Machine learning for medical imaging-based COVID-19 detection and diagnosis.

Area of Science:

• Medical Imaging and Artificial Intelligence
• Infectious Disease Diagnostics

Background:

• Coronavirus disease 2019 (COVID-19) poses a global health challenge due to rapid transmission and high mortality.
• Early COVID-19 detection is vital for pandemic control, but traditional methods like RT-PCR have limitations, including high false negative rates.
• Medical imaging (CT, X-ray) reveals distinct COVID-19 features, differentiating it from healthy cases and other pneumonias.

Purpose of the Study:

• To review recent advancements in machine learning (ML) for COVID-19 detection using medical imaging.
• To focus on ML models applied to CT and X-ray images, as published in high-impact journals.
• To discuss characteristic imaging features of COVID-19 patients.

Main Methods:

• Review of machine learning applications in COVID-19 diagnosis and severity assessment.
• Focus on deep learning algorithms, particularly convolutional neural networks (CNNs).
• Analysis of studies utilizing CT and X-ray imaging data.

Main Results:

• Machine learning models demonstrate significant potential in diagnosing COVID-19 from medical images.
• Deep learning, especially CNNs, is widely used for image segmentation and classification tasks.
• Many ML modules achieve high predictive accuracy, even with small datasets.

Conclusions:

• Machine learning, particularly deep learning on CT and X-ray images, offers a powerful tool for COVID-19 detection.
• ML applications can aid in early diagnosis and potentially improve patient outcomes.
• Further research and application of these models are warranted for pandemic management.