Texture-Based Classification to Overcome Uncertainty between COVID-19 and Viral Pneumonia Using Machine Learning and...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Biology

Background:

• COVID-19 symptoms mimic viral pneumonia, complicating early diagnosis.
• Accurate classification of COVID-19 is challenging due to high-dimensional image data and limitations of previous studies.
• Young individuals, the elderly, and immunocompromised populations are particularly vulnerable.

Purpose of the Study:

• To develop a novel classification model for accurate chest X-ray image analysis.
• To differentiate between normal, COVID-19, and viral pneumonia cases.
• To overcome limitations of previous studies using simplistic algorithms and small datasets.

Main Methods:

• Integration of advanced texture feature extraction methods: Gray-Level Co-occurrence Matrix (GLCM), Gray-Level Dependence Matrix (GLDM), and wavelet transform.
• Application within a deep learning framework for image classification.
• Leveraging unique texture characteristics inherent to each dataset class.

Main Results:

• Superior classification performance compared to traditional methods.
• High accuracy (DLNN: 0.92), recall (DLNN: 0.93), precision (DLNN: 0.87), and F1-Score (DLNN: 0.89) achieved by the deep learning neural network (DLNN).
• Demonstrated effectiveness even with complex and diverse image data.

Conclusions:

• The proposed model represents a significant advancement in AI-based diagnostic systems for COVID-19 and pneumonia.
• The approach offers improved patient outcomes and healthcare management strategies.
• Advanced texture analysis combined with deep learning enhances diagnostic capabilities for respiratory illnesses.