{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,29]],"date-time":"2026-09-29T04:58:42Z","timestamp":1790657922255,"version":"4.1.0"},"reference-count":34,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,9,29]],"date-time":"2026-09-29T00:00:00Z","timestamp":1790640000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Med."],"abstract":"<jats:sec>\n                    <jats:title>Introduction<\/jats:title>\n                    <jats:p>Medical image classification is fundamental to computer-aided diagnosis. Limited labeled samples, subtle inter-class differences, and heterogeneous lesion morphology make it difficult for a single representation to capture all relevant cues. This study investigates MedFuse, a simple dual-stream framework that combines convolutional neural network (CNN) features with representations from a pretrained vision foundation model (VFM).<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>The CNN branch is optimized on the target dataset to emphasize task-adaptive local texture and morphology, whereas the VFM branch is kept frozen to provide a stable global semantic representation. The two pooled feature vectors are concatenated and passed to a common classification head. Accordingly, MedFuse is positioned as a systematic and reproducible fusion baseline. The framework is evaluated across all 12 two-dimensional MedMNISTV2 subsets using multiple CNN-VFM backbone pairs and is compared with alternative weight- and attention-based fusion schemes.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>Simple late feature concatenation was competitive with, and often more stable than, more parameterized fusion modules across the evaluated settings. Moreover, increasing the size of the VFM backbone did not consistently improve classification performance.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Discussion<\/jats:title>\n                    <jats:p>These findings demonstrate that frozen foundation-model representations can complement trainable convolutional features in small-scale medical image classification, while also showing that increased fusion complexity or larger VFM backbones does not necessarily translate into better performance. MedFuse therefore provides a simple and reproducible baseline for systematically studying CNN-VFM feature complementarity in medical image classification.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fmed.2026.1901033","type":"journal-article","created":{"date-parts":[[2026,9,29]],"date-time":"2026-09-29T04:38:06Z","timestamp":1790656686000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["MedFuse: dual-stream fusion of convolutional and vision transformer-based features for enhanced medical image classification"],"prefix":"10.3389","volume":"13","author":[{"given":"Yajing","family":"Ren","sequence":"first","affiliation":[{"name":"Artificial Intelligence and Smart Mine Engineering Technology Center, Xinjiang Institute of Engineering","place":["Urumqi, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ling","family":"Hai","sequence":"additional","affiliation":[{"name":"Artificial Intelligence and Smart Mine Engineering Technology Center, Xinjiang Institute of Engineering","place":["Urumqi, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zheng","family":"Gu","sequence":"additional","affiliation":[{"name":"Artificial Intelligence and Smart Mine Engineering Technology Center, Xinjiang Institute of Engineering","place":["Urumqi, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Wen","family":"Liu","sequence":"additional","affiliation":[{"name":"Artificial Intelligence and Smart Mine Engineering Technology Center, Xinjiang Institute of Engineering","place":["Urumqi, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,9,29]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"2024","DOI":"10.14569\/IJACSA.2024.0150727","article-title":"Recent advances in medical image classification","volume":"15","author":"Dao","year":"2024","journal-title":"Int J Adv Comput Sci Appl"},{"key":"B2","doi-asserted-by":"crossref","DOI":"10.1007\/s10439-026-04241-3","article-title":"Overview of state-of-the-art learning-based classification methods in medical imaging","volume-title":"Ann Biomed Eng","author":"Ghaffar Nia","year":"2026"},{"key":"B3","doi-asserted-by":"publisher","first-page":"1487","DOI":"10.32604\/cmes.2025.070964","article-title":"Deep learning in medical image analysis: a comprehensive review of algorithms, trends, applications, and challenges","volume":"145","author":"Lepcha","year":"2026","journal-title":"Comput Model Eng Sci"},{"key":"B4","first-page":"111226","article-title":"Challenges in deep learning-based small organ segmentation a benchmarking perspective for medical research with limited datasets","volume-title":"Biomed Signal Process Control","author":"Konrad","year":"2026"},{"key":"B5","doi-asserted-by":"publisher","first-page":"105404","DOI":"10.1088\/1361-6501\/ae08d8","article-title":"Mit loss: medical image-aware transfer-calibrated loss for enhanced classification","volume":"36","author":"Pan","year":"2025","journal-title":"Meas Sci Technol"},{"key":"B6","doi-asserted-by":"publisher","first-page":"108634","DOI":"10.1016\/j.cmpb.2025.108634","article-title":"Why does my medical AI look at pictures of birds? Exploring the efficacy of transfer learning across domain boundaries","volume":"261","author":"Jonske","year":"2025","journal-title":"Comput Methods Programs Biomed"},{"key":"B7","doi-asserted-by":"publisher","first-page":"1082","DOI":"10.1007\/s44163-026-02129-8","article-title":"Frequency-aware sparse token vision transformer for medical image segmentation","volume":"6","author":"Sun","year":"2026","journal-title":"Discov Artif Intell"},{"key":"B8","first-page":"104214","article-title":"Prompt-guided foundation model tuning for pathology image classification","volume-title":"Med Image Anal","author":"Lin","year":"2026"},{"key":"B9","doi-asserted-by":"publisher","first-page":"41936","DOI":"10.1038\/s41598-025-25857-w","article-title":"Mednet: a lightweight attention-augmented CNN for medical image classification","volume":"15","author":"Ferdous","year":"2025","journal-title":"Sci Rep"},{"key":"B10","doi-asserted-by":"publisher","first-page":"54040","DOI":"10.1109\/ACCESS.2025.3554184","article-title":"EFFResNet-ViT: a fusion-based convolutional and vision transformer model for explainable medical image classification","volume":"13","author":"Hussain","year":"2025","journal-title":"IEEE Access"},{"key":"B11","doi-asserted-by":"publisher","first-page":"2040","DOI":"10.3390\/diagnostics15162040","article-title":"Neuro-bridge-X: a neuro-symbolic vision transformer with meta-XAI for interpretable leukemia diagnosis from peripheral blood smears","volume":"15","author":"Jammal","year":"2025","journal-title":"Diagnostics"},{"key":"B12","doi-asserted-by":"publisher","first-page":"1527980","DOI":"10.3389\/frai.2025.1527980","article-title":"Medalmighty: enhancing disease diagnosis with large vision model distillation","volume":"8","author":"Ren","year":"2025","journal-title":"Front Artif Intell"},{"key":"B13","doi-asserted-by":"publisher","first-page":"e18744400417835","DOI":"10.2174\/0118744400417835251022042920","article-title":"Multimodal medical image fusion: techniques, databases, evaluation metrics, and clinical applications -a comprehensive review","volume":"18","author":"Goswami","year":"2025","journal-title":"Open Neuroimag J"},{"key":"B14","article-title":"Towards optimal convolutional transfer learning architectures for breast lesion classification and ACL tear detection","volume-title":"arXiv [Preprint]","author":"Frees","year":"2025"},{"key":"B15","doi-asserted-by":"publisher","first-page":"94","DOI":"10.1186\/s13244-026-02271-y","article-title":"Medical image pretraining-based transfer learning for generalizable and robust diagnosis of bone tumors on radiographs: a multi-center study","volume":"17","author":"Li","year":"2026","journal-title":"Insights Imaging"},{"key":"B16","doi-asserted-by":"publisher","first-page":"21622","DOI":"10.1038\/s41598-025-05009-w","article-title":"Lessons learned from RadiologyNET foundation models for transfer learning in medical radiology","volume":"15","author":"Napravnik","year":"2025","journal-title":"Sci Rep"},{"key":"B17","doi-asserted-by":"publisher","first-page":"6512","DOI":"10.1038\/s41598-026-36652-6","article-title":"Clinical validation of lightweight CNN architectures for reliable multi-class classification of lung cancer using histopathological imaging techniques","volume":"16","author":"Raza","year":"2026","journal-title":"Sci Rep"},{"key":"B18","first-page":"2259","article-title":"Advancements in Medical Image Classification Through Fine-Tuning Natural Domain Foundation Models","author":"Mansoori","year":"2025"},{"key":"B19","article-title":"Scaling Self-Supervised and Cross-Modal Pretraining for Volumetric CT Transformers","volume-title":"arXiv [Preprint]","author":"Claessens","year":"2025"},{"key":"B20","doi-asserted-by":"publisher","first-page":"32","DOI":"10.1038\/s43856-025-01293-9","article-title":"Compact vision language models enable efficient and interpretable optical coherence tomography through layer-specific multimodal learning","volume":"6","author":"Haghighi","year":"2026","journal-title":"Commun. Med"},{"key":"B21","first-page":"1","article-title":"Evaluating the Feasibility of Vision-Language Models in Skin Cancer Detection: A Comparative Study with CNNs and Vision Transformers","author":"Chen","year":"2025"},{"key":"B22","doi-asserted-by":"publisher","first-page":"7467","DOI":"10.1038\/s41598-025-92054-0","article-title":"A dual-stream feature decomposition network with weight transformation for multi-modality image fusion","volume":"15","author":"Hu","year":"2025","journal-title":"Sci Rep"},{"key":"B23","doi-asserted-by":"publisher","first-page":"256","DOI":"10.1177\/01617346251362168","article-title":"Experimental assessment of conventional features, CNN-based features and ensemble schemes for discriminating benign versus malignant lesions on breast ultrasound images","volume":"47","author":"Bianconi","year":"2025","journal-title":"Ultrason Imaging"},{"key":"B24","doi-asserted-by":"publisher","first-page":"28273","DOI":"10.1109\/JIOT.2026.3679487","article-title":"AI-driven feature-enhanced stacking ensemble with global-context vision transformers for breast cancer classification in ultrasound images","volume":"13","author":"Vo","year":"2026","journal-title":"IEEE Internet Things J."},{"key":"B25","doi-asserted-by":"publisher","first-page":"1470","DOI":"10.3390\/electronics15071470","article-title":"Lightweight neural network ensemble models for medical image classification with MedMNIST dataset","volume":"15","author":"Prvan","year":"2026","journal-title":"Electronics (Basel)"},{"key":"B26","doi-asserted-by":"publisher","first-page":"128","DOI":"10.1186\/s12880-025-01667-4","article-title":"A systematic literature review: exploring the challenges of ensemble model for medical imaging","volume":"25","author":"Supriyadi","year":"2025","journal-title":"BMC Med Imaging"},{"key":"B27","article-title":"A Heterogeneous Ensemble for Multi-Center COVID-19 Classification from Chest CT Scans","volume-title":"arXiv [Preprint]","author":"Nilay","year":"2026"},{"key":"B28","doi-asserted-by":"publisher","first-page":"30","DOI":"10.3390\/iot6020030","article-title":"A hybrid learnable fusion of convnext and swin transformer for optimized image classification","volume":"6","author":"Qezelbash-Chamak","year":"2025","journal-title":"IoT"},{"key":"B29","doi-asserted-by":"publisher","first-page":"62","DOI":"10.3390\/jimaging12020062","article-title":"AACNN-ViT: adaptive attention-augmented convolutional and vision transformer fusion for lung cancer detection","volume":"12","author":"Rahman","year":"2026","journal-title":"J.Imaging"},{"key":"B30","doi-asserted-by":"publisher","first-page":"10033","DOI":"10.1038\/s41598-026-40906-8","article-title":"A multi-scale hybrid ResNet\u2013transformer with distance-aware learning for interpretable BI-RADS mammographic classification","volume":"16","author":"Singh","year":"2026","journal-title":"Sci Rep"},{"key":"B31","doi-asserted-by":"publisher","first-page":"16959","DOI":"10.1038\/s41598-026-43282-5","article-title":"Benchmarking MedViT and hybrid CNN\u2013ViT architectures for multi-label thoracic disease classification","volume":"16","author":"Agbo","year":"2026","journal-title":"Sci Rep"},{"key":"B32","doi-asserted-by":"publisher","first-page":"1960","DOI":"10.3390\/electronics15091960","article-title":"Pantograph wear classification via dual-backbone feature-fusion ensemble network","volume":"15","author":"Ullah","year":"2026","journal-title":"Electronics (Basel)"},{"key":"B33","doi-asserted-by":"publisher","first-page":"30","DOI":"10.3390\/ai7010030","article-title":"Explainable multi-modal medical image analysis through dual-stream multi-feature fusion and class-specific selection","volume":"7","author":"Ullah","year":"2026","journal-title":"AI"},{"key":"B34","doi-asserted-by":"publisher","first-page":"41","DOI":"10.1038\/s41597-022-01721-8","article-title":"MedMNIST v2\u2014A large-scale lightweight benchmark for 2D and 3D biomedical image classification","volume":"10","author":"Yang","year":"2023","journal-title":"Scientific Data"}],"container-title":["Frontiers in Medicine"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fmed.2026.1901033\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,29]],"date-time":"2026-09-29T04:38:34Z","timestamp":1790656714000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fmed.2026.1901033\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,29]]},"references-count":34,"alternative-id":["10.3389\/fmed.2026.1901033"],"URL":"https:\/\/doi.org\/10.3389\/fmed.2026.1901033","relation":{},"ISSN":["2296-858X"],"issn-type":[{"value":"2296-858X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,29]]},"article-number":"1901033"}}