Investigating the Role of Image Fusion in Brain Tumor Classification Models Based on Machine Learning Algorithm for
R Nanmaran1, S Srimathi2, G Yamuna2
1Department of Biomedical Engineering, Saveetha School of Engineering, Saveetha Institute of Medical and Technical Sciences, Saveetha University, Chennai, 602105 Tamil Nadu, India.
Computational and Mathematical Methods in Medicine
|February 18, 2022
Summary
Image fusion using frequency domain methods enhances brain tumor classification. This AI-driven approach improves diagnostic accuracy for personalized medicine, outperforming traditional methods.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computational Biology
Background:
- Image fusion techniques are crucial for enhancing medical image quality and information content.
- Artificial intelligence (AI) algorithms are increasingly vital for improving healthcare outcomes and enabling personalized medicine.
- Accurate brain tumor classification is essential for effective patient treatment and personalized medicine strategies.
Purpose of the Study:
- To investigate the role of image fusion in improving brain tumor classification models.
- To develop a novel, fusion-based cancer classification model for more effective personalized medicine.
- To enhance the performance of AI classifiers by utilizing high-quality fused medical images.
Main Methods:
- Preprocessing of medical images (MRI, SPECT) using Contrast Limited Adaptive Histogram Equalization.
- Application of a Discrete Cosine Transform-based image fusion method to combine benign and malignant brain tumor images.
- Evaluation of AI classifiers (Support Vector Machine, KNN, Decision Tree) using features extracted from fused images.
Main Results:
- The fusion-based approach significantly improved classification results compared to using individual input images.
- The Support Vector Machine (SVM) classifier achieved a maximum accuracy of 96.8% with fused image features.
- SVM classifier demonstrated superior performance over KNN and Decision Tree classifiers in brain tumor classification.
Conclusions:
- Image fusion, particularly using frequency domain methods, enhances the quality and informativeness of medical images for classification tasks.
- The proposed fusion-based cancer classification model shows significant potential for improving personalized medicine through more accurate brain tumor diagnosis.
- AI algorithms integrated with image fusion offer a promising avenue for advancing diagnostic capabilities in healthcare.
