X-ray image classification using random forests with local wavelet-based CS-local binary patterns.

Area of Science:

• Medical Imaging
• Computer Vision
• Machine Learning

Background:

• Traditional Local Binary Patterns (LBP) methods for image analysis often rely on raw pixel data.
• Enhancing texture descriptors is crucial for improving the accuracy of medical image classification tasks.
• Existing classification methods may face limitations in terms of speed and performance for complex datasets.

Purpose of the Study:

• To develop a novel and efficient method for classifying X-ray images.
• To improve the performance and reduce the computational time of X-ray image classification.
• To introduce a new feature descriptor that captures wavelet-based texture characteristics.

Main Methods:

• Extraction of local wavelet-based centre symmetric LBP (WCS-LBP) descriptors from image regions.
• Utilizing WCS-LBP features to construct random forests, an ensemble of decision trees.
• Classification of X-ray images based on the maximum posterior probability derived from the random forest model.

Main Results:

• The proposed WCS-LBP method demonstrated superior performance compared to other feature descriptors and classifiers.
• The random forest classifier combined with WCS-LBP achieved faster processing times for X-ray image classification.
• Significant improvements in classification accuracy were observed using the novel wavelet-based texture descriptors.

Conclusions:

• The integration of wavelet-based texture analysis with LBP offers a powerful approach for X-ray image classification.
• The proposed method provides a computationally efficient and accurate solution for medical image analysis.
• This technique holds potential for advancing automated diagnostic systems in radiology.