100464","DOI":"10.1016\/j.adro.2020.04.023","article-title":"Automated contouring of contrast and noncontrast computed tomography liver images with fully convolutional networks","volume":"6","year":"2021","journal-title":"Adv Radiat Oncol"},{"key":"2026083110350717000_2026.08.03.26359612v2.18","doi-asserted-by":"publisher","DOI":"10.1186\/s12880-021-00677-2"},{"key":"2026083110350717000_2026.08.03.26359612v2.19","doi-asserted-by":"crossref","unstructured":"Lowekamp BC , Chen DT , Ib\u00e1\u00f1ez L , Blezek D . The design of SimpleITK. Front Neuroinform. 2013;7:45.","DOI":"10.3389\/fninf.2013.00045"},{"key":"2026083110350717000_2026.08.03.26359612v2.20","doi-asserted-by":"crossref","unstructured":"Beare R , Lowekamp B , Yaniv Z . Image segmentation, registration and characterization in R with SimpleITK. J Stat Softw. 2018;86.","DOI":"10.18637\/jss.v086.i08"},{"key":"2026083110350717000_2026.08.03.26359612v2.21","doi-asserted-by":"publisher","DOI":"10.1007\/s10278-017-0037-8"},{"key":"2026083110350717000_2026.08.03.26359612v2.22","doi-asserted-by":"crossref","unstructured":"Roh\u00e9 MM , Datar M , Heimann T , Sermesant M , Pennec X . SVF-Net: learning deformable image registration using shape matching. In: Medical Image Computing and Computer Assisted Intervention\u2212 MICCAI 2017: 20th International Conference, Quebec City, QC, Canada, September 11-13, 2017, Proceedings, Part I 20. Springer; 2017:266\u2013274.","DOI":"10.1007\/978-3-319-66182-7_31"},{"issue":"xxxx","key":"2026083110350717000_2026.08.03.26359612v2.23","first-page":"20170788","article-title":"Respiratory motion correction for free-breathing 3D abdominal MRI using CNN-based image registration: a feasibility study","volume":"91","year":"2018","journal-title":"Br J Radiol"},{"issue":"10","key":"2026083110350717000_2026.08.03.26359612v2.24","doi-asserted-by":"crossref","first-page":"5935","DOI":"10.1002\/mp.15163","article-title":"Detection of vessel bifurcations in CT scans for automatic objective assessment of deformable image registration accuracy","volume":"48","year":"2021","journal-title":"Med Phys"},{"issue":"9","key":"2026083110350717000_2026.08.03.26359612v2.25","doi-asserted-by":"crossref","first-page":"znae165","DOI":"10.1093\/bjs\/znae165","article-title":"Identification of A0 minimum ablative margins for colorectal liver metastases: multicentre, retrospective study using deformable CT registration and artificial intelligence-based autosegmentation","volume":"111","year":"2024","journal-title":"British Journal of Surgery"},{"issue":"2","key":"2026083110350717000_2026.08.03.26359612v2.26","doi-asserted-by":"crossref","first-page":"e221373","DOI":"10.1148\/radiol.221373","article-title":"Ablative margins of colorectal liver metastases using deformable CT image registration and autosegmentation","volume":"307","year":"2023","journal-title":"Radiology"},{"key":"2026083110350717000_2026.08.03.26359612v2.27","doi-asserted-by":"publisher","DOI":"10.1007\/s00330-019-06609-7"}],"container-title":[],"original-title":[],"link":[{"URL":"https:\/\/syndication.highwire.org\/content\/doi\/10.64898\/2026.08.03.26359612","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,31]],"date-time":"2026-08-31T17:35:14Z","timestamp":1788197714000},"score":1,"resource":{"primary":{"URL":"http:\/\/medrxiv.org\/lookup\/doi\/10.64898\/2026.08.03.26359612"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,5]]},"references-count":27,"URL":"https:\/\/doi.org\/10.64898\/2026.08.03.26359612","relation":{},"subject":[],"published":{"date-parts":[[2026,8,5]]},"subtype":"preprint"}}