Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{185449,
        author = {Sanapala Anusha and Prof. K. Venkata Rao},
        title = {3D Deep Learning-Based Brain Tumor Segmentation, Glioma Classification, and Survival Prediction},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {12},
        number = {5},
        pages = {1668-1675},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=185449},
        abstract = {This paper  proposes an integrated 3D deep learning framework for brain tumor analysis using the BraTS 2020 dataset. This begins with preprocessing of multimodal MRI scans (T1, T1CE, T2, and FLAIR), including skull stripping, Z-score normalization, and image resizing. Tumor segmentation was performed using a custom 3D U-Net model trained to predict whole-tumor (WT), tumor core (TC), and enhancing tumor (ET) regions. For tumor-type classification (HGG vs. LGG), a pretrained 3D DenseNet121 model was employed with stratified K-fold cross-validation. Segmentation masks were used to extract the volumetric features of WT, TC, and ET. These tumor features, combined with age and MRI inputs, were used for survival prediction using 3D DenseNet regression analysis. The evaluation included the Dice score for segmentation, accuracy and AUC for classification, and MAE, RMSE, and R² for survival analysis. Results showed improved performance when the tumor features were integrated. This paper  demonstrates a robust multitask pipeline capable of capturing tumor heterogeneity and enhancing predictive accuracy. The framework offers an automated and reproducible approach that can reduce inter-observer variability and assist clinical decision-making. Its modular design allows for future integration of additional clinical or genomic data, paving the way for personalized neuro-oncology applications.},
        keywords = {3D U-Net, BraTS 2020, Brain Tumor, Classification, DenseNet121, HGG, LGG, Segmentation, Stratified K-fold Cross Validation, Survival Prediction},
        month = {October},
        }