Hybrid Discrete Wavelet Transform and Gabor Filter Banks Processing for Features Extraction from Biomedical Images

Area of Science:

• Biomedical Imaging
• Machine Learning
• Image Analysis

Background:

• Accurate feature extraction and classification are crucial for diagnosing diseases from biomedical images.
• Existing methods often struggle with complex textural features in medical scans.
• Developing robust automated systems is essential for improving diagnostic efficiency.

Purpose of the Study:

• To present a new methodology for automatic feature extraction and classification of biomedical images.
• To exploit the spatial orientation of high-frequency textural features for improved classification accuracy.
• To validate the proposed approach on diverse medical imaging datasets.

Main Methods:

• A two-step process involving the two-dimensional discrete wavelet transform (DWT) and a Gabor filter bank.
• Extraction of high-frequency subband images and subsequent Gabor filtering at various orientations.
• Computation of image entropy and uniformity statistics, fed into a support vector machine (SVM) classifier.

Main Results:

• The methodology demonstrated superior classification performance compared to methods using only DWT or Gabor filters.
• Validation on mammograms, retina, and brain MR images confirmed the approach's effectiveness.
• The combined feature extraction technique yielded higher classification accuracies.

Conclusions:

• The proposed automated feature extraction and classification method offers enhanced performance for biomedical image analysis.
• This approach shows significant potential for improving diagnostic accuracy in medical imaging applications.
• The integration of DWT and Gabor filters provides a powerful tool for analyzing complex image textures.