{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,8]],"date-time":"2026-10-08T21:04:12Z","timestamp":1791493452803,"version":"4.3.4"},"reference-count":26,"publisher":"Wiley","issue":"1","license":[{"start":{"date-parts":[[2025,1,27]],"date-time":"2025-01-27T00:00:00Z","timestamp":1737936000000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/onlinelibrary.wiley.com\/termsAndConditions#vor"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["82371921"],"award-info":[{"award-number":["82371921"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["81971578"],"award-info":[{"award-number":["81971578"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["onlinelibrary.wiley.com"],"crossmark-restriction":true},"short-container-title":["Magnetic Resonance Imaging"],"published-print":{"date-parts":[[2025,7]]},"abstract":"<jats:sec><jats:title>Background<\/jats:title><jats:p>The spinal column is a frequent site for metastases, affecting over 30% of solid tumor patients. Identifying the primary tumor is essential for guiding clinical decisions but often requires resource\u2010intensive diagnostics.<\/jats:p><\/jats:sec><jats:sec><jats:title>Purpose<\/jats:title><jats:p>To develop and validate artificial intelligence (AI) models using noncontrast MRI to identify primary sites of spinal metastases, aiming to enhance diagnostic efficiency.<\/jats:p><\/jats:sec><jats:sec><jats:title>Study Type<\/jats:title><jats:p>Retrospective.<\/jats:p><\/jats:sec><jats:sec><jats:title>Population<\/jats:title><jats:p>A total of 514 patients with pathologically confirmed spinal metastases (mean age, 59.3\u2009\u00b1\u200911.2\u2009years; 294 males) were included, split into a development set (360) and a test set (154).<\/jats:p><\/jats:sec><jats:sec><jats:title>Field Strength\/Sequence<\/jats:title><jats:p>Noncontrast sagittal MRI sequences (T1\u2010weighted, T2\u2010weighted, and fat\u2010suppressed T2) were acquired using 1.5\u2009T and 3\u2009T scanners.<\/jats:p><\/jats:sec><jats:sec><jats:title>Assessment<\/jats:title><jats:p>Two models were evaluated for identifying primary sites of spinal metastases: the expert\u2010derived features (EDF) model using radiologist\u2010identified imaging features and a ResNet50\u2010based deep learning (DL) model trained on noncontrast MRI. Performance was assessed using accuracy, precision, recall, F1 score, and the area under the receiver operating characteristic curve (ROC\u2010AUC) for top\u20101, top\u20102, and top\u20103 indicators.<\/jats:p><\/jats:sec><jats:sec><jats:title>Statistical Tests<\/jats:title><jats:p>Statistical analyses included Shapiro\u2013Wilk, <jats:italic>t<\/jats:italic> tests, Mann\u2013Whitney <jats:italic>U<\/jats:italic> test, and chi\u2010squared tests. ROC\u2010AUCs were compared via DeLong tests, with 95% confidence intervals from 1000 bootstrap replications and significance at <jats:italic>P<\/jats:italic>\u2009&lt;\u20090.05.<\/jats:p><\/jats:sec><jats:sec><jats:title>Results<\/jats:title><jats:p>The EDF model outperformed the DL model in top\u20103 accuracy (0.88 vs. 0.69) and AUC (0.80 vs. 0.71). Subgroup analysis showed superior EDF performance for common sites like lung and kidney (e.g., kidney F1: 0.94 vs. 0.76), while the DL model had higher recall for rare sites like thyroid (0.80 vs. 0.20). SHapley Additive exPlanations (SHAP) analysis identified sex (SHAP: \u22120.57 to 0.68), age (\u22120.48 to 0.98), T1WI signal intensity (\u22120.29 to 0.72), and pathological fractures (\u22120.76 to 0.25) as key features.<\/jats:p><\/jats:sec><jats:sec><jats:title>Data Conclusion<\/jats:title><jats:p>AI techniques using noncontrast MRI improve diagnostic efficiency for spinal metastases. The EDF model outperformed the DL model, showing greater clinical potential.<\/jats:p><\/jats:sec><jats:sec><jats:title>Plain Language Summary<\/jats:title><jats:p>Spinal metastases, or cancer spreading to the spine, are common in patients with advanced cancer, often requiring extensive tests to determine the original tumor site. Our study explored whether artificial intelligence could make this process faster and more accurate using noncontrast MRI scans. We tested two methods: one based on radiologists' expertise in identifying imaging features and another using a deep learning model trained to analyze MRI images. The expert\u2010based method was more reliable, correctly identifying the tumor site in 88% of cases when considering the top three likely diagnoses. This approach may help doctors reduce diagnostic time and improve patient care.<\/jats:p><\/jats:sec><jats:sec><jats:title>Level of Evidence<\/jats:title><jats:p>3<\/jats:p><\/jats:sec><jats:sec><jats:title>Technical Efficacy<\/jats:title><jats:p>Stage 2<\/jats:p><\/jats:sec>","DOI":"10.1002\/jmri.29720","type":"journal-article","created":{"date-parts":[[2025,1,27]],"date-time":"2025-01-27T11:21:39Z","timestamp":1737976899000},"page":"176-186","update-policy":"https:\/\/doi.org\/10.1002\/crossmark_policy","source":"Crossref","is-referenced-by-count":6,"title":["Identifying Primary Sites of Spinal Metastases: Expert\u2010Derived Features vs. <scp>ResNet50<\/scp> Model Using Nonenhanced <scp>MRI<\/scp>"],"prefix":"10.1002","volume":"62","author":[{"given":"Ke","family":"Liu","sequence":"first","affiliation":[{"name":"Department of Radiology Peking University Third Hospital  Beijing China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jinlai","family":"Ning","sequence":"additional","affiliation":[{"name":"Department of Informatics King's College London  London UK"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Siyuan","family":"Qin","sequence":"additional","affiliation":[{"name":"Department of Radiology Peking University Third Hospital  Beijing China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jun","family":"Xu","sequence":"additional","affiliation":[{"name":"Department of Radiology Peking University Third Hospital  Beijing China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-3499-4106","authenticated-orcid":false,"given":"Dapeng","family":"Hao","sequence":"additional","affiliation":[{"name":"Department of Radiology Affiliated Hospital of Qingdao University  Qingdao China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1860-2918","authenticated-orcid":false,"given":"Ning","family":"Lang","sequence":"additional","affiliation":[{"name":"Department of Radiology Peking University Third Hospital  Beijing China"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"311","published-online":{"date-parts":[[2025,1,27]]},"reference":[{"key":"e_1_2_7_2_1","doi-asserted-by":"publisher","DOI":"10.1007\/s00590-019-02541-0"},{"key":"e_1_2_7_3_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.jbo.2022.100446"},{"key":"e_1_2_7_4_1","doi-asserted-by":"publisher","DOI":"10.1016\/S0140-6736(17)33326-3"},{"key":"e_1_2_7_5_1","doi-asserted-by":"publisher","DOI":"10.3390\/diagnostics14090962"},{"key":"e_1_2_7_6_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.spinee.2022.07.089"},{"key":"e_1_2_7_7_1","doi-asserted-by":"publisher","DOI":"10.1200\/JCO.2017.72.7362"},{"key":"e_1_2_7_8_1","doi-asserted-by":"publisher","DOI":"10.1111\/os.12552"},{"key":"e_1_2_7_9_1","doi-asserted-by":"publisher","DOI":"10.1056\/NEJMra1303917"},{"key":"e_1_2_7_10_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.jare.2014.11.007"},{"key":"e_1_2_7_11_1","doi-asserted-by":"publisher","DOI":"10.1259\/bjr.20220267"},{"key":"e_1_2_7_12_1","doi-asserted-by":"publisher","DOI":"10.3390\/cancers16152700"},{"key":"e_1_2_7_13_1","doi-asserted-by":"publisher","DOI":"10.1186\/s40425-019-0720-z"},{"key":"e_1_2_7_14_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.mri.2019.02.013"},{"key":"e_1_2_7_15_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41586-021-03512-4"},{"key":"e_1_2_7_16_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41591-024-02915-w"},{"key":"e_1_2_7_17_1","doi-asserted-by":"crossref","unstructured":"HeK ZhangX RenS SunJ.Deep residual learning for image recognition. In:2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR); 2015 p 770\u2013778.","DOI":"10.1109\/CVPR.2016.90"},{"key":"e_1_2_7_18_1","unstructured":"IoffeS SzegedyC.Batch normalization: Accelerating deep network training by reducing internal covariate shift.arXiv2015;abs\/1502.03167."},{"key":"e_1_2_7_19_1","unstructured":"AgarapAF.Deep learning using rectified linear units (ReLU). arXiv 2018;abs\/1803.08375."},{"key":"e_1_2_7_20_1","unstructured":"LundbergSM ErionGG LeeS\u2010I.Consistent individualized feature attribution for tree ensembles. arXiv 2018;abs\/1802.03888."},{"key":"e_1_2_7_21_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41572-020-00216-3"},{"key":"e_1_2_7_22_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.wneu.2018.03.091"},{"key":"e_1_2_7_23_1","doi-asserted-by":"publisher","DOI":"10.1038\/s42256-019-0048-x"},{"key":"e_1_2_7_24_1","unstructured":"Doshi\u2010VelezF KimB.Towards a rigorous science of interpretable machine learning. arXiv: Machine Learning.2017."},{"key":"e_1_2_7_25_1","doi-asserted-by":"publisher","DOI":"10.21037\/atm.2018.05.32"},{"key":"e_1_2_7_26_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.wneu.2023.04.029"},{"key":"e_1_2_7_27_1","doi-asserted-by":"publisher","DOI":"10.3390\/cancers15112974"}],"container-title":["Journal of Magnetic Resonance Imaging"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/pdf\/10.1002\/jmri.29720","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,6,20]],"date-time":"2025-06-20T03:24:28Z","timestamp":1750389868000},"score":1,"resource":{"primary":{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/10.1002\/jmri.29720"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,1,27]]},"references-count":26,"journal-issue":{"issue":"1","published-print":{"date-parts":[[2025,7]]}},"alternative-id":["10.1002\/jmri.29720"],"URL":"https:\/\/doi.org\/10.1002\/jmri.29720","archive":["Portico"],"relation":{},"ISSN":["1053-1807","1522-2586"],"issn-type":[{"value":"1053-1807","type":"print"},{"value":"1522-2586","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,1,27]]},"assertion":[{"value":"2024-10-28","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-01-13","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-01-27","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}