Image Analysis for MRI Based Brain Tumor Detection and Feature Extraction Using Biologically Inspired BWT and SVM.

Area of Science:

• Medical Imaging
• Computer-Aided Diagnosis
• Biomedical Engineering

Background:

• Manual segmentation of brain tumors from MRI is time-consuming and subjective.
• Radiologist experience heavily influences the accuracy of tumor detection and extraction.
• Computer-aided technologies are crucial for overcoming limitations in manual medical image analysis.

Purpose of the Study:

• To develop an efficient and accurate automated brain tumor segmentation method using medical imaging.
• To reduce the complexity and time involved in segmenting tumorous regions in MR images.
• To enhance the performance of classification algorithms for improved diagnostic accuracy.

Main Methods:

• Investigated Berkeley Wavelet Transform (BWT) for brain tumor segmentation in MR images.
• Employed Support Vector Machine (SVM) classifier with feature extraction for improved accuracy.
• Validated performance using metrics: accuracy, sensitivity, specificity, and Dice Similarity Index (DSI).

Main Results:

• Achieved 96.51% accuracy, 97.72% sensitivity, and 94.2% specificity in identifying normal and abnormal tissues.
• Obtained an average Dice Similarity Index coefficient of 0.82, indicating strong overlap with manual segmentation.
• Demonstrated superior performance in quality parameters and accuracy compared to existing state-of-the-art techniques.

Conclusions:

• The proposed BWT-based method offers a highly effective and accurate solution for automated brain tumor segmentation.
• This approach significantly aids in distinguishing between normal and abnormal tissues in MR images.
• The technique shows promise for clinical application, improving diagnostic efficiency and reliability.