Related Experiment Video
Updated: Aug 27, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.9K
Application of Genetic Algorithm and U-Net in Brain Tumor Segmentation and Classification: A Deep Learning Approach
Muhammad Arif1, Anupama Jims2, Ajesh F3
1Department of Computer Science, Superior University, Lahore, Pakistan.
Computational Intelligence and Neuroscience
|September 26, 2022
Summary
This study introduces a deep learning model for accurate brain tumor detection using MRI scans. The proposed GA-UNET method significantly improves detection accuracy and performance over traditional techniques.
Area of Science:
- Neurology
- Medical Imaging
- Artificial Intelligence
Background:
- Brain cancer, characterized by abnormal cell growth in the cerebrum, is classified into benign and malignant types.
- Early detection is crucial but hindered by limitations in traditional tumor detection methods, including low effectiveness and image processing issues.
- Existing methods suffer from small datasets, poor predictive capabilities, and overlooked crucial stages, leading to inaccurate tumor detection.
Purpose of the Study:
- To develop an effective deep learning technique for accurate brain tumor detection.
- To address the limitations of traditional methods in terms of effectiveness, image quality, dataset size, and predictive accuracy.
- To enhance the early detection of brain cancer through advanced computational approaches.
Main Methods:
- Utilized the REMBRANDT dataset with multisequence MRI scans from 130 patients.
- Employed preprocessing techniques including greyscale conversion, skull stripping, and histogram equalization.
- Implemented a pipeline involving genetic algorithm for segmentation, discrete wavelet transform (DWT) for feature extraction, particle swarm optimization for feature selection, and U-Net for classification.
Main Results:
- The proposed GA-UNET model achieved high performance metrics: 0.97 accuracy, 0.98 sensitivity, and 0.98 specificity.
- The model demonstrated superior performance compared to other advanced models in brain tumor detection.
- The integrated approach effectively overcame limitations associated with traditional detection methods.
Conclusions:
- The developed GA-UNET deep learning model offers a highly effective solution for brain tumor detection.
- This technique significantly enhances accuracy, sensitivity, and specificity in identifying cancerous growths.
- The study highlights the potential of advanced deep learning for improving early cancer diagnosis and patient outcomes.
