Run length encoding based wavelet features for COVID-19 detection in X-rays.

Area of Science:

• Medical Imaging Analysis
• Machine Learning in Healthcare
• Radiology

Background:

• Accurate and efficient detection of COVID-19 from chest X-rays is crucial for timely diagnosis and patient management.
• Traditional methods may face challenges in feature extraction and classification accuracy.

Purpose of the Study:

• To develop a novel approach for the recognition of COVID-19 cases using chest X-ray images.
• To leverage the power of discrete Wavelet transform (DWT) for feature extraction and support vector machine (SVM) for classification.

Main Methods:

• Chest X-ray images were decomposed using DWT to obtain approximation coefficients.
• A novel coefficient selection scheme involving hard thresholding and run-length encoding was employed to extract high-magnitude coefficients.
• Feature vectors were generated, unified in length via zero-padding, and classified using an SVM.

Main Results:

• The proposed system demonstrated promising classification accuracy for identifying COVID-19 cases.
• The DWT effectively produced a small set of highly discriminative features from the X-ray images.

Conclusions:

• The DWT-based feature extraction method is effective in generating discriminative features for COVID-19 recognition.
• This approach reduces feature space dimensionality, leading to decreased training data requirements and lower computational complexity.