Analysis and Recognition of Human Lower Limb Motions Based on Electromyography (EMG) Signals

Round 1

Reviewer 1 Report

Comment 1: Based on current write-up, the contribution of this paper is limited. In Section 1, the author emphasized 3 main contributions: (1) 4 muscles that were both exercise related and less relevant were selected among 73 numerous lower extremity muscles by using OpenSim and SPSS software. (2) Lower limb angulation and EMG signals at different road slopes, different gaits, 75 and different movements were collect-ed and analyzed by IMU and EMG sensors. (3) The feasibility of identifying road slope, gait and lower limb movement with the 77 features of EMG signal is verified by using double hidden layer BP neural network.

All three items are not contributions, but experiment details. They could potentially be contributions if the author could clearly justify why they are important and original. Take item 3 example, there exist many studies able to identify locomotion via EMG, what would be the original contribution of this paper to the field? 

Comment 2: The experiments of this paper did not compare the proposed method with any state-of-the-art approaches. Even if it is a feasibility study, it is not convincing because many existing studies have demonstrated the feasibility. Additionally, the experiments did not consider/discuss the difference between amputees and able-bodied subjects. 

Comment 3: The writing of this paper needs improvement. Some descriptions are confusing, informal, not informative or has error in grammar. A few examples, "4 parts of muscles in lower limb", "Select 4 movements (squat, lunge, raise leg and stand up)" in the conclusion section. There also exist a decent number of typos, only a few examples:  "Be-fore", "ex-perimenters".

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors submitted a well prepared and interesting manuscript. However, the manuscript requires minor revision before acceptance. Namely, the architecture of the applied neural network is not clear from the manuscript. The number of nodes in different layers is not shown in Figure 4 and the text also does not report on it. Moreover, the curves of the training of NN is also not depicted in the manuscript. The applied optimisation algorithm is not mentioned or cited (ADAM, AdaGrad, etc.).

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The paper has been significantly improved after the revision. Besides the EMG based human lower limb motion recognition, the authors could consider discuss the benefit of EMG based methods over other approaches such as IMU based, computer vision-based methods.  

Potentially related references but not limited to: 

Zhang, Kuangen, et al. "A subvision system for enhancing the environmental adaptability of the powered transfemoral prosthesis." IEEE transactions on cybernetics 51.6 (2020): 3285-3297.

Zhong, Boxuan, et al. "Environmental context prediction for lower limb prostheses with uncertainty quantification." IEEE Transactions on Automation Science and Engineering 18.2 (2020): 458-470.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf