Deep learning-integrated MRI brain tumor analysis: feature extraction, segmentation, and Survival Prediction using
Deependra Rastogi1, Prashant Johri2, Massimo Donelli3,4
1School of Computer Science and Engineering, IILM University, Greater Noida, Noida, 201306, UP, India.
Scientific Reports
|January 9, 2025
Summary
This study introduces a deep learning model for precise brain glioma segmentation and survival prediction using MRI scans. The approach enhances diagnostic accuracy and treatment planning for brain tumor patients.
Area of Science:
- Neuro-oncology
- Medical Imaging
- Artificial Intelligence
Background:
- Gliomas are the most common malignant brain tumors.
- Accurate tumor segmentation and survival prediction are critical for diagnosis and treatment.
- Current methods require improvement in precision and resilience.
Purpose of the Study:
- To develop a deep learning approach for accurate brain tumor segmentation and survival prediction in glioma patients.
- To improve diagnostic and prognostic capabilities for brain tumors.
- To leverage MRI data for enhanced patient outcomes.
Main Methods:
- Utilized 2D volumetric convolution neural networks with a majority rule for robust brain tumor segmentation.
- Employed a Deep Learning Inspired 3D replicator neural network for radiomic feature selection.
- Integrated MRI scans for comprehensive tumor analysis.
Main Results:
- The model achieved successful segmentation of brain tumors, including enhancing and non-enhancing tumor regions.
- Accurate prediction of patient survival outcomes was demonstrated.
- Evaluation on the BRATS2020 dataset yielded satisfactory and promising results.
Conclusions:
- The proposed deep learning model offers a promising solution for glioma segmentation and survival prediction.
- This approach can aid clinicians in diagnosis, treatment planning, and risk factor identification.
- Further validation may enhance its clinical applicability in neuro-oncology.
