Brain Tumor Detection Using 3D U-Net Segmentation Features and a Hybrid Machine Learning Classifier

• Author(s): Nindujerla Sandeep Rao; Kota Sai Vidith; Korabandi Ajay Babu; Dr. N. Sudheer Kumar
• Paper ID: 1717163
• Page: 3822-3825
• Published Date: 03-05-2026
• Published In: Iconic Research And Engineering Journals
• Publisher: IRE Journals
• e-ISSN: 2456-8880
• Volume/Issue: Volume 9 Issue 10 April-2026
• DOI: https://doi.org/10.64388/IREV9I10-1717163

Abstract

Brain tumors killed roughly 250,000 people worldwide in 2020, with approximately 300,000 new diagnoses that same year. The most aggressive type, Glioblastoma Multiforme (GBM), carries a median survival of just 15 months and a five-year survival rate under 5%. A key clinical complication is the MGMT gene promoter: its methylation status determines how well a patient respond to chemotherapy, but establishing it currently requires an invasive biopsy. This paper proposes a non-invasive framework combining 3D U-Net MRI segmentation with a soft-vote hybrid classifier (KNN + GBC). On the RSNA-MICCAI dataset (585 samples), 3D U-Net segmentation yields 111 volumetric radiomic features versus 54 from a 2D U-Net. The hybrid model achieves 99.4% classification accuracy on the richer 3D features, well above deep learning baselines (39–49%) tested on the same data, confirming that feature quality and ensemble diversity outperform model complexity on small clinical datasets.

Keywords

brain tumor detection, 3D U-Net segmentation, radiomic features, hybrid ensemble learning, gradient boosting, MGMT prediction

Citations

IRE Journals:
Nindujerla Sandeep Rao, Kota Sai Vidith, Korabandi Ajay Babu, Dr. N. Sudheer Kumar "Brain Tumor Detection Using 3D U-Net Segmentation Features and a Hybrid Machine Learning Classifier" Iconic Research And Engineering Journals Volume 9 Issue 10 2026 Page 3822-3825 https://doi.org/10.64388/IREV9I10-1717163

IEEE:
Nindujerla Sandeep Rao, Kota Sai Vidith, Korabandi Ajay Babu, Dr. N. Sudheer Kumar "Brain Tumor Detection Using 3D U-Net Segmentation Features and a Hybrid Machine Learning Classifier" Iconic Research And Engineering Journals, vol. 9, no. 10, Apr. 2026, doi: https://doi.org/10.64388/IREV9I10-1717163

APA:
Nindujerla Sandeep Rao, Kota Sai Vidith, Korabandi Ajay Babu, Dr. N. Sudheer Kumar (2026). Brain Tumor Detection Using 3D U-Net Segmentation Features and a Hybrid Machine Learning Classifier. Iconic Research And Engineering Journals, 9(10). doi: https://doi.org/10.64388/IREV9I10-1717163

MLA:
Nindujerla Sandeep Rao, Kota Sai Vidith, Korabandi Ajay Babu, Dr. N. Sudheer Kumar "Brain Tumor Detection Using 3D U-Net Segmentation Features and a Hybrid Machine Learning Classifier" Iconic Research And Engineering Journals, vol. 9, no. 10, Apr. 2026. Crossref, https://doi.org/10.64388/IREV9I10-1717163

BibTeX

@article{1717163,
author = {Nindujerla Sandeep Rao, Kota Sai Vidith, Korabandi Ajay Babu, Dr. N. Sudheer Kumar},
title = {Brain Tumor Detection Using 3D U-Net Segmentation Features and a Hybrid Machine Learning Classifier},
journal = {Iconic Research And Engineering Journals},
year = {2026},
volume = {9},
number = {10},
pages = {3822-3825},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1717163.pdf},
abstract = {Brain tumors killed roughly 250,000 people worldwide in 2020, with approximately 300,000 new diagnoses that same year. The most aggressive type, Glioblastoma Multiforme (GBM), carries a median survival of just 15 months and a five-year survival rate under 5%. A key clinical complication is the MGMT gene promoter: its methylation status determines how well a patient respond to chemotherapy, but establishing it currently requires an invasive biopsy. This paper proposes a non-invasive framework combining 3D U-Net MRI segmentation with a soft-vote hybrid classifier (KNN + GBC). On the RSNA-MICCAI dataset (585 samples), 3D U-Net segmentation yields 111 volumetric radiomic features versus 54 from a 2D U-Net. The hybrid model achieves 99.4% classification accuracy on the richer 3D features, well above deep learning baselines (39–49%) tested on the same data, confirming that feature quality and ensemble diversity outperform model complexity on small clinical datasets.},
keywords = {brain tumor detection, 3D U-Net segmentation, radiomic features, hybrid ensemble learning, gradient boosting, MGMT prediction},
month = {April}
}