Next Article in Journal
Exact Solutions to Cancer Laser Ablation Modeling
Previous Article in Journal
Noise Reduction in LED-Based Photoacoustic Imaging
Previous Article in Special Issue
Complementary Metal Oxide Semiconductor-Based Optical Detection System for Fluidic Cellular Medium pH Quantification
 
 
Article
Peer-Review Record

Microstructured Waveguide Sensors for Point-of-Care Health Screening

Photonics 2025, 12(4), 399; https://doi.org/10.3390/photonics12040399
by Svetlana S. Konnova 1,*, Pavel A. Lepilin 1, Anastasia A. Zanishevskaya 1, Alexey Y. Gryaznov 1, Natalia A. Kosheleva 2, Victoria P. Ilinskaya 3, Julia S. Skibina 1 and Valery V. Tuchin 4,5,6
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Photonics 2025, 12(4), 399; https://doi.org/10.3390/photonics12040399
Submission received: 28 February 2025 / Revised: 14 April 2025 / Accepted: 17 April 2025 / Published: 20 April 2025
(This article belongs to the Special Issue Optical Sensors for Advanced Biomedical Applications)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The authors present the results of experimental studies on label-free sensors based on hollow-core microstructured optical waveguides (HC-MOW) for human blood analysis. Here, my comments on the manuscript:

  1. In recent years, substantial research efforts have been devoted to the development of ultra-sensitive biosensors. Therefore, the literature review should be expanded to include more recent studies, particularly those reporting key performance metrics. Relevant references include: High-resolved near-field sensing by means of dielectric grating with a box-like resonance shape. IEEE Sensors Journal, 24(5), 6045-6053, 2024.; Label-free monitoring of human IgG/anti-IgG recognition using Bloch surface waves on 1D photonic crystals. Biosensors, 8(3), 71, 2018.; Label-free specific detection of immunoglobulin G antibody using nanoporous hydrogel photonic crystals. Sensors and Actuators B: Chemical, 180, 107-113, 2013.; Optical slot-assisted metasurface for IgG protein detection. Journal of Physics: Conference Series (Vol. 2725, No. 1, p. 012001), IOP Publishing, 2024; Human IgG detection in serum on polymer-based Mach-Zehnder interferometric biosensors. Journal of Biophotonics, 9(3), 218-223, 2016.
  2. The novelty of the proposed approach is not clearly articulated. Numerous studies on microstructured optical fibers have already been reported in the literature. Therefore, the authors should explicitly highlight the unique contributions of their work. Furthermore, they should clarify the advantages of using a hollow-core microstructured waveguide in comparison to other sensing solutions.
  3. The authors should provide a justification for the selected operating wavelength, including considerations related to the fiber design. This would strengthen the rationale behind the device’s configuration and potential advantages.
  4. The spectral plots should be enhanced for better visibility. It would also be beneficial to overlay spectra in specific spectral regions to facilitate the evaluation of shifts and trends. This would improve data interpretation and allow for a clearer assessment of the sensor's performance.
  5. All figures should be scaled, centered, and formatted with a consistent font size. This would improve the overall readability and professionalism of the manuscript.

 

Comments on the Quality of English Language

The manuscript's English should be carefully revised for clarity and correctness. For example, in Figure 6, "Wavelenght" should be corrected to "Wavelength." A thorough proofreading is recommended to ensure technical accuracy and readability.

Author Response

  1. In recent years, substantial research efforts have been devoted to the development of ultra-sensitive biosensors. Therefore, the literature review should be expanded to include more recent studies, particularly those reporting key performance metrics. Relevant references include: High-resolved near-field sensing by means of dielectric grating with a box-like resonance shape. IEEE Sensors Journal, 24(5), 6045-6053, 2024.; Label-free monitoring of human IgG/anti-IgG recognition using Bloch surface waves on 1D photonic crystals. Biosensors, 8(3), 71, 2018.; Label-free specific detection of immunoglobulin G antibody using nanoporous hydrogel photonic crystals. Sensors and Actuators B: Chemical, 180, 107-113, 2013.; Optical slot-assisted metasurface for IgG protein detection. Journal of Physics: Conference Series (Vol. 2725, No. 1, p. 012001), IOP Publishing, 2024; Human IgG detection in serum on polymer-based Mach-Zehnder interferometric biosensors. Journal of Biophotonics, 9(3), 218-223, 2016.

 

Thank you for the comment, We have added information ( line 60-63) to the introduction and inserted links.

 

  1. The novelty of the proposed approach is not clearly articulated. Numerous studies on microstructured optical fibers have already been reported in the literature. Therefore, the authors should explicitly highlight the unique contributions of their work. Furthermore, they should clarify the advantages of using a hollow-core microstructured waveguide in comparison to other sensing solutions.

Thank you, the information added to the manuscript. Line 79-82

Microstructured waveguides with a hollow core and one row of capillaries have not yet been used to study blood serum in general. Glass waveguides are easy to manufacture and use, chemically stable and neutral, no special additional components applied to the surface are required, liquids enter the capillary without additional manipulation due to capillary forces.

 

3. The authors should provide a justification for the selected operating wavelength, including considerations related to the fiber design. This would strengthen the rationale behind the device’s configuration and potential advantages.

 

Thank you for your important comment. The main maxima of the comb of our waveguide are located in this range, which increases its sensitivity. Besides, this range of radiation is the most gentle for biological molecules

 

4. The spectral plots should be enhanced for better visibility. It would also be beneficial to overlay spectra in specific spectral regions to facilitate the evaluation of shifts and trends. This would improve data interpretation and allow for a clearer assessment of the sensor's performance.

 

Thank you for your comment, to improve the manuscript we have replaced Figure 5 and supplemented Figures 1 and 10. We hope that after this the work will become more understandable and professional.

 

 5. All figures should be scaled, centered, and formatted with a consistent font size. This would improve the overall readability and professionalism of the manuscript.

Thank you, the text was centered and aligned and the figures were formatted.

Comments on the Quality of English Language

The manuscript's English should be carefully revised for clarity and correctness. For example, in Figure 6, "Wavelenght" should be corrected to "Wavelength." A thorough proofreading is recommended to ensure technical accuracy and readability.

Thank you, the error has been fixed.

Reviewer 2 Report

Comments and Suggestions for Authors

Manuscript ID: Photonics-3531323

Title: Microstructured waveguide sensors for point-of-care health screening

The paper presents a biosensor system using hollow-core microstructured optical waveguides (HC-MOWs) for blood serum analysis. While the concept of label-free optical biosensing for point-of-care diagnostics is promising and aligns with current research trends, the manuscript requires substantial revisions to address methodological gaps, data interpretation issues, and clarity concerns before publication.

Major Weaknesses

  1. Sample Limitations
    • Small sample size (n=21) with no power calculation to justify statistical significance.
    • All subjects are female (35–45 years), limiting generalizability.
    • No details on inclusion/exclusion criteria or ethical approval for human samples.
  2. Clinical Relevance
    • Fails to link PCA results to specific biomarkers or clinical outcomes.
    • No ROC curves or sensitivity/specificity metrics to validate diagnostic utility.
  3. Methodological Concerns
    • HC-MOW manufacturing process lacks critical details (e.g., glass type, capillary dimensions).
    • No discussion of signal-to-noise ratio or detection limits.
    • Incomplete description of blood serum dilution protocol (saline concentration unspecified).
  4. Data Analysis
    • PCA implementation lacks transparency:
      • No variance explained by principal components.
      • No justification for wavelength range selection (350–1050 nm).
      • Missing scree plot or component loading analysis.
    • Double smoothing of spectra (Steps 2 and 7) risks data distortion.

Other Recommendations

  1. Introduction & Background
  • Clarify the novelty compared to prior HC-MOW biosensors (e.g.,).
  • Add a table comparing performance metrics (sensitivity, cost, speed) with existing POC devices.
  1. Methods
  • Provide:
    • Ethical approval documentation for human samples.
    • HC-MOW fabrication details (glass composition, drawing parameters).
    • Spectral resolution of the AvaSpec-ULS4096CL-EVO system.
    • Validation of Savitzky-Golay filter parameters (why window width=29?).
  1. Results
  • Include:
    • Raw vs. processed spectra in supplementary materials.
    • Error bars in PCA plots (Fig. 3).
    • Statistical comparison between HC-MOW and standard photometry results.
  1. Discussion
  • Address:
    • Clinical translation challenges (e.g., interference from hemolyzed samples).
    • Limitations of PCA for diagnostic classification.
    • Cost-benefit analysis vs. commercial biochemical analyzers.

Figures & Tables

  • Fig. 1: Add scale bar to cross-sectional image.
  • Fig. 2: Label critical optical components (e.g., collimator specs).
  • Table 1: Clarify units for wall thickness (d) and refractive indices.

Additional Notes

  • Consider adding machine learning approaches (beyond PCA) for spectral classification.
  • Discuss sterilization/reusability of HC-MOWs for POC applications.
  • Compare performance with similar waveguide-based biosensors.

Conclusion

While the HC-MOW biosensor concept shows potential, the current evidence is insufficient to support claims about clinical utility. Substantial revisions are required to strengthen the statistical analysis, validate clinical relevance, and improve technical transparency.

Author Response

Small sample size (n=21) with no power calculation to justify statistical significance.

A similar calculation is performed at the planning stage of the experiment before it is carried out. To implement it, it is necessary to know or assume the quality of the classifier used. In our case, this was impossible. The purpose of the study was to obtain a “proof of concept”

 

All subjects are female (35–45 years), limiting generalizability.

Indeed, this may negatively affect the generalizing ability of the model, however, in our case, such a sample structure is necessary to standardize the study

 

No details on inclusion/exclusion criteria or ethical approval for human samples.

The groups were formed only in accordance with the data of biochemical blood tests based on the presence of deviations in the parameters, other factors such as diagnoses, medical history, and medications used were not considered. The article is about the fact that blood changes can, in principle, be detected in this type of waveguide in visible light, using the appropriate spectral data processing method. The remaining statements require a larger amount of data. There is information about the approval by the ethics committee at the end of the article.

 

Fails to link PCA results to specific biomarkers or clinical outcomes.

In the discussion section, we present some reflections on this topic. The composition of the blood affects the spectrum as a whole, the shift of the waveguide comb and the intensity of the glow. Therefore, it is impossible to say what specifically affected the spectrum, you can only correlate with the data obtained by other methods and collect more statistics. With a large accumulated pool of data, it will be possible to try using AI.

 

No ROC curves or sensitivity/specificity metrics to validate diagnostic utility.

We do not have a sufficient sample and precise parameters for constructing ROC curves. In this paper, we only assume the diagnostic value of these waveguides and demonstrate the possibilities for population screening.

 

HC-MOW manufacturing process lacks critical details (e.g., glass type, capillary dimensions).

Thank you for your comment.

Line 147 Corrected in manuscript. Glass type Ar-Glass, Schott; The external and internal dimensions of the capillaries are shown in Figure 1.

 

No discussion of signal-to-noise ratio or detection limits.

 

Line 163-170 The noise in our opto-electrical measuring system is around 1% of the total intensity, and is corrected by the built-in spectrometer software and smoothing.

 

Incomplete description of blood serum dilution protocol (saline concentration unspecified).

 

Line 142 - 143 Thank you for your comment, we have added the information. 

 

No variance explained by principal components.

Line 321-334 Thank you for your important comment. The proportion of explained variance of the first three principal components is presented in Figure 8 in the text of the article. However, the manuscript does not actually provide numerical indication of these values. To increase the transparency of the results, we have added the numerical values ​​of the proportions of explained variance of the first three components directly to the text of the article and to the figure caption. 

 

 

 

No justification for wavelength range selection (350–1050 nm).

The main maxima of the comb of our waveguide are located in this range, which increases its sensitivity in this range. Besides, this range of radiation is the most gentle for biological molecules. 

 

Missing scree plot or component loading analysis.

Scree plot analysis is represented on the picture 8

 

Double smoothing of spectra (Steps 2 and 7) risks data distortion.

We agree with the remark that repeated smoothing of the data is associated with the risk of distortion of the spectra. That is why the Savitzky-Golay filter window was selected manually, with careful control to ensure that the spectral peaks were not shifted or distorted noticeably after the second smoothing stage. We made sure that the selected filter parameters provide the necessary noise suppression with minimal distortion of the original shape of the spectra.

Other Recommendations

Introduction & Background

 

Clarify the novelty compared to prior HC-MOW biosensors (e.g.,).

Thank you for the comment, we have added this information to the introduction.

 

Add a table comparing performance metrics (sensitivity, cost, speed) with existing POC devices.

The presented work demonstrate a principle, but not a final version of a biosensor.

 

Methods

Provide:

Ethical approval documentation for human samples.

Thank you for your comment, we have the documents and are ready to provide them if necessary or add to supplemental materials.

 

HC-MOW fabrication details (glass composition, drawing parameters).

Thank you for the comment, the we have added the information about glass on line 150 to the materials and methods, the procedure is standard for stuck-and-draw technology.

 

Spectral resolution of the AvaSpec-ULS4096CL-EVO system.

Line 165 Spectral resolution is 0.25 nm

 

Validation of Savitzky-Golay filter parameters (why window width=29?).

 Lines 281 - 286: The filter window width of 29 points was chosen empirically, based on the analysis of spectra with different degrees of noise and testing several window width options (e.g. 15, 21, 25, 29, 31, 35). The window width of 29 was optimal based on the results of visual and quantitative analysis, allowing for effective noise suppression with minimal impact on the location and relative intensity of spectral peaks. 

 

Results

Include:

 

Raw vs. processed spectra in supplementary materials.

We will put this data to the supplementary materials.

 

Error bars in PCA plots (Fig. 3).

The figure shows the comb maxima, on which the corresponding wavelengths are marked.

 

Statistical comparison between HC-MOW and standard photometry results.

 

Thank you for your comment. We showed it on the Picture 4 and discussed on line 251 - 257

 

Discussion

Address:

 

Clinical translation challenges (e.g., interference from hemolyzed samples).

 

 Thank you for your comment. Sorry, but we we didn't understand your remark

 

Limitations of PCA for diagnostic classification.

 

In the current study, we have shown that PCA can effectively separate groups based on optical characteristics, but we do not propose to use PCA directly as a diagnostic classifier without additional validation. In future studies, supervised machine learning methods (e.g., logistic regression, support vector machines, or neural network classifiers) trained on PCA results are planned for clinical diagnostics.

 

Cost-benefit analysis vs. commercial biochemical analyzers.

Thank you, we do not discuss these aspects in this article, but obviously the costs of our biosensor for general screening will be low, at least due to the absence of additional reagents and short screening time.

Figures & Tables

Fig. 1: Add scale bar to cross-sectional image.

Thank you for the comment, we have added scale bar (Figure 1.).

 

Fig. 2: Label critical optical components (e.g., collimator specs).

 

All components of the optical scheme are presented in the figure caption (Figure 2.).

 

Table 1: Clarify units for wall thickness (d) and refractive indices.

 

Sorry, There is no such parameters in this table.

Additional Notes

Consider adding machine learning approaches (beyond PCA) for spectral classification.

Line 379-382

In future studies, supervised machine learning methods (e.g., logistic regression, support vector machines, or neural network classifiers) trained on PCA results are planned for clinical diagnostics.

 

Discuss sterilization/reusability of HC-MOWs for POC applications.

 

Line 153 - 154

 

Compare performance with similar waveguide-based biosensors.

 

At this stage of work we cannot discuss the performance of the system, as it requires validation and further development.

 

 

Reviewer 3 Report

Comments and Suggestions for Authors

The data in Fig. 4(a) does not make sense. In Fig. 6, the authors presented the refractive index difference of the sample from the patient group and the control group. Only minimal differences from the two set can be observed, and some sample from the patient group have similar optical properties with the control group. In fig. 4, however, the authors show that there is a significant difference between the two group, even though there is no physics evidence support that this could happen. It looks to me that the difference from the two group were due to using two pieces of equipment. 

Comments on the Quality of English Language
  1. The authors should grammar check their manuscript. For example: in abstract, “In this paper, the results of experimental studies of label-free sensors based on hollow-core microstructured optical waveguide (HC-MOW) for human blood serum analysis obtained.” obtained -> were obtained. 
  2. In methods section,  «sick group» should be rephrased to a more professional term. 

Author Response

The data in Fig. 4(a) does not make sense. In Fig. 6, the authors presented the refractive index difference of the sample from the patient group and the control group. Only minimal differences from the two set can be observed, and some sample from the patient group have similar optical properties with the control group. In fig. 4, however, the authors show that there is a significant difference between the two group, even though there is no physics evidence support that this could happen. It looks to me that the difference from the two group were due to using two pieces of equipment. 

 

As shown in the article in section 3.1, the refractive index in the waveguide core determines the position of the transmission maxima in the waveguide spectrum. As can be seen in Figures 4 and 6, the spectra of "sick" and "healthy" samples have a significant difference in the intensity of the peaks in the wavelength range corresponding to the complex absorption of various blood components. In our opinion, the significant difference in the spectra is explained by the large interaction length of light in the waveguide, as a result of which the absorption effect is enhanced.

 

Comments on the Quality of English Language

The authors should grammar check their manuscript. For example: in abstract, “In this paper, the results of experimental studies of label-free sensors based on hollow-core microstructured optical waveguide (HC-MOW) for human blood serum analysis obtained.” obtained -> were obtained. 

In methods section,  «sick group» should be rephrased to a more professional term. 

 

Thank you for your comments, we have corrected some errors in the language, I hope after this the work has become more professional.

 

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The Authors have modified the manuscript according to the Reviewers' comments.

Author Response

The team of authors expresses deep gratitude to the reviewers for their attentive attitude to our work and valuable comments.

Reviewer 2 Report

Comments and Suggestions for Authors

Evaluation of Revised Manuscript: "Microstructed waveguide sensors for point-of-care health screening"

The authors have made substantial revisions addressing many of the technical points raised in the previous review. Key information regarding experimental setup, data processing, and results interpretation (e.g., ethics statement, component specifications, variance explained by PCA, loading plots, confidence ellipses) has been added. The paper is improved and clearer.

However, several significant concerns remain, primarily related to the study design limitations and the depth of discussion surrounding practical application and potential interferents. While the authors position this as a "proof-of-concept" study, the limitations impacting the generalizability and immediate applicability of the findings need to be more explicitly acknowledged and discussed within the manuscript text itself.

Point-by-Point Evaluation Based on Previous Comments and Author Replies:

  1. Small sample size (n=21) / No power calculation:
    • Status: Addressed partially. The authors correctly state power calculations are typically done a priori and justify the current work as "proof of concept" (Lines 118-121). However, the manuscript should more strongly emphasize in the Discussion/Limitations section that the small sample size limits statistical power, prevents robust subgroup analysis, and restricts the generalizability of the findings. The "proof of concept" nature needs to be framed with these caveats.
  2. All subjects female (35–45 years):
    • Status: Addressed partially. The sample characteristics are stated (Lines 127-128). The authors' reply mentions standardization as the reason, but this justification is not included in the manuscript text. This rationale should be added to the Methods, and the resulting significant limitation on generalizability must be clearly stated in the Discussion/Limitations.
  3. Inclusion/Exclusion Criteria & Ethical Approval:
    • Status: Addressed partially. The ethics statement is now present (Lines 445-447). However, the description of participant groups (Lines 127-132) still lacks formal inclusion/exclusion criteria beyond "healthy women preparing for IVF" vs. "patients of a cardiology clinic with chronic diseases". What specific diseases? Were controls screened for underlying conditions? This vagueness remains a weakness.
  4. Linking PCA to Biomarkers/Outcomes:
    • Status: Addressed. The Discussion (Lines 390-400, 410-414) now attempts to correlate PCA outliers (samples 12, 16, 21) with specific elevated biochemical parameters (glucose, creatinine, cholesterol). While speculative, this analysis adds value and is appropriate for the study stage.
  5. ROC Curves / Sensitivity / Specificity:
    • Status: Addressed. The authors correctly state the sample size is insufficient for these metrics (Author Reply). The manuscript appropriately avoids making definitive diagnostic claims, framing the potential for screening (Lines 425-429). This limitation should still be explicitly mentioned in the Discussion.
  6. HC-MOW Manufacturing Details:
    • Status: Addressed. Glass type (Ar-Glass, Schott) is specified (Line 144), and dimensions are labeled in Figure 1.
  7. Signal-to-Noise Ratio (SNR) / Detection Limits (LOD):
    • Status: Weakly addressed. Lines 159-161 mention ~1% noise relative to total intensity. This is minimal information. A more formal discussion of SNR or estimated LOD (even if preliminary) would strengthen the paper's claims about sensitivity. The current statement is insufficient.
  8. Dilution Protocol:
    • Status: Addressed. Saline concentration (0.9%) and preparation are now specified (Lines 139-140).
  9. Variance Explained by PCs:
    • Status: Addressed. Percentages are provided in the text (Lines 287, 290, 292) and Figure 8 caption.
  10. Wavelength Range Justification:
    • Status: Not Addressed in Manuscript. The authors provided a valid justification in their response (comb maxima location, gentle range) but did not add this rationale to the manuscript text (e.g., in Methods 2.3 or Discussion). This should be included.
  11. Scree Plot / Component Loadings:
    • Status: Addressed. Figure 8a (scree plot) and 8b (loading plot) are included and discussed.
  12. Double Smoothing:
    • Status: Addressed. The authors acknowledge the risk and provide justification (Lines 275-280) for the empirical selection of the filter window width (29 points) to balance noise reduction and signal integrity.

Evaluation of Responses to "Other Recommendations":

  • Novelty: Marginally addressed. Intro positions the work but could state the specific novel contribution more sharply.
  • Comparison Table: N/A (Acceptable justification for POC stage).
  • Ethical Docs: Addressed (Statement included).
  • Fabrication Details: Marginally addressed (Standard process cited, specific parameters omitted).
  • Spectral Resolution: Addressed (0.25 nm stated, Line 161).
  • Savitzky-Golay Validation: Addressed (Justification added, Lines 275-280).
  • Raw vs Processed Spectra: Not Addressed (Check if supplementary material provided).
  • Error Bars/Ellipses on PCA: Addressed (Confidence ellipses added to Fig 9, mentioned Line 301).
  • Statistical Comparison MOW vs Photometry: Addressed (Fig 4, Lines 245-252).
  • Clinical Translation Challenges (Hemolysis): Weakly addressed. Discussion touches on hemoglobin/lipemia (Lines 343, 338-342) but doesn't explicitly frame common issues like hemolysis as key challenges/interferents needing investigation.
  • PCA Limitations for Diagnosis: Addressed (Discussion clarifies PCA role, mentions future ML plans, Lines 370-373, 374-383).
  • Cost-Benefit: N/A (Acceptable justification for POC stage).
  • Fig 1 Scale Bar: Not Addressed (Lacks a graphical scale bar).
  • Fig 2 Labels: Addressed (Components labeled/captioned; specs in text).
  • Table 1 Units: N/A (Comment was likely mistaken).
  • Machine Learning: Addressed (Mentioned as future work, Lines 370-373).
  • Sterilization/Reusability: Addressed (Single use stated, Lines 148-149).
  • Comparison w/ Similar Sensors: N/A (Acceptable justification for POC stage).

Recommendation:

Major Revision

While significantly improved, the manuscript still requires major revisions to adequately address the remaining weaknesses before it can be considered for publication.

Key Revisions Needed:

  1. Strengthen Limitations Discussion: Explicitly state the limitations imposed by the small sample size (n=21), homogeneous sample characteristics (female-only, specific age range), and lack of detailed inclusion/exclusion criteria. Discuss how these factors limit statistical power, prevent robust subgroup analysis, and restrict the generalizability of the current findings. Frame the "proof-of-concept" nature clearly within these constraints.
  2. Justify Methodological Choices in Text: Add the explicit rationale for using a female-only sample (e.g., standardization for this initial study) and the selected wavelength range (e.g., alignment with waveguide comb maxima, minimizing potential photodamage) directly into the manuscript's Methods or Discussion sections.
  3. Clarify Participant Groups: Provide more specific details on the inclusion/exclusion criteria for both the "healthy" and "sick" groups in the Methods section (e.g., specific chronic diseases considered, any baseline screening for controls).
  4. Elaborate on Sensor Performance Metrics: While acknowledging the proof-of-concept stage, enhance the discussion on sensor performance. Provide a more quantitative discussion, estimation, or at least a qualitative assessment of key metrics like Limit of Detection (LOD), sensitivity, and Signal-to-Noise Ratio (SNR). Authors could consult works like Lotfiani et al. (2022) [Journal of Lightwave Technology, 40(4), pp.1231-1237.], Jahromi & Lotfiani (2022) [IEEE Sensors Journal, 22(21), pp.20430-20437.], and Dehdashti Jahrom (2024) [Optics Express, 32(24), pp.43475-43489.] for methodologies and examples of how these parameters are characterized and discussed for optical/photonic sensors, which might inform future characterization of their HC-MOW system.
  5. Address Potential Clinical Interferents: Explicitly discuss potential interference from common clinical sample variables (e.g., hemolysis, varying levels of lipemia beyond simple visual assessment, specific medications) as significant challenges for practical point-of-care application that warrant targeted investigation in future studies.
  6. Add Scale Bar: Include a graphical scale bar on the cross-sectional image in Figure 1.
  7. Enhance Novelty and Contextualization: More clearly articulate the specific novelty of using this particular HC-MOW configuration and PCA analysis for undiluted serum screening compared to previous HC-MOW work or other standard techniques. Briefly contrasting the approach or potential advantages/disadvantages with other emerging sensor platforms (e.g., advanced plasmonic sensors described in [Journal of Lightwave Technology, 40(4), pp.1231-1237] or THz sensors like in Khan et al. (2023) [Applied Sciences, 13(9), p.5784.]) could help position the work within the broader field.
  8. Expand on Future AI/ML Integration: Elaborate meaningfully on the planned use of AI/ML briefly mentioned in the discussion (Lines 370-373). Suggesting specific approaches beyond generic terms would be beneficial. Referencing methodologies for AI-aided sensor modeling or spectral analysis, such as those explored in Hamedi et al. (2023) [Engineering Applications of Artificial Intelligence, 118, p.105646.], Hamedi & Jahromi (2021) [Expert Systems with Applications, 178, p.115029.], or Jahromi & Hamedi (2021) [Materials Research Bulletin, 141, p.111371], could demonstrate a clearer path forward for leveraging AI to potentially improve classification accuracy, handle complex spectral features, and move beyond basic PCA in subsequent research.
  9. Provide Supplementary Data: Ensure the raw and processed spectra are readily available as supplementary material, as indicated in the author responses.

 

Addressing these points will significantly strengthen the manuscript's rigor, transparency, and appropriate framing of the findings within the context of a proof-of-concept study.

Author Response

Point-by-Point Evaluation Based on Previous Comments and Author Replies:

  1. Small sample size (n=21) / No power calculation:
  • Status:Addressed partially. The authors correctly state power calculations are typically done a priori and justify the current work as "proof of concept" (Lines 118-121). However, the manuscript should more strongly emphasize in the Discussion/Limitations section that the small sample size limits statistical power, prevents robust subgroup analysis, and restricts the generalizability of the findings. The "proof of concept" nature needs to be framed with these caveats.

Thank you for the comment, added information on line 467 -472

    1. All subjects female (35–45 years):
  • Status:Addressed partially. The sample characteristics are stated (Lines 127-128). The authors' reply mentions standardization as the reason, but this justification is not included in the manuscript text. This rationale should be added to the Methods, and the resulting significant limitation on generalizability must be clearly stated in the Discussion/Limitations.

In this study, when forming groups, we started from a group of healthy women, since we had access to samples from the IVF clinic. For this reason, the group of patients was also recruited of one (female) gender. Of course, it is interesting to compare these data with men and groups of other ages. The value of the method is that we can add data to the study, thus forming a database for further training of artificial intelligence.

Line 129 - 149

Line 467 - 472

    1. Inclusion/Exclusion Criteria & Ethical Approval:
  • Status:Addressed partially. The ethics statement is now present (Lines 445-447). However, the description of participant groups (Lines 127-132) still lacks formal inclusion/exclusion criteria beyond "healthy women preparing for IVF" vs. "patients of a cardiology clinic with chronic diseases". What specific diseases? Were controls screened for underlying conditions? This vagueness remains a weakness.

In this study, a comprehensive blood test and a full examination were not conducted. Only some biochemical parameters were studied, which we considered essential for them to have an impact on the blood serum spectra.

Healthy women did not have chronic diseases, as this is a mandatory condition for receiving a quota from the state for the IVF procedure. Before undergoing further procedures, they underwent an initial examination and questionnaire according to the standard procedure. The list of questions includes, among other things, items on the presence of chronic diseases and the drugs used.

For the "patients" group, there were hospital cards. The patients had different diagnoses and all of them were registered in a cardiology clinic, having a history of cardiological problems at various stages of treatment.

Some details of the selection were covered in the materials and methods section.

Line 129 - 149

Line 467 - 472

  1. Linking PCA to Biomarkers/Outcomes:
  • Status: The Discussion (Lines 390-400, 410-414) now attempts to correlate PCA outliers (samples 12, 16, 21) with specific elevated biochemical parameters (glucose, creatinine, cholesterol). While speculative, this analysis adds value and is appropriate for the study stage.
    1. ROC Curves / Sensitivity / Specificity:
  • Status: The authors correctly state the sample size is insufficient for these metrics (Author Reply). The manuscript appropriately avoids making definitive diagnostic claims, framing the potential for screening (Lines 425-429). This limitation should still be explicitly mentioned in the Discussion.
  • Line 466-471
    1. HC-MOW Manufacturing Details:
  • Status: Glass type (Ar-Glass, Schott) is specified (Line 144), and dimensions are labeled in Figure 1.
    1. Signal-to-Noise Ratio (SNR) / Detection Limits (LOD):
  • Status:Weakly addressed. Lines 159-161 mention ~1% noise relative to total intensity. This is minimal information. A more formal discussion of SNR or estimated LOD (even if preliminary) would strengthen the paper's claims about sensitivity. The current statement is insufficient.

Detection limits and sensitivity have been investigated in earlier publications. Added a reference to these data on the lines 380-384

 

  1. Dilution Protocol:
  • Status: Saline concentration (0.9%) and preparation are now specified (Lines 139-140).
    1. Variance Explained by PCs:
  • Status: Percentages are provided in the text (Lines 287, 290, 292) and Figure 8 caption.
    1. Wavelength Range Justification:
  • Status:Not Addressed in Manuscript. The authors provided a valid justification in their response (comb maxima location, gentle range) but did not add this rationale to the manuscript text (e.g., in Methods 2.3 or Discussion). This should be included.
  • Added substantiation on the lines 192-197
    1. Scree Plot / Component Loadings:
  • Status: Figure 8a (scree plot) and 8b (loading plot) are included and discussed.
    1. Double Smoothing:
  • Status: The authors acknowledge the risk and provide justification (Lines 275-280) for the empirical selection of the filter window width (29 points) to balance noise reduction and signal integrity.

Evaluation of Responses to "Other Recommendations":

  • Novelty:Marginally addressed. Intro positions the work but could state the specific novel contribution more sharply.
  • Comparison Table:N/A (Acceptable justification for POC stage).
  • Ethical Docs:Addressed (Statement included).
  • Fabrication Details:Marginally addressed (Standard process cited, specific parameters omitted).
  • Spectral Resolution:Addressed (0.25 nm stated, Line 161).
  • Savitzky-Golay Validation:Addressed (Justification added, Lines 275-280).
  • Raw vs Processed Spectra:Not Addressed (Check if supplementary material provided).
  • Error Bars/Ellipses on PCA:Addressed (Confidence ellipses added to Fig 9, mentioned Line 301).
  • Statistical Comparison MOW vs Photometry:Addressed (Fig 4, Lines 245-252).
  • Clinical Translation Challenges (Hemolysis):Weakly addressed. Discussion touches on hemoglobin/lipemia (Lines 343, 338-342) but doesn't explicitly frame common issues like hemolysis as key challenges/interferents needing investigation.

Line 411-415

  • PCA Limitations for Diagnosis:Addressed (Discussion clarifies PCA role, mentions future ML plans, Lines 370-373, 374-383).
  • Cost-Benefit:N/A (Acceptable justification for POC stage).
  • Fig 1 Scale Bar:Not Addressed (Lacks a graphical scale bar).
  • Fixed
  • Fig 2 Labels:Addressed (Components labeled/captioned; specs in text).
  • Fixed
  • Table 1 Units:N/A (Comment was likely mistaken).
  • Machine Learning:Addressed (Mentioned as future work, Lines 370-373).
  • Sterilization/Reusability:Addressed (Single use stated, Lines 148-149).
  • Comparison w/ Similar Sensors:N/A (Acceptable justification for POC stage).

Recommendation:

Major Revision

While significantly improved, the manuscript still requires major revisions to adequately address the remaining weaknesses before it can be considered for publication.

Key Revisions Needed:

  1. Strengthen Limitations Discussion:Explicitly state the limitations imposed by the small sample size (n=21), homogeneous sample characteristics (female-only, specific age range), and lack of detailed inclusion/exclusion criteria. Discuss how these factors limit statistical power, prevent robust subgroup analysis, and restrict the generalizability of the current findings. Frame the "proof-of-concept" nature clearly within these constraints.

 

Thank you for your important comment, we added information on Line 444-471 

 

  1. Justify Methodological Choices in Text:Add the explicit rationale for using a female-only sample (e.g., standardization for this initial study) and the selected wavelength range (e.g., alignment with waveguide comb maxima, minimizing potential photodamage) directly into the manuscript's Methods or Discussion sections.

Thank you for your important comment, we added information on Line 129-149, 444-471 

 

  1. Clarify Participant Groups:Provide more specific details on the inclusion/exclusion criteria for both the "healthy" and "sick" groups in the Methods section (e.g., specific chronic diseases considered, any baseline screening for controls).

Thank you for your important comment, we added information on Line 129-149, 444-471 

 

  1. Elaborate on Sensor Performance Metrics:While acknowledging the proof-of-concept stage, enhance the discussion on sensor performance. Provide a more quantitative discussion, estimation, or at least a qualitative assessment of key metrics like Limit of Detection (LOD), sensitivity, and Signal-to-Noise Ratio (SNR). Authors could consult works like Lotfiani et al. (2022) [Journal of Lightwave Technology, 40(4), pp.1231-1237.], Jahromi & Lotfiani (2022) [IEEE Sensors Journal, 22(21), pp.20430-20437.], and Dehdashti Jahrom (2024) [Optics Express, 32(24), pp.43475-43489.] for methodologies and examples of how these parameters are characterized and discussed for optical/photonic sensors, which might inform future characterization of their HC-MOW system.

Detection limits and sensitivity have been investigated in earlier publications. Added a reference to these data on the lines 370-390.

In this paper, we did not provide calculations of sensitivity and specificity, since they are possible with an accurate and objective determination of true positive and true negative samples. In this case, we have a system with many unknown parameters, so the calculation of sensitivity and specificity is not in our favor.

 

  1. Address Potential Clinical Interferents:Explicitly discuss potential interference from common clinical sample variables (e.g., hemolysis, varying levels of lipemia beyond simple visual assessment, specific medications) as significant challenges for practical point-of-care application that warrant targeted investigation in future studies.

 

Thank you for pointing this out. We have accordingly modified information on Line 153-158 and 411-415

 

  1. Add Scale Bar:Include a graphical scale bar on the cross-sectional image in Figure 1.

 

Fixed

 

  1. Enhance Novelty and Contextualization:More clearly articulate the specific novelty of using this particular HC-MOW configuration and PCA analysis for undiluted serum screening compared to previous HC-MOW work or other standard techniques. Briefly contrasting the approach or potential advantages/disadvantages with other emerging sensor platforms (e.g., advanced plasmonic sensors described in [Journal of Lightwave Technology, 40(4), pp.1231-1237] or THz sensors like in Khan et al. (2023) [Applied Sciences, 13(9), p.5784.]) could help position the work within the broader field.

 

Thank you for the comment, in the article we point out that “Experiments with whole and diluted serum have shown that dilution of blood serum does not affect the determination of concentrations of substances in these solutions, since all parameters of the particle content in serum change proportionally” [Line 364-367].

 

  1. Expand on Future AI/ML Integration:Elaborate meaningfully on the planned use of AI/ML briefly mentioned in the discussion (Lines 370-373). Suggesting specific approaches beyond generic terms would be beneficial. Referencing methodologies for AI-aided sensor modeling or spectral analysis, such as those explored in Hamedi et al. (2023) [Engineering Applications of Artificial Intelligence, 118, p.105646.], Hamedi & Jahromi (2021) [Expert Systems with Applications, 178, p.115029.], or Jahromi & Hamedi (2021) [Materials Research Bulletin, 141, p.111371], could demonstrate a clearer path forward for leveraging AI to potentially improve classification accuracy, handle complex spectral features, and move beyond basic PCA in subsequent research.

 

Added reasoning for further integration of ML/AI on the lines 453-465

 

  1. Provide Supplementary Data:Ensure the raw and processed spectra are readily available as supplementary material, as indicated in the author responses.

 

Files were attached.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

In this revision, the authors addressed the concerns in the previous comments. The PCA algorithm seem to work well to differentiate the two group. This manuscript is good for publication. 

Author Response

The team of authors expresses deep gratitude to the reviewers for their attentive attitude to our work and valuable comments.

Back to TopTop