Combining UNet 3+ and Transformer for Left Ventricle Segmentation via Signed Distance and Focal Loss
Round 1
Reviewer 1 Report
Comments file is attached
Comments for author File: Comments.pdf
Author Response
Please see the attachment.Author Response File: Author Response.docx
Reviewer 2 Report
Review of the article
Combining UNet 3+ and Transformer for Left Ventricle Segmentation via Signed Distance and Focal Loss
The authors describe that Left ventricle (LV) segmentation of cardiac magnetic resonance (MR) images is essential to evaluate cardiac function parameters and diagnosing cardiovascular diseases (CVDs). Doing the segementation manually it takes 20 minutes for a clinician with some obstacles during segmentation. Existing segementation algorithms are dificult to be applied in clinical settings and are based on machine learning relying on the initialization step. In the presented work, a network based on an encoder-decoder architecture for automatic LV segmentation of short-axis cardiac MR images was proposed.
Overall, the paper is well written with sufficient details for all performed steps, which would need only some minor spell checking e.g. the use of „s“ or „ing“ or not.
Personally I would like to see some more MR images since data of 145 subjects were analyzed showing visually the differences using the different segmentation techniques highlighting the benefits for clinical staff.
Minor things:
- Inconsistent use of abbrevations e.g. in line 1 use „magnetic resonance (MR) images“ versus in line 20 „magnetic resonance imaging (MRI)“
- Please check correct spelling, partly errors like line 30 „to segment a patient’s cardiac MR images“ need to be corrected to „to segment a patient’s cardiac MR image“
- I personally would wish to see some sample images under Part 2 that show the segmentation result e.g. of the traditional segmentation methods, as well as introductory Figure of heart parts (epicardium, endocardium, …)
- Regardind the structure of a classical research article part 2 could be included into the introduction and subheading 2 is then Methods
- Figure description could give some more details like Figure 1: Which model? Segmentation of what?
- in figure 2 the text and probably also the MR images are blurred/pixelated. Space is missing behind (a) in „(a)Cardiac MR image“. I would recommend to use arrows to point the reader to the important things in MR images.
- Line 206: Please give information which version of PyTorch was used.
- Line 256: Where or what is Table II?
Introduction
Cardiovascular diseases (CVDs) is a bit too general, e.g. in line 26 turns up stroke, something specific about which CVDs are addressed, which changes in heart etc. could be explained.
Author Response
Please see the attachment.
Author Response File: Author Response.docx