https://doi.org/10.1140/epjs/s11734-025-01842-2
Regular Article
Cascading GLCM and T-SNE for detecting tumor on kidney CT images with lightweight machine learning design
1
Faculty of Engineering, Electrical-Electronics Engineering, Cankiri Karatekin University, Cankiri, Türkiye
2
Faculty of Engineering, Computer Engineering, Cankiri Karatekin University, Cankiri, Türkiye
a tahaetem@hotmail.com, tahaetem@karatekin.edu.tr
Received:
29
May
2025
Accepted:
30
July
2025
Published online:
7
August
2025
This study presents a novel machine learning framework that combines Gray-Level Co-occurrence Matrix (GLCM) and T-Distributed Stochastic Neighbor Embedding (T-SNE) for feature extraction and dimensionality reduction in kidney tumor detection. The proposed method aims to achieve high accuracy while reducing computational overhead, making it suitable for real-time applications in resource-constrained settings. GLCM is employed to extract texture-based features from CT images, which are then reduced to lower dimensions using T-SNE. This approach not only improves classification accuracy but also enhances model interpretability by preserving essential features in a reduced-dimensional space. The model is trained and tested on two large CT datasets. Various classifiers, Fine KNN, Bagged Trees, Fine Tree, Tri-layered Neural Network, and Fine Gaussian SVM, are evaluated. Results indicate that the combination of GLCM and T-SNE achieved state-of-the-art performance, with Fine KNN reaching nearly perfect accuracy 99.98%. Moreover, Hybrid feature extraction and dimensionality reduction methods show improved prediction speed and reduced model size, demonstrating the effectiveness of dimensionality reduction. Future directions include the integration of different models, expanding the dataset for broader applicability, and improving model explainability to enhance clinical-level applications.
Copyright comment Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
© The Author(s), under exclusive licence to EDP Sciences, Springer-Verlag GmbH Germany, part of Springer Nature 2025
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
