High-Sensitivity Bone Tumor Classification in Radiographs Using Deep Learning and Ensemble Learning Strategies
Resumo
A classificação de tumores ósseos em imagens radiográficas é uma tarefa desafiadora devido à variabilidade visual das lesões. Este trabalho apresenta uma abordagem baseada em aprendizado profundo para a classificação binária de tumores ósseos em benignos e malignos, utilizando imagens radiográficas do conjunto de dados BTXRD. Foram avaliadas seis arquiteturas profundas pré-treinadas, incluindo modelos ConvNeXt, MaxViT e Swin Transformer. Além disso, foram investigadas estratégias de ensemble baseadas em soft voting, weighted voting e stacking. No conjunto de teste, o método proposto alcançou acurácia de 97,87%, F1-score de 97,92% e AUC-ROC de 99,59%. Além disso, o método obteve recall de 100,00% para a classe maligna.
Referências
Ajay, A. et al. (2026). Bonevisionnet: A deep learning approach for the classification of bone tumours. Computerized Medical Imaging and Graphics.
Bokhari, S. M. and Shafi, M. (2026). Deep learning driven radiographic classification of primary bone tumors using attention augmented hybrid models. Biomedical Signal Processing and Control, 113:108888.
He, Y. et al. (2020). Deep learning-based classification of primary bone tumors on radiographs: A preliminary study. eBioMedicine, 62.
Liu, Y. et al. (2025). Bone tumor recognition strategy. Scientific Reports.
Liu, Z. et al. (2022a). A convnet for the 2020s. arXiv.
Liu, Z., Hu, H., Lin, Y., Yao, Z., Xie, Z., Wei, Y., Ning, J., Cao, Y., Zhang, Z., Dong, L., Wei, F., and Guo, B. (2022b). Swin transformer v2: Scaling up capacity and resolution. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.
Mohammed, A. and Kora, R. (2023). A comprehensive review on ensemble deep learning: Opportunities and challenges. Journal of King Saud University - Computer and Information Sciences, 35(2):757–774.
National Cancer Institute (2018). Primary bone cancer fact sheet.
Pullan, J. E. and Lotfollahzadeh, S. (2025). Primary bone cancer. StatPearls.
Saran, S. and Phulware, R. H. (2022). World health organization classification of bone tumors (fifth edition): What a radiologist needs to know? Indian Journal of Musculoskeletal Radiology, 4:73–86.
Shorten, C. and Khoshgoftaar, T. (2019). Image data augmentation survey. Journal of Big Data.
Song, L. et al. (2024). Deep learning with multimodal images. Cancer Imaging.
Tu, Z., Talebi, H., Zhang, H., Yang, F., Milanfar, P., Bovik, A., and Li, Y. (2022). Maxvit: Multi-axis vision transformer. In European Conference on Computer Vision.
Wang, H. et al. (2025a). Deep learning models for bone tumors. npj Precision Oncology.
Wang, H., He, Y., Wan, L., et al. (2025b). Deep learning models in classifying primary bone tumors and bone infections based on radiographs. npj Precision Oncology, 9:72.
Yao, S. et al. (2025). A radiograph dataset for the classification, localization, and segmentation of primary bone tumors. Scientific Data.
Zhao et al. (2024). Convnext-based mask r-cnn for bone tumor segmentation. Journal of Bone Oncology.
Bokhari, S. M. and Shafi, M. (2026). Deep learning driven radiographic classification of primary bone tumors using attention augmented hybrid models. Biomedical Signal Processing and Control, 113:108888.
He, Y. et al. (2020). Deep learning-based classification of primary bone tumors on radiographs: A preliminary study. eBioMedicine, 62.
Liu, Y. et al. (2025). Bone tumor recognition strategy. Scientific Reports.
Liu, Z. et al. (2022a). A convnet for the 2020s. arXiv.
Liu, Z., Hu, H., Lin, Y., Yao, Z., Xie, Z., Wei, Y., Ning, J., Cao, Y., Zhang, Z., Dong, L., Wei, F., and Guo, B. (2022b). Swin transformer v2: Scaling up capacity and resolution. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.
Mohammed, A. and Kora, R. (2023). A comprehensive review on ensemble deep learning: Opportunities and challenges. Journal of King Saud University - Computer and Information Sciences, 35(2):757–774.
National Cancer Institute (2018). Primary bone cancer fact sheet.
Pullan, J. E. and Lotfollahzadeh, S. (2025). Primary bone cancer. StatPearls.
Saran, S. and Phulware, R. H. (2022). World health organization classification of bone tumors (fifth edition): What a radiologist needs to know? Indian Journal of Musculoskeletal Radiology, 4:73–86.
Shorten, C. and Khoshgoftaar, T. (2019). Image data augmentation survey. Journal of Big Data.
Song, L. et al. (2024). Deep learning with multimodal images. Cancer Imaging.
Tu, Z., Talebi, H., Zhang, H., Yang, F., Milanfar, P., Bovik, A., and Li, Y. (2022). Maxvit: Multi-axis vision transformer. In European Conference on Computer Vision.
Wang, H. et al. (2025a). Deep learning models for bone tumors. npj Precision Oncology.
Wang, H., He, Y., Wan, L., et al. (2025b). Deep learning models in classifying primary bone tumors and bone infections based on radiographs. npj Precision Oncology, 9:72.
Yao, S. et al. (2025). A radiograph dataset for the classification, localization, and segmentation of primary bone tumors. Scientific Data.
Zhao et al. (2024). Convnext-based mask r-cnn for bone tumor segmentation. Journal of Bone Oncology.
Publicado
19/10/2026
Como Citar
REIS, Francisco J. dos Santos; CARVALHO FILHO, Antonio O. de; ARAÚJO, José D. L..
High-Sensitivity Bone Tumor Classification in Radiographs Using Deep Learning and Ensemble Learning Strategies. In: ENCONTRO NACIONAL DE INTELIGÊNCIA ARTIFICIAL E COMPUTACIONAL (ENIAC), 23. , 2026, Cuiabá/MT.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2026
.
p. 1850-1861.
ISSN 2763-9061.
DOI: https://doi.org/10.5753/eniac.2026.31423.
