Transcranial Doppler-Based Neurofeedback to Improve Hemispheric Lateralization

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Review of “ Transcranial-Doppler-based Neurofeedback to improve hemispheric lateralization”.

The authors aim to explore and validate TCD-based neurofeedback as a potential method for training and enhancing the control of brain activity lateralization. Ultimately, they seek to assess its applicability in the fields of rehabilitation and BCIs.  

MAJOR COMMENTS

The topic is very interesting, and the manuscript is generally well written and well organized.

#1# However, I recommend that the authors more clearly articulate the novelty of their study in comparison to previous TCD-based neurofeedback research. Specifically, they should address in the introduction:

(1) what is novel about their approach?

(2) how does it differ from prior work?

(3) why are these differences important and impactful?

While this information is presented in the discussion section (lines 322-341 and 360-364), it would strengthen the manuscript to present it earlier (ideally in the introduction) as part of the motivation and objectives of the study.

#2# The statistical analysis section requires clarification, particularly in terms of the goal and approach used.

Line 179-190: It is not clear what was compared and how the statistical tests were applied. (i) Did you intend to compare all four sessions to assess whether there was a difference between left and right conditions overall?

(ii) Were the curves averaged across all subjects before comparison (of two traces) as implied in the first sentence?

(iii) Or were the values averaged per session and subject and then statistically tested?

To properly infer lateralization effects at the group level, the standard approach would be to use session-level values per subject and evaluate those across sessions. A repeated-measures ANOVA or a non-parametric equivalent (e.g., the Friedman test) would be appropriate to analyze the overall lateralization index (LI) and the effect of the session.

Additionally, it seems that the methods described (lines 179-190) might relate to the generation of Figure 2. If  so, please clarify the following steps explicitly in the manuscript:

 (1) the first sentence in the paragraph aims to explain how data were managed prior to the visualization in Figure 2, which shows the mean +/- std over all the subjects per session for the left and right conditions.

(2) That you used a Wilcoxon signed-rank test (paired samples) at each time point to assess statistical differences—comparing, at each point, the 14 paired values for Left vs. Right conditions.

Line 211-215: The statistical approach used to compare results across the four sessions needs to be reconsidered. Since the same participants are measured in all four sessions, the data represent related samples. Therefore, an appropriate analysis would be a repeated-measures ANOVA or its non-parametric equivalent, the Friedman test, along with the corresponding post-hoc tests if significant differences are found.

The Kruskal-Wallis test (and its parametric counterpart, one-way ANOVA) assumes independent samples, which does not apply to your within-subjects design.

Please revise the statistical analysis accordingly and clarify the methods used to ensure they match the structure of your data.

 

MINOR COMMENTS

Line 178:  correct the sentence “This ais was approached in different ways.”

Author Response

Please see attached file

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

• Please remove the caption: "Highlighted Application: Authors are encouraged to provide a concise description of the specific or potential application of the work. This section is not mandatory."
• Add hyperlinks to tables, figures, references, and equations.
• Add a brief description between section and subsection titles.
• At the end of the introduction, add a short paragraph describing the manuscript's structure.
• In the introduction section, add the main contributions of the work to the state of the art.
• The authors mention that Fourteen Healthy was involved but do not clearly and thoroughly describe the inclusion and exclusion criteria. The term "healthy" is ambiguous.
• There are some typographical errors. For example, some figure captions have periods at the end, but others do not. Please review the spelling.
• The quality of the images needs to be improved.
• The title of Figure 2 is not correctly positioned because there is text between the title and the figure.
• A control group (without neurofeedback or with sham feedback) was not included. This makes it impossible to distinguish the effects of specific training from a possible placebo effect or task familiarization. Please justify.
• Although it is stated that the left and right signals are given in random order, it is not clear whether the task order is completely counterbalanced. This can lead to order bias or uncontrolled learning.
• It is not specified whether the researchers analyzing the data were blinded to the sessions or conditions, which is essential to reduce interpretation bias.
• Although the lateralization formula is clear, the signal processing is not detailed with sufficient precision: What type of filtering was applied? How are motion artifacts handled?
• The algorithm for calculating the optimal threshold (Youden Index) is not detailed. This is relevant for replicability and for interpreting the classifier's actual performance.
• A "delay of several seconds" in the feedback is mentioned, but how it affects the interpretation of the subject's performance is not quantified or analyzed.
• There is no justification for why the training lasts 15 minutes or whether this was optimal or arbitrary. There is also no analysis of whether a longer duration impacts performance.
• Wilcoxon is used only at individual points in the time curves without correction for multiple comparisons. This increases the risk of false positives.
• Improvements in AUC from 0.55 to 0.69 and precision from 0.51 to 0.65 are reported, but there is no discussion of whether these values ​​are clinically relevant or statistically significant. No confidence intervals are included.
• Although the ITR (0.084 bits/min) is calculated, this is very low for real-world BCI applications. This should be mentioned as a serious limitation.
• No important individual differences that could arise from the data are discussed, nor is there any analysis of whether some subjects do not respond to training.
• Implications for post-stroke neurorehabilitation are mentioned without any evidence in patients or measures of functional transfer, which is speculative.
• The possible effects of fatigue or habituation are not discussed. Four sessions on different days may introduce habituation or motivational effects that were not controlled for or assessed.
• Most of the references are old. Although they are relevant to the topic of study, they require updating.
• A deeper analysis of the figures and images presented in the results is required

Author Response

Please see attached file

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

My comments were addressed, and the work can be considered for publication.