A Survey on Deep Learning in COVID-19 Diagnosis

Area of Science:

• Artificial Intelligence
• Medical Imaging
• Machine Learning

Background:

• The COVID-19 pandemic has caused widespread global health and economic disruption.
• Accurate and rapid diagnosis of COVID-19 is crucial for controlling its spread and mitigating harm.
• Artificial intelligence (AI), particularly deep learning, offers promising avenues for medical image analysis in disease diagnosis.

Purpose of the Study:

• To review and introduce the latest deep learning methods and techniques for diagnosing COVID-19 using chest X-ray and CT images.
• To analyze the performance of Convolutional Neural Network (CNN) based systems for COVID-19 diagnosis.
• To discuss the potential for improving CNN performance in medical image analysis for infectious diseases.

Main Methods:

• Review of deep learning techniques applied to chest X-ray and CT image classification for COVID-19.
• Detailed explanation of CNN architectures including AlexNet, ResNet, DenseNet, VGG, and GoogleNet.
• Analysis of CNN diagnostic systems based on metrics like sensitivity, accuracy, precision, specificity, and F1 score.

Main Results:

• CNNs demonstrate essential value and good performance in the experimental diagnosis of COVID-19 from medical images.
• Various CNN techniques like rectified linear units, batch normalization, data augmentation, and dropout are integral to system performance.
• Existing CNN systems show promising results in classifying COVID-19 cases using chest imaging data.

Conclusions:

• Convolutional Neural Networks are a valuable tool for assisting in the diagnosis of COVID-19 via chest X-ray and CT imaging.
• Further improvements in CNN performance can be achieved by expanding datasets, incorporating GPU acceleration, and refining data preprocessing techniques.
• This review contributes to future research by highlighting the capabilities and potential of CNNs in medical image-based disease diagnosis.