Related Experiment Video
Updated: Oct 25, 2025
08:05
Lung CT Segmentation to Identify Consolidations and Ground Glass Areas for Quantitative Assesment of SARS-CoV Pneumonia
Published on: December 19, 2020
14.4K
Wavelet and deep learning-based detection of SARS-nCoV from thoracic X-ray images for rapid and efficient testing
Amar Kumar Verma1, Inturi Vamsi2, Prerna Saurabh3
1Department of Electrical and Electronics, Birla Institute of Technology and Science-Pilani, Hyderabad Campus, 500078, India.
Expert Systems with Applications
|August 9, 2021
Summary
This study introduces a rapid COVID-19 detection method using wavelet and deep learning on X-rays, achieving up to 98.87% accuracy. This AI approach offers a faster alternative to traditional genetic testing for Severe Acute Respiratory Coronavirus Syndrome (SARS-nCoV).
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Signal Processing
Background:
- Current Severe Acute Respiratory Coronavirus Syndrome (SARS-nCoV) diagnosis relies on time-consuming real-time Reverse Transcriptase-Polymerase Chain Reaction (rRT-PCR).
- There is a need for rapid and efficient diagnostic tools to manage the pandemic effectively.
Purpose of the Study:
- To develop and validate a wavelet and deep learning-enabled procedure for fast and accurate COVID-19 detection using thoracic X-ray images.
- To compare the performance of different wavelet families in feature extraction for COVID-19 diagnosis.
Main Methods:
- Utilized a deep learning approach with Convolutional Neural Network (CNN) architectures pre-trained on ImageNet.
- Employed Discrete Wavelet Transforms (DWT) with various mother wavelets (Haar, Daubechies, Symlet, Biorthogonal, Coiflet, Discrete Meyer) for two-level decomposition.
- Extracted features like ground-glass opacities from thoracic X-rays to enhance signal-to-noise ratio.
- Validated models using k-Fold Cross Validation (k-Fold CV) on two-class (COVID vs. Normal) and four-class (COVID-19 vs. PNA bacterial vs. PNA viral vs. Normal) datasets.
Main Results:
- The Symlet 7 approximation component achieved the highest test accuracy of 98.87%, followed by Biorthogonal 2.6 at 98.73%.
- Haar and Daubechies wavelets demonstrated excellent validation accuracy on unseen data.
- For the four-class classification, the proposed algorithm achieved 98% precision, 98% recall, and 99% Dice similarity coefficient.
Conclusions:
- Wavelet-based deep learning models show high accuracy and efficiency for detecting COVID-19 from thoracic X-rays.
- The proposed method offers a practical and feasible real-time diagnostic solution for Severe Acute Respiratory Coronavirus Syndrome (SARS-nCoV).
- This AI-driven approach can significantly reduce diagnostic time compared to traditional rRT-PCR methods.
