Automatic method for classifying COVID-19 patients based on chest X-ray images, using deep features and PSO-optimized...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Biology

Background:

• The COVID-19 pandemic poses a significant global health challenge.
• Screening infected patients via chest X-ray (CXR) is crucial but time-consuming and variable.
• Automated techniques are needed to assist specialists in CXR-based COVID-19 diagnosis.

Purpose of the Study:

• To develop and evaluate an automated computational method for identifying COVID-19 patients from chest radiographs.
• To improve the efficiency and consistency of COVID-19 screening using medical imaging.

Main Methods:

• A dataset of chest X-ray images was acquired from public databases.
• Image preprocessing and standardization were performed.
• Deep features were extracted using VGG19, Inception-v3, and ResNet50 networks.
• Classification was conducted using eXtreme Gradient Boosting (XGBoost) optimized by particle swarm optimization (PSO).

Main Results:

• The proposed method achieved high performance metrics: 98.71% accuracy, 98.89% precision, 99.63% recall, and 99.25% F1-score.
• The combination of deep features and XGBoost optimized by PSO proved effective for COVID-19 classification.
• The automated approach demonstrated efficiency in distinguishing COVID-19 from non-COVID-19 cases.

Conclusions:

• Automated classification of CXR images for COVID-19 detection is feasible and efficient.
• The proposed deep learning and XGBoost-PSO method offers a valuable tool for clinicians.
• This technique can significantly aid in combating the COVID-19 pandemic through rapid and reliable screening.