Classification of Red Blood Cells Using Time-Distributed Convolutional Neural Networks from Simulated Videos
Round 1
Reviewer 1 Report
The present article discusses a new approach to classifying RBC elasticity. The authors demonstrate the potential of using CNNs and simulation-based data to accurately classify RBC elasticity.
However, the article lacks a discussion of the shortcomings of the approach chosen by the authors.
I recommend that the authors add to the text of the article (in the discussion section) a paragraph devoted to discussing the limitations of using the CNNs method in question.
Author Response
Response to Reviewer 1 Comments
Point 1:
The present article discusses a new approach to classifying RBC elasticity. The authors demonstrate the potential of using CNNs and simulation-based data to accurately classify RBC elasticity.
However, the article lacks a discussion of the shortcomings of the approach chosen by the authors.
I recommend that the authors add to the text of the article (in the discussion section) a paragraph devoted to discussing the limitations of using the CNNs method in question.
Response 1:
We have addressed the point by:
- adding the penultimate paragraph in the "Introduction" section;
- adding the "Discussion" subsection 4.4.
Reviewer 2 Report
The authors proposed a novel approach to determine RBC elasticity by analyzing video recordings and using a convolutional neural network for classification. The paper can be published in the journal after minor revisions:
1) Please compare the characteristics of numerical simulation and video classification briefly.
2) You need to explain the properties of cell classification or cite a relevant paper, for instance: https://doi.org/10.1515/bmt-2022-0232
3) What is "simulation experiments"?
4) Did you perform numerical simulations? If so, you need to present their characteristics, for example, governing equations and boundary conditions.
5) Fig. 2 does not show a channel.
6) You need to provide the numerical aspects of deformability, surface tension, etc.
Minor editing of English language required.
Author Response
Response to Reviewer 2 Comments
Point 1: Please compare the characteristics of numerical simulation and video classification briefly.
Response 1: The point was partially addressed in the original manuscript in lines 42-57. We added the penultimate paragraph in the "Introduction" section and the "Discussion" subsection 4.4 to deal with this more extensively.
Point 2: You need to explain the properties of cell classification or cite a relevant paper, for instance: https://doi.org/10.1515/bmt-2022-0232
Response 2: Last sentence of the second paragraph (lines 27-28) was added.
Point 3: What is "simulation experiments"?
Response 3: The terminology of the article adopts concepts from the disciplines it develops. Several concepts meet in the article, one of them is a numerical model of blood flow and "simulation experiment" means the simulation of the flow of blood plasma and red blood cells moving in it. This movement takes place with specified input conditions, such as the speed of individual points of the fluid and the speed acting on the particle on the surface of the blood cell. These velocities affect the blood flow at the beginning of the simulation, but gradually the flow stabilizes. At this time, we will store data, that is, information about the position and speed of some particles on the surface of the cell. These data serve as an input dataset, on which neural networks are subsequently trained. The simulation experiment is described in the "Data preparation" section 3.
Point 4: Did you perform numerical simulations? If so, you need to present their characteristics, for example, governing equations and boundary conditions.
Response 4: We added new information in the first paragraph of the 3rd section, including citations [25], [26] which refer to a more detailed description how the simulations were generated.
Point 5: Fig. 2 does not show a channel.
Response 5: We have replaced the figure with a new one and changed the caption of the figure.
Point 6: You need to provide the numerical aspects of deformability, surface tension, etc.
Response 6: The first paragraph of the added "Discussion" subsection 4.4 contains explanation related to this point. In case this point was related to how the numerical simulation was created and how it incorporates these aspects, this should be addressed similarly as point 4 - by the added reference to a previous work.