Automated brain tumor segmentation with deep learning
DOI:
https://doi.org/10.52292/j.laar.2025.3614Keywords:
3D U-Net, Attention U-Net, Brats 2018, Accuracy.Abstract
Accurate tumor delineation is crucial for effective diagnosis, treatment planning and risk factor identification. This study presents an advanced Deep Learning (DL) framework designed for the precise segmentation of brain tumors and reliable survival prediction for tumor patients. In this work, a cutting-edge approach that leverages an ensemble strategy, combining two distinct 3D UNet architectures (3D U-Net and Attention 3D U-Net) is incorporated for segmentation purpose. This ensemble approach employs a majority rule mechanism and ensures a more reliable and comprehensive delineation of tumor regions. The proposed system's effectiveness and performance are evaluated using BraTS 2018 dataset. The proposed deep learning network trained and validated for brain tumor segmentation achieved promising results on the online final dataset, as evidenced by the Dice coefficient and Hausdorff metric scores. Specifically, the performance metrics for different tumor regions were as follows: Enhancing Tumor (ET), 0.805 Dice Score and Hausdorff Distance about 2.779. For Tumor Core (TC), its about 0.851 and 6.378 and for Necoratic core 0.904 and 6.323.
Published
Issue
Section
License
Copyright (c) 2025 Latin American Applied Research - An international journal

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
Once a paper is accepted for publication, the author is assumed to have transferred its copyright to the Publisher. The Publisher will not, however, put any limitation on the personal freedom of the author to use material from the paper in other publications. From September 2019 it is required that authors explicitly sign a copyright release form before their paper gets published. The Author Copyright Release form can be found here