Multiresolution analysis for COVID-19 diagnosis from chest CT images: wavelet vs. contourlet transforms.

Area of Science:

• Medical Imaging
• Computer Vision
• Signal Processing

Background:

• Chest computer tomography (CT) is crucial for COVID-19 diagnosis.
• Wavelet and contourlet transforms offer localized, multiresolution image analysis.
• Feature extraction from CT images is key for accurate disease classification.

Purpose of the Study:

• To investigate transform-based features for COVID-19 classification using chest CT images.
• To determine the optimal transform and decomposition level for feature extraction.
• To develop an efficient and accurate COVID-19 detection method.

Main Methods:

• Applied multiresolution analysis using wavelet and contourlet transforms at different decomposition levels.
• Extracted textural and statistical features from subbands.
• Utilized particle swarm optimization (PSO) for feature selection.
• Employed a support vector machine (SVM) classifier for classification.

Main Results:

• Contourlet features from the first decomposition level (L1) yielded the most reliable classification.
• Feature vector computation was rapid (<25 ms for 256x256 images).
• Achieved 100% accuracy, sensitivity, specificity, precision, and F-score with the optimized feature set.
• Outperformed several deep learning approaches in COVID-19 detection.

Conclusions:

• Transform-based features are reliable for COVID-19 detection from chest CT scans.
• The contourlet transform method offers reduced computational complexity.
• This approach is suitable for real-time automatic screening systems for initial COVID-19 detection.