Article Sidebar
Published:
Mar 24, 2026Keywords:
Article Details
Submission of a manuscript to this journal is a representation that the manuscript has not been published previously and is not under consideration for publication elsewhere.
All authors named in each manuscript would be required to sign a form (to be supplied by the Editor) so that they may retain their copyright in the article but to assign to us (the Publishers) and its licensees in perpetuity, in all forms, formats and media (whether known or created in the future) to (i) publish, reproduce, distribute, display and store the contribution, (ii) translate the contribution into other languages, create adaptations, reprints, include within collections and create summaries, extracts and/or abstracts of the contribution, (iii) create any other derivative works(s) based on the contribution, (iv) to exploit all subsidiary rights in the contribution, (v) the inclusion of electronic links from the contribution to third party material where-ever it may be located, and (vi) license any thrid party to do any or all of the above.
Main Article Content
Volumetric T-stage classification and segmentation of lung cancer with deep learning and machine learning in CT images: A 3D U-net-radiomics hybrid framework
Abstract
Purpose: To develop a 3D U-Net-Radiomics hybrid architecture of automated lung cancer segmentation and T-stage classification based on chest CT images.
Method: The 3D U-Net achieves high-fidelity voxel-level cancer segmentation, i.e., voxel-level volumetric mask, and radiomics features, such as shape, intensity, and texture features, are produced on the segmented regions. A Support Vector Machine (SVM) classifier was used to determine the cancer stages (T1-T4) by these features.
Results: The framework trained on the LIDC-IDRI dataset (1,018 cases) with a train/validation/test split of 70/10/20 indicates that the framework is more accurate in segmentation (97.4 %), has a Dice similarity score of 0.94, and is more accurate in T-stage classification (94.8 %), compared to baseline 2D CNN and handcrafted feature-based methods. The stability of the extracted radiomics descriptors was confirmed, and their reliability and reproducibility were validated using feature stability analysis.
Conclusion: The findings suggest that volumetric deep learning with radiomics is a high-performing and clinically interpretable solution to the automated staging of lung cancer, which helps in making better diagnostic and treatment decisions.
