COVID-19 Detection System Using Chest CT Images and Multiple Kernels-Extreme Learning Machine Based on Deep Neural...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Biology

Background:

• Coronavirus disease (COVID-19) is a fatal epidemic originating in Wuhan, China.
• Diagnosis relies on radiological images and RT-PCR tests.
• Automated analysis of chest CT images is crucial for combating infectious diseases.

Purpose of the Study:

• To propose a novel Multiple Kernels-ELM-based Deep Neural Network (MKs-ELM-DNN) method for COVID-19 detection.
• To leverage deep learning for automated analysis of chest CT scans.
• To improve the accuracy and efficiency of COVID-19 diagnosis.

Main Methods:

• Utilized a pre-trained DenseNet201 Convolutional Neural Network (CNN) for deep feature extraction from CT images.
• Employed Extreme Learning Machine (ELM) classifiers with various activation functions (ReLU-ELM, PReLU-ELM, TanhReLU-ELM).
• Implemented a majority voting method for final class label prediction.

Main Results:

• The MKs-ELM-DNN model achieved an accuracy score of 98.36% on a public dataset.
• The proposed model demonstrated superior performance compared to state-of-the-art algorithms.
• Experimental validation confirmed the model's effectiveness in identifying COVID-19 cases.

Conclusions:

• The MKs-ELM-DNN model offers an effective approach for COVID-19 identification.
• Automated CT image analysis using this deep learning model can aid in disease control.
• This method shows significant potential in combating the COVID-19 pandemic.