Developing a Deep Neural Network model for COVID-19 diagnosis based on CT scan images

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computer Science

Background:

• Chest X-rays and computed tomography (CT) scans show potential in diagnosing COVID-19.
• Current artificial intelligence (AI) CT diagnosis systems face challenges with complex neural networks, leading to training difficulties and high computation rates.

Purpose of the Study:

• To develop a lightweight 6-layer Deep Neural Network (DNN) model for improved COVID-19 diagnosis using CT scan images.
• To enhance the accuracy of classifying individuals with and without COVID-19.

Main Methods:

• A 6-layer Deep Neural Network (DNN) model was developed for COVID-19 diagnosis from CT images.
• A global feature extractor operator was employed for image feature extraction.
• The 10-fold cross-validation technique was used for data partitioning (training, testing, validation).
• The DNN model was trained without neuron dropout.

Main Results:

• The proposed lightweight DNN model achieved a diagnostic accuracy of 96.71%.
• The DNN model demonstrated superior performance compared to other investigated classification models (decision trees, random forests, standard neural networks).

Conclusions:

• The developed 6-layer DNN model offers a highly accurate and efficient method for COVID-19 diagnosis using CT scans.
• This AI-driven approach provides a promising alternative or supplement to traditional COVID-19 testing methods.