Geometric and Topological Bases of a New Classification of Wood Vascular Tissues, Part 2: Classification of Vessels According to Their Grouping
Round 1
Reviewer 1 Report
The paper deals with how a new classification of wood vessels can improve the use of computer pattern recognition. In general the paper has merit as it proposes an effective way to do wood identification utilizing software in an innovative way.
Overall the paper requires many corrections. the introduction of the study is not clear and is written in the first person. This is not acceptable for a scientific publication, thus the paper need in depth editing.
The methods are not clear, there is a broad description of the vessel patterns but no mention about the software were the data will be uploaded, how it will be uploaded. Even the vessel arrangement description becomes confusing. Also there is no mention on how to make the software identify differences between vessel and parenchyma arrangements.
The result section seems as a continuation of the methods, so they do require more clarity. Several formulas are stated this like the density coefficient and that should be explained in more detail in the methods section.
The discussion is good, but without clear results and methods is not easy to understand.
Finally the references are not well cited. Including just the link of a publication is disrespectful to the reviewers and readers.
Author Response
Comments from Reviewer 1
The paper deals with how a new classification of wood vessels can improve the use of computer pattern recognition. In general the paper has merit as it proposes an effective way to do wood identification utilizing software in an innovative way.
Comment 1: Overall the paper requires many corrections. the introduction of the study is not clear and is written in the first person. This is not acceptable for a scientific publication, thus the paper need in depth editing.
Author response: We gratefully accept the recommendation for a precise and detailed editing of the text. In the introduction a precise stylistic editing of the text is made. It has been significantly shortened in order to make it clearer.
Comment 2: The methods are not clear, there is a broad description of the vessel patterns but no mention about the software were the data will be uploaded, how it will be uploaded. Even the vessel arrangement description becomes confusing. Also there is no mention on how to make the software identify differences between vessel and parenchyma arrangements.
Author response: Thank you for this suggestion. It would have been interesting to explore the differences between vessel and parenchyma arrangements. However, in the case of our study, it seems slightly out of scope because we do not discuss the vessel detection algorithm. That is a completely different study, explored in our previous researches. We have measured some properties, such as the major and minor axis length, the orientation, the centroid and the perimeter of the segmented cell cavity, while trying to draw an ellipse (or circle) over the edges. In the current algorithm, we assume, that the data entries are already detected vessels, stored in a matrix after the detection process. (Normally, it can be saved in a csv file, but we do not think that is needed.) We do not propose an application yet.
In classifying the species, we have been guided by the rule that one arrangement of vessels does not fall into two classification groups. There are limits to the range in which the indicator falls in order to determine the affiliation of a group in one or another class. In addition, the ratio of the distances between the vessels is monitored (gmax/gmin).
Comment 3: The result section seems as a continuation of the methods, so they do require more clarity. Several formulas are stated this like the density coefficient and that should be explained in more detail in the methods section.
Author response: The article uses models that include the following main indicators (A, B, etc.). In the edited text they have been moved from the introduction to Part 2. Methods and materials. They are defined by formulas and described in detail in another publication of the authors. The same article is cited in the edited text.
Comment 4: The discussion is good, but without clear results and methods is not easy to understand.
Author response: Thank you for the comment. After the corrections made we hope that it become more clear.
Comment 5: Finally the references are not well cited. Including just the link of a publication is disrespectful to the reviewers and readers.
Author response: Thank you for pointing this out. We agree with this comment. Therefore, we have correct the citations in the text and the list of references.
In addition to the above comments, all spelling and grammatical errors have been corrected.
We look forward to hearing from you in due time regarding our submission and to respond to any further questions and comments you may have.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The work aims to compare and analyze the values of observable and calculated quantitative indicators (coefficients) of the anatomical structure of the wood, to classify the tree species according to their mutual position (topology) in groups of neighboring vessels. The manuscript is very well written, understandable, and contains clear interpretations.
Author Response
Thank you!
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Please submit again as the file in the manuscript section corresponds to the author's response.
Author Response
We would like to inform you that we have read your recommendations very carefully. We have made the necessary changes to the text we are enclosing here (sustainability-1474775 19.12.docx ). Please accept our thanks for your work on the text, which increased the quality of the article.
Author Response File: 
Author Response.docx
Round 3
Reviewer 1 Report
The introduction lacks any reference. Many sources refer to the classification of vessel elements. Also, the introduction states that it is highly complex. There is enough information about vessel elements classified by their distribution (diffuse, ring-porous, and semi-ring-porous) and their numbers. The way the introduction wants to make a statement about the difficulty of this classification is vague and needs work.
The state of the arts section mentions that the CNN will be used based on the texture of the wood. For wood anatomists, the texture refers to fine, medium, or coarse wood. Is this what the authors are referring to? If so, there needs to be further information about this as readers would find it difficult to understand. Also, when mentioning previous works by author name, please include them in the manuscript next to the reference number. Besides this, the section overall lacks detail. Would you mind working on this?
The materials section refers to analyzing tree species. This can be unclear, as if it's trees that are being analyzed, then it would be part of the field of dendrology. Also, after mentioning that trees are being studied, then this means that the proposal is based on a non-destructive test. If the wood is being photographed for analysis, the methods should state that wood from 76 species will be used for the study. Also, add a table displaying the names of the used species.
Please be consistent with the species naming section explaining the vessel arrangement. Some species have their common name and only the genus in parenthesis (this also has to be fixed as the complete name should be stated), and other sections only have the genus mentioned. Would you mind correcting and following a single format?
The results section looks more precise and delivers better than the previous version. Although there is a significant improvement, similar errors as those stated above remain in this part. Please correct.
Overall the paper has merit, but I would recommend the authors to do a more exhaustive literature review, as methods for machine learning for wood identification have been done previously, and it would be good to see how this method can improve or build on current technologies. I would refer to the work of the Forest Products Lab and the work on the XiloTron by Dr. Wiedenhoeft.
Author Response
Comments from Reviewer 1
Comment 1: The introduction lacks any reference. Many sources refer to the classification of vessel elements. Also, the introduction states that it is highly complex. There is enough information about vessel elements classified by their distribution (diffuse, ring-porous, and semi-ring-porous) and their numbers. The way the introduction wants to make a statement about the difficulty of this classification is vague and needs work.
Author response: The way the introduction wants to make a statement about the difficulty of this classification is cleared. I acknowledge the remark and we made corrections
Comment 2: The state of the arts section mentions that the CNN will be used based on the texture of the wood. For wood anatomists, the texture refers to fine, medium, or coarse wood. Is this what the authors are referring to? If so, there needs to be further information about this as readers would find it difficult to understand. Also, when mentioning previous works by author name, please include them in the manuscript next to the reference number. Besides this, the section overall lacks detail. Would you mind working on this?
Author response: In the field of image processing, a texture of an image is a set of features, computed from the input image, aiming to measure the visible texture (pattern) of an image.
Also, when mentioning previous works by author name, please include them in the manuscript next to the reference number. – it is corrected
Besides this, the section overall lacks detail. Would you mind working on this? - As future work, the proposed method will be used as input for computer vision algorithms, based on both supervised and unsupervised learning.
Comment 3: The materials section refers to analyzing tree species. This can be unclear, as if it's trees that are being analyzed, then it would be part of the field of dendrology. Also, after mentioning that trees are being studied, then this means that the proposal is based on a non-destructive test. If the wood is being photographed for analysis, the methods should state that wood from 76 species will be used for the study. Also, add a table displaying the names of the used species.
Author response: Indeed, the recognition of species by external morphological features is the subject of Dendrology, but we have nothing to do with it. The article analyzes the structure of wood, not trees. The table was removed due to the large volume, but now we are returning it. I accept the remark and we made adjustments.
Comment 4: Please be consistent with the species naming section explaining the vessel arrangement. Some species have their common name and only the genus in parenthesis (this also has to be fixed as the complete name should be stated), and other sections only have the genus mentioned. Would you mind correcting and following a single format?
Author response: I acknowledge the remark and we made corrections.
Comment 5: The results section looks more precise and delivers better than the previous version. Although there is a significant improvement, similar errors as those stated above remain in this part. Please correct.
Author response: I acknowledge the remark and we made corrections.
Comment 6: Overall the paper has merit, but I would recommend the authors to do a more exhaustive literature review, as methods for machine learning for wood identification have been done previously, and it would be good to see how this method can improve or build on current technologies. I would refer to the work of the Forest Products Lab and the work on the XiloTron by Dr. Wiedenhoeft.
Author response: Thank you for the comment. We get familiar with the work of Dr. Wiedenhoeft and we referred it in our research.
The neural network can accept different types of information, such as pixels, vectors, or entire images. The work of Dr, Wiedenhoeft use images as entries of the NN architecture. Yes, the proposed classification could be useful for quantitative wood anatomy studies such as comparison of clones, differentiating growing patterns, and cell aggregations. They can also be used in algorithms aimed at the elaboration of image analysis. The obtained results will be high quality entries for constructions of machine learning architectures. The best information is with small quantity (we are talking about the length of the vector for example, not for the number of vectors) of input information to give quality (descriptive) information. For example, see the Iris dataset and how it is used for clustering. This example illustrates how a self-organizing map neural network can cluster iris flowers into classes topologically, providing insight into the types of flowers and a useful tool for further analysis:
https://www.mathworks.com/help/deeplearning/ug/iris-clustering.html
The appropriate correction can be found in the text (see p. 3, 4).
We look forward to hearing from you in due time regarding our submission and to respond to any further questions and comments you may have.
Sincerely,
Assoc. Prof. Nikolai Bardarov, PhD
Round 4
Reviewer 1 Report
The current version is improved. My only observation is regarding the wood species table. The scientific name (Latin) should be written in italics with their authorship.
Author Response
We agree with the remarks and have made adjustments
