Evaluation of Morlet Wavelet Analysis for Artifact Detection in Low-Frequency Commercial Near-Infrared Spectroscopy Systems

Round 1

Reviewer 1 Report

Non-invasive Near-Infrared Spectroscopy (NIRS) equipment is used to test brain tissue oxygenation using a technique called regional cerebral oxygen saturation (rSO2). One of the main drawbacks is the presence of artifacts in recorded signals. The manual removal of these artifacts requires a lot of time and resources. The aim of the study was to assess the suitability of wavelet analysis as an automated technique for the straightforward removal of signal loss artifacts from rSO2 data acquired from devices that are sold commercially. Using the current populations of patients with traumatic brain injury (TBI), patients undergoing elective spine surgery (SP), and healthy controls (HC), a retrospective observational analysis was carried out. 

For each patient, data on arterial blood pressure (ABP) and rSO2 were gathered. Using wavelet coefficients and coherence to identify signal loss artifacts in rSO2 signals, wavelet analysis was found to be effective in eliminating basic signal loss errors. In the HC, SP, and TBI populations, the removal success rates were 100%, 99.8%, and 99.7%, respectively (although the precision in pinpointing the precise moment was somewhat compromised). Future research is required, although wavelets may prove helpful in a layered approach NIRS signal artifact tool employing higher frequency data, even if there were little advantages to employing wavelet analysis over thresholding techniques.

Although the authors provide a method, seems not novel. Application of existing method do not significantly contribute to the body of knowledge. More explaination of the work is needed. 

It seems tha the groups are  not balanced. Hwo the authors tackle this. Please clarify. 

Which statistical tool was applied? Whether the imbalanced data incorporated in this? how the p values are calculated. 

It is not clear, how the number of subjects was calculated for such a study.

Demonstration of the method's validity is not necessarily required to have many subjects. Moreover the title is misleading. Please change it accordingly the work performed. 

The authors must balance in referencing the articles in terms of Gender/area/expertise/countrywise. 

Author Response

Please see attachment. 

Author Response File: Author Response.docx

Reviewer 2 Report

Some concerns in this study.

1. The motivation of this study is unclear. Why this evaluation is needed.

2. Why is the motivation of using wavelet analysis.

3. The implementation details of the wavelet should be clarified.

4. Please highlight the key conclusion of this study.

5. Please well summarize the main contributions of this study.

none

Author Response

Please see the attachment. 

Author Response File: Author Response.docx

Reviewer 3 Report

This paper presents an evaluation of wavelet analysis for artifact detection in commercial low-frequency near-infrared spectroscopy systems.

 The article presents an interesting and current topic of the application of wavelet analysis. The authors used appropriate methodology and tools to conduct experiments and analyze data. This paper does not provide a sufficient theoretical introduction to wavelet analysis and its applications in NIRS. It does not clearly state the purpose and research hypotheses, nor does he formulate conclusions and practical implications of his work. The effectiveness of the solution and the thesis depends largely on the selection of criteria and the appropriate data selection.

 Minor remarks:

1) I propose to describe in more detail on what basis and what results from the scope of selection of input data for the research problem.

2) The authors could refer more broadly to other research and justify the choice of the presented solution.

3) What features of the presented model are an original approach in the studied discipline? I propose to formulate the purpose and research hypotheses more precisely.

4) The choice of wavelet analysis parameters, such as the type of wavelet, decomposition level, threshold and thresholding method, could be further justified.

5) The article should contain more precise information about the theory and principles of wavelet analysis, its advantages and limitations, and its relationship to NIRS.

Author Response

Please see the attachment. 

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The authors have addressed comments.

Reviewer 2 Report

No further question.

Minor refinement