Harnessing deep learning for enhanced detection of thymic epithelial tumors
DOI:
https://doi.org/10.14419/p2j8w965Keywords:
Self-Supervised Learning; Contrastive Learning; Adaptive Neuro-Evolution; Deep Learning; Feature Selection; Tumor Classification; TCGA-THYM; SimCLR; Moco; Medical ImagingAbstract
The increasing reliance on deep learning for medical image classification has significantly improved tumor detection. However, traditional deep learning models often suffer from redundant feature extraction, poor feature separability, and suboptimal classifica-tion performance. This study proposes an advanced hybrid framework integrating Adaptive Neuro-Evolutionary Pruned Embedding (ANPE) and Self-Supervised Embedding Transformation (SSET) for tumor classification to overcome these limitations. The ANPE module optimizes feature selection by employing multi-scale feature extraction, neuro-evolutionary pruning, and self-attention ranking to retain only the most discriminative features, thus enhancing classification accuracy. The SSET module, on the other hand, leverages contrastive learning techniques, specifically SimCLR (Simple Contrastive Learning) and MoCo (Momentum Con-trastive Learning), to improve feature representation and cluster separation. The proposed ANPE-SSET framework was evaluated on the TCGA-THYM dataset, a benchmark dataset for histopathological tumor classification. Experimental results demonstrate that ANPE-SSET outperforms baseline models such as CNN, ResNet, and Vision Transformers (ViTs). The proposed model achieved an accuracy of 96.2%, significantly surpassing CNN (85.3%), ResNet (90.5%), and ViT (92.1%). The AUC-ROC score of 98.3% high-lights its superior ability to distinguish between tumor classes. In conclusion, the ANPE-SSET hybrid model effectively integrates evolutionary feature selection and self-supervised contrastive learning, leading to state-of-the-art performance in tumor classifica-tion. The results demonstrate its superiority over traditional models, making it a promising approach for medical image analysis. Future work will focus on extending this framework to other medical imaging datasets and further optimizing it for clinical applica-tions.
References
- Boehme, L., Yayon, N., Teichmann, S. A., & Taghon, T. (2025). From cortex to medulla: Navigating the thymic landscape of T cell education. Mo-lecular Therapy. https://doi.org/10.1016/j.ymthe.2025.03.006.
- Dubey, S., & Sikarwar, S. S. (2025). Applications of AI in cancer detection—A review of the specific ways in which AI is being used to detect and diagnose various types of cancer. In AI in Disease Detection: Advancements and Applications (pp. 147–166). https://doi.org/10.1002/9781394278695.ch7.
- Lin, P., Lin, Y., Mai, Z., Zheng, Y., Zheng, J., Zhou, Z., Zhao, X., & Cui, L. (2025). Targeting cancer with precision: Strategical insights into TCR-engineered T cell therapies. Theranostics, 15(1), 300. https://doi.org/10.7150/thno.104594.
- Zeng, J., Lin, Z., Zhang, X., Zheng, T., Xu, H., & Liu, T. (2025). Leveraging artificial intelligence for neoantigen prediction. Cancer Research. https://doi.org/10.1158/0008-5472.CAN-24-2553.
- Moll, M., & Baumjohann, D. (2025). Boosting human immunology: Harnessing the potential of immune organoids. EMBO Molecular Medicine, 1–10. https://doi.org/10.1038/s44321-025-00193-8.
- Wang, G., Huang, J., Chen, H., Jiang, C., Jiang, L., Feng, W., & Tian, G. (2025). Exploring novel biomarkers and immunotherapeutic targets for biofeedback therapies to reveal the tumor-associated immune microenvironment through a multimetric analysis of kidney renal clear cell carcinoma. Discover Oncology, 16(1), 1–17. https://doi.org/10.1007/s12672-025-02090-5.
- Hao, D. (2025). Trustworthy deep learning on medical images (PhD dissertation, University of Pittsburgh).
- Meenakshi Devi, P., Gnanavel, S., Narayana, K. E., & Sangeethaa, S. N. (2024). Novel methods for diagnosing glaucoma: Segmenting optic discs and cups using ensemble learning algorithms and CDR ratio analysis. IETE Journal of Research, 70(8), 6828–6847. https://doi.org/10.1080/03772063.2024.2302104.
- Sacco, M. A., Pietroluongo, E., Husain, A. N., Arif, Q., Esposito, A., Di Lello, A., Garassino, M. C., Marino, M., & Dolezal, J. M. (2024). AB041. Deep learning discriminates thymic epithelial tumors histological subtypes using digital pathology. Mediastinum, 8, AB041. https://doi.org/10.21037/med-24-ab041.
- Lv, Q., Liang, K., Tian, C., Zhang, Y., Li, Y., Deng, J., Yue, W., & Li, W. (2024). Unveiling thymoma typing through hyperspectral imaging and deep learning. Journal of Biophotonics. https://doi.org/10.1002/jbio.202400325.
- Kussaibi, H. (2024). AI-enhanced subtyping of thymic tumors: Attention-based MIL with pathology-specific feature extraction. Preprint. https://doi.org/10.1101/2024.06.07.24308609.
- Zhou, H., Bai, H. X., Jiao, Z., Cui, B., Wu, J., Zheng, H., Yang, H., & Liao, W. (2023). Deep learning-based radiomic nomogram to predict risk categorization of thymic epithelial tumors: A multicenter study. European Journal of Radiology, 168, 111136. https://doi.org/10.1016/j.ejrad.2023.111136.
- Lay, N., Harmon, S., Latanski, J., Bingham, M., McAdams, M. J., Sirajuddin, A., Swift, S., Feierabend, C., Sansone, S., DaSilva, L., Rajan, A., Turkbey, B., & Zhao, C. (2023). AB011. Quantification of thymic epithelial tumor volumes through deep learning. Mediastinum. https://doi.org/10.21037/med-23-ab011.
- Liu, W., Wang, W., Guo, R., Zhang, H., & Guo, M. (2024). Deep learning for risk stratification of thymoma pathological subtypes based on pre-operative CT images. BMC Cancer, 24(1). https://doi.org/10.1186/s12885-024-12394-4.
- Yang, Y., Cheng, J., Chen, L., Cui, C., Liu, S., & Zuo, M. (2024). Application of machine learning for the differentiation of thymomas and thymic cysts using deep transfer learning: A multi-center comparison of diagnostic performance based on different dimensional models. Thoracic Cancer. https://doi.org/10.1111/1759-7714.15454.
- Dolezal, J. M., Guo, W., Bestvina, C., Vokes, E. E., Donington, J. S., Husain, A. N., & Garassino, M. C. (2022). AB012. Automated histologic subtyping of thymic epithelial tumors with deep learning. Mediastinum, 6, AB012. https://doi.org/10.21037/med-22-ab012.
- Amoah, L. E., Ji, X., Gad, S. E., & Zanello, F. (2022). Fully automatic quantitative measurement of 18F-FDG PET/CT in thymic epithelial tumors using a convolutional neural network. Clinical Nuclear Medicine, 47(7), 590–598. https://doi.org/10.1097/RLU.0000000000004146.
- Han, S. Y., Oh, J. S., Kim, Y., Seo, S. Y., Lee, G. D., Park, M., Choi, S., Kim, H. R., Kim, Y.-H., Kim, H. K., Park, S.-I., & Ryu, J.-S. (2022). Ful-ly automatic quantitative measurement of 18F-FDG PET/CT in thymic epithelial tumors using a convolutional neural network. Clinical Nuclear Medicine, 47, 590–598. https://doi.org/10.1097/RLU.0000000000004146.
- Yang, Y., Cheng, J., Peng, Z., Yi, L., Lin, Z., He, A., Jin, M., Cui, C., Liu, Y., Zhong, Q., & Zuo, M. (2023). Development and validation of con-trast-enhanced CT-based deep transfer learning and combined clinical-radiomics model to discriminate thymomas and thymic cysts: A multicenter study. Academic Radiology. https://doi.org/10.1016/j.acra.2023.10.018.
- Chang, C. C., Tang, E. K., Wei, Y.-F., Lin, C. Y., Wu, F. Z., Wu, M. T., Liu, Y.-S., Yen, Y.-T., Ma, M., & Tseng, Y. L. (2023). Clinical radiomics-based machine learning versus three-dimensional convolutional neural network analysis for differentiation of thymic epithelial tumors from other prevascular mediastinal tumors on chest computed tomography scan. Frontiers in Oncology, 13. https://doi.org/10.3389/fonc.2023.1105100.
- Nakajo, M., Harada Takeda, A., Katsuki, A., Jinguji, M., Ohmura, K., Tani, A., Sato, M., & Yoshida, T. (2022). The efficacy of 18F-FDG-PET-based radiomic and deep-learning features using a machine-learning approach to predict the pathological risk subtypes of thymic epithelial tumors. British Journal of Radiology, 95(1134). https://doi.org/10.1259/bjr.20211050.
- Onozato, Y., Suzuki, H., Matsumoto, H., Ito, T., Yamamoto, T., Tanaka, K., Sakairi, Y., Matsui, Y., Iwata, T., Iida, T., Iizasa, T., & Yoshino, I. (2023). Machine learning models from computed tomography to diagnose thymic epithelial tumors requiring combined resection. Journal of Thorac-ic Disease, 16, 4935–4946. https://doi.org/10.21037/jtd-23-1840.
- Tan, S., Chen, J., Wang, Z.-G., Zhao, X., Sun, Y., Zhu, Y., Gong, Z., & Zhang, D. (2022). Genomic characterization of thymic epithelial tumors reveals critical genes underlying tumorigenesis and poor prognosis. Clinical Genetics, 103(5), 529–539. https://doi.org/10.1111/cge.14285.
- Zhang, X., Zhang, P., Cong, A., Feng, Y., Chi, H., Xia, Z., & Tang, H. (2023). Unraveling molecular networks in thymic epithelial tumors: Deci-phering the unique signatures. Frontiers in Immunology, 14. https://doi.org/10.3389/fimmu.2023.1264325.
- Li, Z., Wang, F., Zhang, H., Zheng, H., Zhou, X., Wang, Z., Xie, S.-L., Peng, L., Wang, X., & Wang, Y. (2023). The predictive value of a comput-ed tomography-based radiomics model for the surgical separability of thymic epithelial tumors from the superior vena cava and the left innominate vein. Quantitative Imaging in Medicine and Surgery, 13, 5622–5640. https://doi.org/10.21037/qims-22-1050.
- Zhang, H., Chen, H., Qin, J., Wang, B., Ma, G., Wang, P., Zhong, D., & Liu, J. (2022). MC-ViT: Multi-path cross-scale vision transformer for thy-moma histopathology whole slide image typing. Frontiers in Oncology, 12. https://doi.org/10.3389/fonc.2022.925903.
- Caron, M., Misra, I., Mairal, J., Goyal, P., Bojanowski, P., & Joulin, A. (2020). Unsupervised learning of visual features by contrasting cluster as-signments. Advances in Neural Information Processing Systems, 33, 9912–9924.
- Chen, T., Kornblith, S., Norouzi, M., & Hinton, G. (2020). A simple framework for contrastive learning of visual representations. Proceedings of the 37th International Conference on Machine Learning, 119, 1597–1607.
- He, K., Fan, H., Wu, Y., Xie, S., & Girshick, R. (2020). Momentum contrast for unsupervised visual representation learning. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR 2020), 9729–9738. https://doi.org/10.1109/CVPR42600.2020.00975.
- Möhrmann, L., Rostock, L., Werner, M., Oleś, M., Arnold, J. S., Paramasivam, N., Jöhrens, K., et al. (2025). Genomic landscape and molecularly informed therapy in thymic carcinoma and other advanced thymic epithelial tumors. Med. https://doi.org/10.1016/j.medj.2025.100612.
- Grill, J.-B., Strub, F., Altché, F., Tallec, C., Richemond, P. H., Buchatskaya, E., Doersch, C., Pires, B. A., Guo, Z., Azar, M. G., Piot, B., Kavuk-cuoglu, K., Munos, R., & Valko, M. (2020). Bootstrap your own latent: A new approach to self-supervised learning. Advances in Neural Infor-mation Processing Systems, 33, 21271–21284.
Downloads
How to Cite
Received date: March 26, 2025
Accepted date: April 26, 2025
Published date: May 2, 2025