Distinctive approach in brain tumor detection and feature extraction using biologically inspired DWT method and SVM

Area of Science:

• Medical Imaging Analysis
• Computational Biology
• Machine Learning in Medicine

Background:

• Brain tumor segmentation and classification are critical for treatment but remain challenging due to tumor heterogeneity.
• Existing methods often struggle with the diverse characteristics of brain tumors, impacting diagnostic accuracy.

Purpose of the Study:

• To develop and evaluate an enhanced method for brain tumor segmentation and classification.
• To improve the accuracy and efficiency of medical image analysis for brain tumors.

Main Methods:

• Utilized Discrete Wavelet Transform (DWT) and Principal Component Analysis (PCA) for feature enhancement and selection.
• Employed Otsu's method for initial segmentation, followed by PCA for dimensionality reduction.
• Extracted texture features using the grey-level co-occurrence matrix (GLCM).
• Classified tumors using a Support Vector Machine (SVM) with various kernels.

Main Results:

• Achieved high performance metrics: 86.9% recall, 95.2% precision, and 90.9% F-measure.
• Demonstrated a Dice Similarity Index coefficient of 0.82, indicating strong agreement between automated and manual segmentation.
• The proposed method showed improved quality and accuracy over existing techniques.

Conclusions:

• The combined DWT-PCA approach with SVM classification offers a robust and accurate solution for brain tumor segmentation and classification.
• This method enhances the reliability of medical image analysis, potentially improving patient outcomes.