Multi-texture features and optimized DeepNet for COVID-19 detection using chest x-ray images

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Biology

Background:

• The COVID-19 pandemic necessitates rapid and accurate diagnostic tools.
• Radiological imaging, particularly chest X-rays (CXRs), plays a crucial role in diagnosing COVID-19.
• Existing diagnostic methods require optimization for early and reliable detection.

Purpose of the Study:

• To propose a novel and effective method for early COVID-19 detection using chest X-ray images.
• To develop an optimized deep learning model integrated with advanced feature extraction techniques.
• To enhance the accuracy and reliability of AI-based COVID-19 diagnosis.

Main Methods:

• A multi-local texture features (MLTF) approach was developed for robust feature extraction.
• An Improved Weed Sea-based DeepNet (IWS-based DeepNet) was designed, optimizing Deep Convolutional Neural Network (Deep CNN) structure.
• The IWS algorithm combines Improved Invasive Weed Optimization (IIWO) and Sea Lion Optimization (SLnO).
• Region of Interest (RoI) extraction using adaptive thresholding was employed to reduce noise in CXR images.

Main Results:

• The proposed MLTF and IWS-based DeepNet achieved high performance metrics.
• The system demonstrated a True Positive Rate (TPR) of 0.933%.
• A True Negative Rate (TNR) of 0.890% and an overall accuracy of 0.919% were recorded.

Conclusions:

• The developed IWS-based DeepNet with MLTF offers a promising approach for early and accurate COVID-19 detection from CXR images.
• This AI-driven method can significantly aid radiologists and clinicians in diagnosing COVID-19.
• The optimization techniques employed enhance the potential of deep learning models in medical diagnostics.