Optimal Heart Sound Segmentation Algorithm Based on K-Mean Clustering and Wavelet Transform
, Xingguang Geng 1,2, Zhixing Gao 1,2, Hao Yang 1,2, Zhiwei Dai 1,2 and Haiying Zhang 1,2,*
Round 1
Reviewer 1 Report
This paper proposes a new optimal heart sound segmentation algorithm based on k-mean clustering and haar wavelet transform to provide important information for heart state detection and identification. It first uses the Viola integral combined with the Shannon energy algorithm to extract the heart sound envelope energy function It then extracts the time-frequency domain features from different dimensions. The k-mean clustering and haar wavelet transform is used to localize the heart sound in the time domain.
In general, the topic conducted in the paper is interesting. The paper is well-written and organized. The extra point of this paper is to apply data mining in a real-life application. To further improve the quality of the paper, the authors can consider the following points.
- In the Introduction, highlight the main contributions of the paper. Introduce the mechanism for using k-means in the clustering step.
- In section 2, insert a table of notations used in the paper. Some mathematical notations used in the paper are not in a good format. Make their sizes as normal text's font size.
- In section 2, theoretically discuss the complexity of the proposed framework.
- In section 3, show the standard deviation (+-) for the clustering results shown in tables 1, 2.
- In section 4, discuss other possible methods that can perform the clustering tasks with missing points, which is a problem in this research. the authors can refer to these works in the discussion https://doi.org/10.1016/j.ins.2021.04.076 and https://doi.org/10.3390/sym14010060
-
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
The paper provides an optimized segmentation approach for the heart sound based on wavelet transform and k-mean clustering. The authors first extract the time-frequency domain features from the heart sound envelope energy function using viola integration and Shannon energy. Next, they localize the heart sound's S1 and S2 using the k-mean clustering and Haar wavelet technique.
The presentation is of very poor quality, which obscures the importance of the work's content and its soundness.
Because of the English's poor quality, it is exceedingly challenging to understand the research work. The English language needs to be thoroughly edited.
The majority of the figure captions need to be rephrased and made more informative.
The lower computing complexity is mentioned. However, the article does not display the computational cost result.
Instead of being fast, normal, or slower in Table 2, the computational complexity should be either low or high.
Since none of the figures have legends, it is hard to comprehend the findings from them.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
The paper has been improved based on my comments. I have checked and verified that the current version reaches the level of acceptance.
