An Analysis of New Feature Extraction Methods Based on Machine Learning Methods for Classification Radiological...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Pulmonology

Background:

• COVID-19 primarily affects the lungs, causing inflammation and potentially leading to respiratory failure and multi-organ damage.
• Radiological pulmonary evaluation is critical for managing critically ill COVID-19 patients.
• Interpreting radiological images requires specialized expertise from radiologists.

Purpose of the Study:

• To investigate the efficacy of machine learning techniques for diagnosing COVID-19 using chest X-ray images.
• To identify subtle morphological differences in the lungs of COVID-19 patients that may not be apparent to the human eye.
• To compare the performance of various machine learning algorithms in COVID-19 detection from radiological data.

Main Methods:

• Utilized chest X-ray images from publicly available datasets containing COVID-19 positive cases.
• Extracted image features using the gray level co-occurrence matrix (GLCM) method.
• Applied several machine learning classifiers including K-nearest neighbor, support vector machine, linear discrimination analysis, naïve Bayes, and convolutional neural network (CNN).

Main Results:

• Machine learning methods demonstrated promising results in diagnosing COVID-19 from chest X-ray images.
• Convolutional neural networks (CNNs) exhibited superior performance compared to traditional machine learning approaches.
• AI-driven analysis of radiological images showed potential for reduced human involvement and enhanced diagnostic accuracy.

Conclusions:

• Deep learning techniques, specifically CNNs, are effective tools for automated COVID-19 detection via chest X-rays.
• AI can identify subtle radiological patterns indicative of COVID-19, complementing human expert interpretation.
• The findings suggest a significant role for artificial intelligence in improving the diagnostic workflow for respiratory infections like COVID-19.