{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,18]],"date-time":"2026-08-18T02:42:03Z","timestamp":1787020923040,"version":"build-2736575974"},"reference-count":32,"publisher":"Springer Science and Business Media LLC","issue":"1","license":[{"start":{"date-parts":[[2025,12,24]],"date-time":"2025-12-24T00:00:00Z","timestamp":1766534400000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2026,1,27]],"date-time":"2026-01-27T00:00:00Z","timestamp":1769472000000},"content-version":"vor","delay-in-days":34,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"funder":[{"name":"Xinjiang Uygur Autonomous Region \"Tianshan talents\" medical and health high-level talent training plan","award":["TSYC202301B087"],"award-info":[{"award-number":["TSYC202301B087"]}]},{"name":"Natural Science Foundation General Project of the Science and Technology Department of Xinjiang Uygur Autonomous Region","award":["2021D01A161"],"award-info":[{"award-number":["2021D01A161"]}]}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["Eur J Med Res"],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>The differentiation of brucellar spondylitis (BS) from tuberculous spondylitis (TS) on magnetic resonance imaging (MRI) is a critical clinical challenge. While deep learning holds promise, the optimal architectural strategy for integrating information from multi-sequence MRI remains unclear. This study systematically compared distinct deep learning architectures to identify a valid and effective integration strategy for this diagnostic problem.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>\n                      In this retrospective, single-center diagnostic study, we included 235 patients with surgically and pathologically confirmed BS (\n                      <jats:italic>n<\/jats:italic>\n                      \u2009=\u200982) or TS (\n                      <jats:italic>n<\/jats:italic>\n                      \u2009=\u2009153) from January 2014 to December 2024. We systematically evaluated four distinct architectural strategies for processing sagittal T1-weighted, T2-weighted, and fat-suppressed MRI sequences: (1) baseline models trained on single sequences; (2) a single-branch model that fused sequences as input channels; (3) a heterogeneous multi-branch model using different backbones for each sequence; and (4) a homogeneous multi-branch model using identical backbones. Models were developed on patient-level data splits for training (70%), validation (15%), and internal testing (15%). The primary performance metric was the area under the receiver operating characteristic curve (AUC) on the test set. Statistical significance of performance differences between models was assessed using the DeLong test, with\n                      <jats:italic>P<\/jats:italic>\n                      values adjusted for multiple comparisons using the Benjamini\u2013Hochberg procedure.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>The single-branch fusion model, which treated the three sequences as channels in a single input, failed to learn, yielding performance equivalent to random chance (test AUC range: 0.474\u20130.538). In stark contrast, both the single-sequence and multi-branch architectures proved to be effective. The best single-sequence model achieved a test AUC of 0.765 (95% CI 0.759\u20130.771). The optimal multi-branch model, which successfully integrated all three sequences, achieved a comparable test AUC of 0.764 (95% CI 0.757\u20130.770).<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusions<\/jats:title>\n                    <jats:p>The choice of architecture for integrating multi-sequence MRI data is a critical determinant of model viability. Our findings demonstrate that naive channel wise fusion is an invalid strategy for this task. In contrast, both processing a single MRI sequence and utilizing a multi-branch parallel-processing architecture are valid and effective strategies, achieving comparable diagnostic performance. This study clarifies the architectural principles required for successfully applying deep learning to this multi-modal diagnostic challenge.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1186\/s40001-025-03731-9","type":"journal-article","created":{"date-parts":[[2025,12,24]],"date-time":"2025-12-24T14:03:54Z","timestamp":1766585034000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":1,"title":["Comprehensive comparative analysis of explainable deep learning model for differentiation of brucellar spondylitis and tuberculous spondylitis through MRI sequences"],"prefix":"10.1186","volume":"31","author":[{"given":"Parhat","family":"Yasin","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Abudouresuli","family":"Tuersun","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Anuar","family":"Ashir","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yerlan","family":"Makhambetov","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jie","family":"Sheng","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xinghua","family":"Song","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2025,12,24]]},"reference":[{"issue":"12","key":"3731_CR1","doi-asserted-by":"publisher","first-page":"2509","DOI":"10.1016\/j.spinee.2015.09.024","volume":"15","author":"H Erdem","year":"2015","unstructured":"Erdem H, Elaldi N, Batirel A, Aliyu S, Sengoz G, Pehlivanoglu F, et al. Comparison of brucellar and tuberculous spondylodiscitis patients: results of the multicenter \u201cBackbone-1 Study.\u201d Spine J. 2015;15(12):2509\u201317.","journal-title":"Spine J"},{"issue":"6","key":"3731_CR2","doi-asserted-by":"publisher","first-page":"739","DOI":"10.31616\/asj.2020.0262","volume":"15","author":"F Hammami","year":"2021","unstructured":"Hammami F, Koubaa M, Feki W, Chakroun A, Rekik K, Smaoui F, et al. Tuberculous and brucellar spondylodiscitis: comparative analysis of clinical, laboratory, and radiological features. Asian Spine J. 2021;15(6):739\u201346.","journal-title":"Asian Spine J"},{"issue":"10","key":"3731_CR3","doi-asserted-by":"publisher","first-page":"4289","DOI":"10.1007\/s10067-021-05752-1","volume":"40","author":"S Lan","year":"2021","unstructured":"Lan S, He Y, Tiheiran M, Liu W, Guo H. The angiopoietin-like protein 4: a promising biomarker to distinguish brucella spondylitis from tuberculous spondylitis. Clin Rheumatol. 2021;40(10):4289\u201394.","journal-title":"Clin Rheumatol"},{"key":"3731_CR4","doi-asserted-by":"publisher","DOI":"10.1155\/2016\/8903635","volume":"2016","author":"B Yang","year":"2016","unstructured":"Yang B, Hu H, Chen J, He X, Li H. The evaluation of the clinical, laboratory, and radiological findings of 16 cases of brucellar spondylitis. Biomed Res Int. 2016;2016:8903635.","journal-title":"Biomed Res Int"},{"issue":"4","key":"3731_CR5","doi-asserted-by":"publisher","first-page":"762","DOI":"10.3904\/kjim.2013.019","volume":"31","author":"EJ Joo","year":"2016","unstructured":"Joo EJ, Yeom JS, Ha YE, Park SY, Lee CS, Kim ES, et al. Diagnostic yield of computed tomography-guided bone biopsy and clinical outcomes of tuberculous and pyogenic spondylitis. Korean J Intern Med. 2016;31(4):762\u201371.","journal-title":"Korean J Intern Med"},{"issue":"1","key":"3731_CR6","doi-asserted-by":"publisher","first-page":"96","DOI":"10.1016\/j.carj.2018.09.003","volume":"70","author":"M Sertic","year":"2019","unstructured":"Sertic M, Parkes L, Mattiassi S, Pritzker K, Gardam M, Murphy K. The efficacy of computed tomography-guided percutaneous spine biopsies in determining a causative organism in cases of suspected infection: a systematic review. Can Assoc Radiol J. 2019;70(1):96\u2013103.","journal-title":"Can Assoc Radiol J"},{"issue":"3","key":"3731_CR7","doi-asserted-by":"publisher","first-page":"e244","DOI":"10.1016\/j.ejrad.2011.02.007","volume":"81","author":"CM Heyer","year":"2012","unstructured":"Heyer CM, Brus LJ, Peters SA, Lemburg SP. Efficacy of CT-guided biopsies of the spine in patients with spondylitis\u2013an analysis of 164 procedures. Eur J Radiol. 2012;81(3):e244-249.","journal-title":"Eur J Radiol"},{"issue":"3","key":"3731_CR8","doi-asserted-by":"publisher","first-page":"126","DOI":"10.1093\/ons\/opab148","volume":"21","author":"L Harris","year":"2021","unstructured":"Harris L, Rajashekar D, Sharma P, David KM. Performance of computed tomography-guided spine biopsy for the diagnosis of malignancy and infection. Oper Neurosurg (Hagerstown). 2021;21(3):126\u201330.","journal-title":"Oper Neurosurg (Hagerstown)"},{"issue":"2","key":"3731_CR9","doi-asserted-by":"publisher","first-page":"113","DOI":"10.1097\/BRS.0000000000001697","volume":"42","author":"M Gao","year":"2017","unstructured":"Gao M, Sun J, Jiang Z, Cui X, Liu X, Wang G, et al. Comparison of tuberculous and brucellar spondylitis on magnetic resonance images. Spine (Phila Pa 1976). 2017;42(2):113\u201321.","journal-title":"Spine (Phila Pa 1976)"},{"issue":"9","key":"3731_CR10","doi-asserted-by":"publisher","first-page":"1183","DOI":"10.1016\/j.acra.2018.01.028","volume":"25","author":"X Liu","year":"2018","unstructured":"Liu X, Li H, Jin C, Niu G, Guo B, Chen Y, et al. Differentiation between brucellar and tuberculous spondylodiscitis in the acute and subacute stages by MRI: a retrospective observational study. Acad Radiol. 2018;25(9):1183\u20139.","journal-title":"Acad Radiol"},{"key":"3731_CR11","doi-asserted-by":"publisher","DOI":"10.1016\/j.ejrad.2024.111655","volume":"178","author":"J Chen","year":"2024","unstructured":"Chen J, Guo X, Liu X, Sheng Y, Li F, Li H, et al. Differentiation of tuberculous and brucellar spondylitis using conventional MRI-based deep learning algorithms. Eur J Radiol. 2024;178:111655.","journal-title":"Eur J Radiol"},{"issue":"1","key":"3731_CR12","doi-asserted-by":"publisher","first-page":"13124","DOI":"10.1038\/s41598-018-31486-3","volume":"8","author":"K Kim","year":"2018","unstructured":"Kim K, Kim S, Lee YH, Lee SH, Lee HS, Kim S. Performance of the deep convolutional neural network based magnetic resonance image scoring algorithm for differentiating between tuberculous and pyogenic spondylitis. Sci Rep. 2018;8(1):13124.","journal-title":"Sci Rep"},{"issue":"8","key":"3731_CR13","doi-asserted-by":"publisher","first-page":"300060523119515","DOI":"10.1177\/03000605231195156","volume":"51","author":"W Wang","year":"2023","unstructured":"Wang W, Fan Z, Zhen J. MRI radiomics-based evaluation of tuberculous and brucella spondylitis. J Int Med Res. 2023;51(8):3000605231195156.","journal-title":"J Int Med Res"},{"issue":"6","key":"3731_CR14","doi-asserted-by":"publisher","first-page":"508","DOI":"10.1016\/j.jor.2019.08.014","volume":"16","author":"A Aljawadi","year":"2019","unstructured":"Aljawadi A, Jahangir N, Jeelani A, Ferguson Z, Niazi N, Arnall F, et al. Management of pyogenic spinal infection, review of literature. J Orthop. 2019;16(6):508\u201312.","journal-title":"J Orthop"},{"key":"3731_CR15","unstructured":"Gildenblat J, contributors: PyTorch library for CAM methods. GitHub."},{"issue":"3","key":"3731_CR16","doi-asserted-by":"publisher","first-page":"837","DOI":"10.2307\/2531595","volume":"44","author":"ER DeLong","year":"1988","unstructured":"DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44(3):837\u201345.","journal-title":"Biometrics"},{"issue":"1","key":"3731_CR17","doi-asserted-by":"publisher","first-page":"289","DOI":"10.1111\/j.2517-6161.1995.tb02031.x","volume":"57","author":"Y Benjamini","year":"1995","unstructured":"Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Roy Stat Soc Ser B (Methodol). 1995;57(1):289\u2013300.","journal-title":"J Roy Stat Soc Ser B (Methodol)"},{"key":"3731_CR18","doi-asserted-by":"publisher","DOI":"10.3389\/fsurg.2022.955761","volume":"9","author":"P Yasin","year":"2022","unstructured":"Yasin P, Mardan M, Xu T, Cai X, Abulizi Y, Wang T, et al. Development and validation of a diagnostic model for differentiating tuberculous spondylitis from brucellar spondylitis using machine learning: a retrospective cohort study. Front Surg. 2022;9:955761.","journal-title":"Front Surg"},{"key":"3731_CR19","doi-asserted-by":"publisher","first-page":"5585","DOI":"10.2147\/JIR.S429593","volume":"16","author":"P Yasin","year":"2023","unstructured":"Yasin P, Mardan M, Abliz D, Xu T, Keyoumu N, Aimaiti A, et al. The potential of a CT-based machine learning radiomics analysis to differentiate Brucella and pyogenic spondylitis. J Inflamm Res. 2023;16:5585\u2013600.","journal-title":"J Inflamm Res"},{"key":"3731_CR20","doi-asserted-by":"publisher","DOI":"10.1016\/j.ebiom.2021.103402","volume":"68","author":"FR Eweje","year":"2021","unstructured":"Eweje FR, Bao B, Wu J, Dalal D, Liao WH, He Y, et al. Deep learning for classification of bone lesions on routine MRI. EBioMedicine. 2021;68:103402.","journal-title":"EBioMedicine"},{"issue":"8","key":"3731_CR21","doi-asserted-by":"publisher","first-page":"1944","DOI":"10.1158\/1078-0432.CCR-19-0374","volume":"26","author":"IL Xi","year":"2020","unstructured":"Xi IL, Zhao Y, Wang R, Chang M, Purkayastha S, Chang K, et al. Deep learning to distinguish benign from malignant renal lesions based on routine MR imaging. Clin Cancer Res. 2020;26(8):1944\u201352.","journal-title":"Clin Cancer Res"},{"key":"3731_CR22","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2023.103044","volume":"92","author":"L Han","year":"2024","unstructured":"Han L, Tan T, Zhang T, Huang Y, Wang X, Gao Y, et al. Synthesis-based imaging-differentiation representation learning for multi-sequence 3D\/4D MRI. Med Image Anal. 2024;92:103044.","journal-title":"Med Image Anal"},{"issue":"1","key":"3731_CR23","doi-asserted-by":"publisher","first-page":"20331","DOI":"10.1038\/s41598-020-77389-0","volume":"10","author":"J Lee","year":"2020","unstructured":"Lee J, Wang N, Turk S, Mohammed S, Lobo R, Kim J, et al. Discriminating pseudoprogression and true progression in diffuse infiltrating glioma using multi-parametric MRI data through deep learning. Sci Rep. 2020;10(1):20331.","journal-title":"Sci Rep"},{"issue":"35","key":"3731_CR24","doi-asserted-by":"publisher","first-page":"5978","DOI":"10.3748\/wjg.v27.i35.5978","volume":"27","author":"R Stollmayer","year":"2021","unstructured":"Stollmayer R, Budai BK, Toth A, Kalina I, Hartmann E, Szoldan P, et al. Diagnosis of focal liver lesions with deep learning-based multi-channel analysis of hepatocyte-specific contrast-enhanced magnetic resonance imaging. World J Gastroenterol. 2021;27(35):5978\u201388.","journal-title":"World J Gastroenterol"},{"issue":"2","key":"3731_CR25","doi-asserted-by":"publisher","first-page":"1243","DOI":"10.1007\/s00330-019-06417-z","volume":"30","author":"N Aldoj","year":"2020","unstructured":"Aldoj N, Lukas S, Dewey M, Penzkofer T. Semi-automatic classification of prostate cancer on multi-parametric MR imaging using a multi-channel 3D convolutional neural network. Eur Radiol. 2020;30(2):1243\u201353.","journal-title":"Eur Radiol"},{"issue":"1","key":"3731_CR26","doi-asserted-by":"publisher","first-page":"173","DOI":"10.1002\/jmri.28008","volume":"56","author":"J Jian","year":"2022","unstructured":"Jian J, Li Y, Xia W, He Z, Zhang R, Li H, et al. MRI-based multiple instance convolutional neural network for increased accuracy in the differentiation of borderline and malignant epithelial ovarian tumors. J Magn Reson Imaging. 2022;56(1):173\u201381.","journal-title":"J Magn Reson Imaging"},{"issue":"1","key":"3731_CR27","doi-asserted-by":"publisher","first-page":"181","DOI":"10.1186\/s13014-020-01618-z","volume":"15","author":"L Bielak","year":"2020","unstructured":"Bielak L, Wiedenmann N, Berlin A, Nicolay NH, Gunashekar DD, Hagele L, et al. Convolutional neural networks for head and neck tumor segmentation on 7-channel multiparametric MRI: a leave-one-out analysis. Radiat Oncol. 2020;15(1):181.","journal-title":"Radiat Oncol"},{"issue":"1","key":"3731_CR28","doi-asserted-by":"publisher","first-page":"125","DOI":"10.1186\/s40001-021-00598-4","volume":"26","author":"H Guo","year":"2021","unstructured":"Guo H, Lan S, He Y, Tiheiran M, Liu W. Differentiating brucella spondylitis from tuberculous spondylitis by the conventional MRI and MR T2 mapping: a prospective study. Eur J Med Res. 2021;26(1):125.","journal-title":"Eur J Med Res"},{"key":"3731_CR29","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1016\/j.neuroimage.2019.03.068","volume":"196","author":"S Aslani","year":"2019","unstructured":"Aslani S, Dayan M, Storelli L, Filippi M, Murino V, Rocca MA, et al. Multi-branch convolutional neural network for multiple sclerosis lesion segmentation. Neuroimage. 2019;196:1\u201315.","journal-title":"Neuroimage"},{"key":"3731_CR30","doi-asserted-by":"publisher","DOI":"10.1016\/j.jvcir.2020.102880","volume":"72","author":"X Liu","year":"2020","unstructured":"Liu X, Zhou X, Qian X. Transparency-guided ensemble convolutional neural network for the stratification between pseudoprogression and true progression of glioblastoma multiform in MRI. J Vis Commun Image Represent. 2020;72:102880.","journal-title":"J Vis Commun Image Represent"},{"issue":"7","key":"3731_CR31","doi-asserted-by":"publisher","first-page":"2153","DOI":"10.1109\/TBME.2021.3137561","volume":"69","author":"M Rahimpour","year":"2022","unstructured":"Rahimpour M, Bertels J, Radwan A, Vandermeulen H, Sunaert S, Vandermeulen D, et al. Cross-modal distillation to improve MRI-based brain tumor segmentation with missing MRI sequences. IEEE Trans Biomed Eng. 2022;69(7):2153\u201364.","journal-title":"IEEE Trans Biomed Eng"},{"key":"3731_CR32","doi-asserted-by":"publisher","first-page":"166","DOI":"10.3389\/fonc.2020.00166","volume":"10","author":"Y Ye","year":"2020","unstructured":"Ye Y, Cai Z, Huang B, He Y, Zeng P, Zou G, et al. Fully-automated segmentation of nasopharyngeal carcinoma on dual-sequence MRI using convolutional neural networks. Front Oncol. 2020;10:166.","journal-title":"Front Oncol"}],"container-title":["European Journal of Medical Research"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/link.springer.com\/article\/10.1186\/s40001-025-03731-9","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1186\/s40001-025-03731-9.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1186\/s40001-025-03731-9.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,1,27]],"date-time":"2026-01-27T12:31:56Z","timestamp":1769517116000},"score":1,"resource":{"primary":{"URL":"https:\/\/link.springer.com\/10.1186\/s40001-025-03731-9"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,12,24]]},"references-count":32,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2026,12]]}},"alternative-id":["3731"],"URL":"https:\/\/doi.org\/10.1186\/s40001-025-03731-9","relation":{},"ISSN":["2047-783X"],"issn-type":[{"value":"2047-783X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,12,24]]},"assertion":[{"value":"6 February 2025","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"16 December 2025","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"24 December 2025","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"This study adhered to the principles outlined in the Helsinki Declaration and received approval from the Ethics Committees of several institutions: The Sixth Affiliated Hospital of Xinjiang Medical University. The Ethical Committee waived the requirement for informed consent due to the use of de-identified data, which posed no potential risk to patients and had no connection to the researchers. All procedures were conducted in compliance with relevant guidelines and regulations, and no biological specimens were involved in this study.","order":2,"name":"Ethics","group":{"name":"EthicsHeading","label":"Ethics approval and consent to participate"}},{"value":"Not applicable.","order":3,"name":"Ethics","group":{"name":"EthicsHeading","label":"Consent for publication"}},{"value":"Not applicable.","order":4,"name":"Ethics","group":{"name":"EthicsHeading","label":"Clinical trial number"}},{"value":"The authors declare that the study was carried out without any commercial or financial relationships that may be served as a potential conflict of interest.","order":5,"name":"Ethics","group":{"name":"EthicsHeading","label":"Competing interests"}}],"article-number":"146"}}