A Two-Step Learning Model for the Diagnosis of Coronavirus Disease-19 Based on Chest X-ray Images with 3D Rotational Augmentation

Round 1

Reviewer 1 Report

Overall, the paper is well structured and written. Just a few minor suggestions from me.

Please make sure to include the date of access for your online references, such as the WHO website, etc.

Ensure that every abbreviation is fully introduced in its first appearance.

In the introduction (in the last paragraph) please state your aims, as it is not clear what research gap you are trying to address. Also, clinical relevance must be addressed.

To me, your discussion looks more like an extended results section at the moment. Please discuss how your findings address your research questions, how they contribute to a broader field of knowledge, and their clinical relevance.

Author Response

We attached the reply letter. Thank you.

Author Response File: Author Response.docx

Reviewer 2 Report

The manuscript “Deep Learning for Diagnosis of Coronavirus Disease-19 Based 2 on Chest X-ray Images with You Only Look Once v4 Algorithm” by Kwon et al. proposed a deep learning model based on YOLOv4 to classify normal, COVID-19, lung opacity, and viral pneumonia using Chest X-Ray (CXR) images.

The proposed method involves a lung detection model for automatically generating label files, a 3D rotational augmentation based on 3DPI, and a two-step learning model based on YOLOv4 which comprises a two-class model that classifies normal and disease images, and a three-class model that classifies COVID-19, lung opacity, and viral pneumonia images among the diseases classified in the two-class model. It is shown that the two-step learning model with 3D rotational augmentation shows better detection performance than other alternative methods.

The manuscript is written well. The results are also convincing. I would recommend it for publication.

For detecting covid, rapid antigen test should also be mentioned. It is one of the most popular methods today.

The proposed method seems promising in detecting covid-19 based on CRX images. However, the significance of the work may be limited since not every covid-19 patient has symptoms in the lung. The latest variants of the virus have milder symptoms and may not infect lungs at all. The authors may consider testing the method for lung cancer detection.

There are a few minor English errors.

Author Response

We attached the reply letter. Thank you.

Author Response File: Author Response.docx

Reviewer 3 Report

The paper is remarkable and astounding. However I would suggest to change the title that represent general and sounds more quantitative.

Author Response

We attached the reply letter. Thank you.

Author Response File: Author Response.docx