Opportunistic Osteoporosis Screening from Routine Knee Radiographs Using a Deep Learning-Based Texture-Branch Cross-Attention Network

Basit Öğe Kaydı
dc.contributor.authorKurt, Mehmet
dc.contributor.authorGedik, Mehmet Ali
dc.contributor.authorKestel, Selim
dc.contributor.authorKurt, Hasan Samet
dc.contributor.authorYilmaz, Selcuk
dc.contributor.authorOzdemir, Cuneyt
dc.date.accessioned2026-10-23T18:14:12Z
dc.date.available2026-10-23T18:14:12Z
dc.date.issued2026
dc.departmentSiirt Üniversitesi
dc.description.abstractThis study systematically investigates the impact of architectural components on deep learning-based osteoporosis screening using knee radiographs from a single-center archive. A total of 580 posteroanterior and lateral knee radiographs (normal: n = 389; osteoporosis: n = 191) were included. ImageNet-pretrained backbones (ResNet50V2, VGG16, and DenseNet121) were progressively extended with texture channel packages, dual-stream (global + texture) cross-attention fusion, ordinal-aware supervised contrastive loss, and input-conditioned learnable test-time augmentation (L-TTA). All models were evaluated by fivefold stratified cross-validation using the out-of-fold area under the ROC curve (OOF AUC). TBA-Net v1 achieved the highest performance (OOF AUC = 0.8727), exceeding the strongest baseline VGG16 (0.8497) by 0.0230 AUC and ResNet50V2 (0.8101) and DenseNet121 (0.8092) by 0.0626 and 0.0635 AUC. Ablation analysis demonstrated that the texture branch, cross-attention mechanism, and contrastive loss reduced performance when introduced sequentially, but produced a synergistic effect when jointly optimized; a Shapley decomposition showed that this gain arises from the interaction between components rather than from any component alone. L-TTA Net (OOF AUC = 0.8427) generated clinically interpretable attention patterns. At the current dataset scale, performance plateaued at approximately 0.87 OOF AUC, indicating that further improvements are constrained by sampling variability rather than architectural complexity. Architectural components in small medical imaging datasets should be evaluated within a holistic optimization framework rather than in isolation, and learnable test-time augmentation weights provide interpretability beyond predictive performance. The framework is positioned as a triage aid for prioritizing densitometry referral rather than as a substitute for it.
dc.identifier.doi10.1007/s10278-026-02276-5
dc.identifier.issn2948-2925
dc.identifier.issn2948-2933
dc.identifier.orcid0000-0002-9252-5888
dc.identifier.orcid0009-0008-1495-3319
dc.identifier.pmid42736491
dc.identifier.scopus2-s2.0-105050664361
dc.identifier.scopusqualityN/A
dc.identifier.urihttps://doi.org/10.1007/s10278-026-02276-5
dc.identifier.urihttps://hdl.handle.net/20.500.12604/12326
dc.identifier.wosWOS:001874086300001
dc.identifier.wosqualityQ2
dc.indekslendigikaynakWeb of Science
dc.indekslendigikaynakScopus
dc.indekslendigikaynakPubMed
dc.language.isoen
dc.publisherSpringer
dc.relation.ispartofJournal of Imaging Informatics in Medicine
dc.relation.publicationcategoryMakale - Uluslararası Hakemli Dergi - Kurum Öğretim Elemanı
dc.rightsinfo:eu-repo/semantics/closedAccess
dc.snmzKA_WOS_20261017
dc.subjectOsteoporosis Screening
dc.subjectKnee Radiograph
dc.subjectDeep Learning
dc.subjectCross-Attention Mechanism
dc.subjectTest-Time Augmentation
dc.subjectAblation Analysis
dc.titleOpportunistic Osteoporosis Screening from Routine Knee Radiographs Using a Deep Learning-Based Texture-Branch Cross-Attention Network
dc.typeArticle

Dosyalar

Koleksiyon

WOS İndeksli Yayınlar Koleksiyonu
PubMed İndeksli Yayın Koleksiyonu
Scopus İndeksli Yayınlar Koleksiyonu