Detection of COVID-19 using CXR and CT images using Transfer Learning and Haralick features

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Pulmonology

Background:

• Accurate COVID-19 detection from clinical images is challenging.
• COVID-19 shares similarities with other pulmonary diseases like pneumonia.
• Conventional classification methods require separate models for each disease.

Purpose of the Study:

• To apply transfer learning for COVID-19 detection using clinical images.
• To investigate the similarity between COVID-19 and viral pneumonia.
• To develop an efficient model for early COVID-19 prediction.

Main Methods:

• Transfer learning technique applied to pulmonary disease images.
• Haralick features used for texture feature extraction to focus on areas of interest.
• Statistical analyses employed for disease detection.

Main Results:

• Transfer learning demonstrated that COVID-19 is similar to viral pneumonia.
• The proposed model efficiently identified COVID-19, outperforming existing methods.
• Texture feature extraction using Haralick features improved detection accuracy.

Conclusions:

• Transfer learning eliminates the need for separate COVID-19 classification models.
• The proposed model accelerates prediction, aiding medical professionals and reducing virus spread.
• Early prediction through efficient models can help save lives.