COVID-19 detection based on pre-trained deep networks and LSTM model using X-ray images enhanced contrast with...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Infectious Diseases

Background:

• Coronavirus disease (COVID-19) is a global pandemic with significant mortality.
• Current diagnostic methods for COVID-19 can be time-consuming.
• Biomedical imaging presents a potential avenue for rapid COVID-19 detection.

Purpose of the Study:

• To develop an automated and efficient diagnostic tool for COVID-19 detection.
• To classify COVID-19 cases from chest X-rays using deep learning and image enhancement.
• To improve the accuracy and speed of COVID-19 diagnosis.

Main Methods:

• A hybrid model combining pre-trained deep networks and Long Short-Term Memory (LSTM) was proposed.
• Contrast-enhanced chest X-rays were utilized for feature extraction.
• The Artificial Bee Colony (ABC) algorithm was employed for image transformation optimization.
• Softmax classification was used to differentiate between COVID-19, normal, and pneumonia cases.

Main Results:

• The proposed model achieved high performance metrics on the "COVID-19 radiography" dataset.
• Accuracy reached 98.97%, precision 98.80%, and sensitivity 98.70%.
• ABC-based image enhancement led to a 2.5% increase in classification performance compared to other methods.

Conclusions:

• The proposed hybrid deep learning model demonstrates high efficiency and accuracy for COVID-19 classification from chest X-rays.
• The integration of ABC algorithm-based image enhancement significantly improved diagnostic performance.
• This approach offers a promising automated solution for faster and more reliable COVID-19 diagnosis.