Detection of Human Visceral Leishmaniasis Parasites in Microscopy Images from Bone Marrow Parasitological Examination

Round 1

Reviewer 1 Report

This this article the authors aims  in detecting amastigotes from microscopy images using deep learning techniques. The proposed methodology consists of segmenting the Leishmania parasites in the images, precisely indicating the location of the amastigotes in the image. This study is very interesting and is of publication value.

Comments.

The images considered must be of high quality.

Some slides of bone marrow of bad qualities and other intracellular parasites may be also considered.

Acceptable

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear Authors,

I have read your article “Segmentation of human visceral leishmaniasis parasites in microscopy images from bone marrow parasitological examination” and can recommend it for publication after an appropriate revision has been done. You have an interesting text that may be used in medical practice. Below you will find my detailed ccomments.

Minor

1. Lines 3 and 23: “more than 95%” sounds awkward. What mean percentage did you find in literature?

2. Line 3: I would remove “current and”

3. Line 5: please omit tautology

4. Lines 5-6: I Abstract, it is better to write “computer analysis” while in the main text you may use “computer vision”

5. Lines 10-11: please expand acronyms

6. Line 16: I would remove “”According to theWorld Health Organization (WHO)”. Since which times the WHO is a great medical authority? To refer to CDC or like organisations would be more proper

7. Line 122 and so forth: I think, the purpose of your work is to detect amastigotes and count them rather than to segment them. Segmentation is barely one of your methods, not your goal. Please think about rephrasing.

8. Line 214: “to randomly reduce…” is a split infinitive. Please rephrase

9. Line 215: “not to lose” instead of “to not lose”

10. Line 223: the same

Major

1. Please add an information about the ways of VL transmission to Introduction. Why are there so many infected people in Brazil? Are there local niduses of VL in Brazil? Is the disease inborn or acquired? Which people’s habits can lead to parasite Leischmania chagasi infiltrating a human organism?

2. Please add Limitations subchapter to Results and Discussion where you could specify possible drawbacks of your detection methodology. What may be the sources of potential mistakes in detecting the parasites?

3. Please make you conclusions very brief statements of what you have achieved in this works and how your achievements can be used in healthcare rather than what you plan to do further.

4. Please correct the title of your paper so that "segmentation" should transform to "detection".

Thank you,
the Reviewer

There are several recommendations on minor English improvement on my side that the authors can do themselves.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

Use of the English language is fine. However, several sentences require re-formatting to correct grammatical errors that could otherwise lead to wrong interpretations and difficulties in understanding. 

Passive voice is normally used in sceintific writing (eg: The "the test was carried out" instead of "we did the test"). 

References should be revisited to ensure adherence to journal guidelines. Minor variations in formatting was noted. 

This work includes imaging and IT technology. But the results would be of interest to clinicians and laboratorians. Therefore, authors should preferably  include a few operational definitions for some key scientific terms in this paper, for better understanding. eg: to explain the procedures behind masks, binary masks, clipping etc.

Techniques explained in this paper have their own advantages as compared to traditional detection methods (better accuracy, minimize the demand for training persons for microscopy, avoid the need for molecular and serological assays and resultant lab costs etc.). Authors could include this information for completeness. 

Presented techniques also have their own limitations as compared to clinical naked eye based visualization techniques. Challenge to maintain clinical and wet lab competencies, misdiagnosis of atypical morphologies, dependency on the quality of original images, inability to carry out without trained experts etc. Such limitations should be discussed in the paper. 

There should be consistency in using abbreviations. Use the term VL and avoid LV to indicate visceral leishmaniasis. VL is the accepted and standard abbreviation. 

Comments for author File: Comments.pdf

English is of reasonable standard. However minor modifications are required throughout the paper.

1. Grammatical corrections are required to interpret what authors wish to interpret in an accurate manner.

2. Authors are reminded of the usage of passive voice in scientific writing. 

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 4 Report

The data might be interesting but it has to be improved substantially since there are many concept misleadings and not explained that the reader will bot be able to understand the correct meaning of it. Before its further evalaution the soundness of the science behind, the description of the methods and the of the results has to be really improved.

It has to be improved

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Though the authors have agreed to the limitation but still it would be better to use other intracellular disease images for detection specificity.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors have paid attention to all my remarks and made the improvements truly conscientiously. The text may be published. No more comments.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

The aim of this study was to evaluate the potential assistance that deep learning techniques may have for detecting Leishmania amastigotes in microscopy images. The fundamental ground of the authors is that specialized laboratories carry out the LV diagnosis and the sampling, from the bone marrow, and direct observation of parasites is rather difficult; additionally, they suggest that the used methodologies have great automation power through automatic methods based on computer vision.

 The manuscript herein presented, is cut edging and interesting and ingenious and illustrates and approach for improving automation of diagnosis in leishmaniasis.

 However, still some issues have to be commented.

1. The English language has improved considerably, still there are some caveats that the editors should check before publication.
2. It would be interesting to mention why the authors claim that this methodology is only useful for visceral leishmaniasis.
3. If the aim is that the methodology would be used in many contexts, the authors should better explain the terminology used. Already in the abstract the following phrase is stated “In parasite segmentation, in this methodology, a Dice of 80.4% was obtained, Intersection over Union (IoU) of 75.2%”, many readers not experts in these methodologies may run away and not get the message the authors want to transmit.
4. The authors mention the following: “Tegumentary Leishmaniasis (TL), also known as cutaneous leishmaniasis” in the introduction. This is a misleading concept that MUST be corrected since as can be described in many validated publications American tegumentary leishmaniasis is characterized by a spectrum of clinical manifestations including localized cutaneous, diffuse cutaneous, disseminated (DL), and mucocutaneous (MCL) leishmaniasis. The authors must include references for this issue and there are many in the literature that would help them to make the appropriate correction.
5. Again, there is a misleading concept when the authors mention that: “VL is a disease caused by protozoa of the Leishmania chagasi species”. This is true only for the Americas (please for example refer to the page of PAHO https://www3.paho.org/hq/index.php?option=com_content&view=article&id=6420:2012-leishmaniasis-visceral&Itemid=39347&lang=es#gsc.tab=0, under etiology).
6. In materials and methods please explain the following phrase in page 4: “This work’s objective is to detect amastigotes in slide field images. This process has as output the identification of amastigotes, which directly reflects the degree of infection of the patient” What the authors mean by degree of infection?
7. In page 5 please define “Panotic staining”, as well as “blade field images” and “pre-processing” of the images for the readers not familiar with the technique.
8. In page 6, when the authors mention that: “The images were divided into clippings to use the segmentation methodology” please specify if this clipping is perpendicular or parallel to the image and why.  Again, it would make the article easier to understand for the readers.
9. In page 7, when the authros mention “96 x 96”; please clarify if these dimensions are pixel the same is true further down where it says “8 x” and through the manuscript.
10. In page 8, please clarify how validation was performed.
11. In page 9, include a reference for the readers to better understand the “Unet” model, especially since it was originally designed for neural networks. Also, the authors should mention which is the main way to predict if the parasites are present or not and why.
12. In page 10, please clarify the following phrase: “The results obtained for the best segmentation model were a Dice of 0.804, IoU of 0.752, and AUC of 286 0.859. Also, False Positive Rate (FPR) = 0.004 and False Negative Rate (FNR) = 0.278. The FPR metric indicates the background regions incorrectly segmented as amastigotes, and the FNR measure indicates the regions containing amastigotes and incorrectly identified as image backgrounds”. It is important for the readers not familiar with the technique.
13. Finally, it is important that the authors stress the limitations of general use of this technique since the cost as well as the training of the technicians and the potential “real use” of it in the field can represent a challenge to its wide implementation.

For all these reasons I recommend to evaluate again the manuscript after these changes are included.

1. The English language has improved considerably, still there are some caveats that the editors should check before publication.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 3

Reviewer 1 Report

No comments

Reviewer 4 Report

I congratulate the authors for the improvement on this manuscript which is very interesting.

I only have a minor request.  In line 28, after infantum include: (called Leishmania chagasi in the Americas).

Thanks