Subjects
In this study, twelve subjects (five females and seven males) with normal swallowing capabilities were recruited. All subjects were examined by an occupational therapist to make sure that they had no history of dysphagia or any other medical problems that might affect swallowing. The mean age of the subjects was 27 years with a range of 23–38. The protocol of this study was approved by the Institutional Review Board, and the procedures were performed in accordance with the ethical standards of the committee on human experimentation at the Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences. Before the study, the subjects received a complete explanation of the purpose, risks, and procedures of the investigation. All the subjects gave written informed consent and provided permission for publication of photographs with scientific and educational purposes.
HD sEMG signal acquisition
The HD sEMG signals were acquired from all the enrolled subjects. Firstly, each electrode was properly cleaned by using alcohol pad and then coated with a conductive gel to ensure good adhesiveness and conductivity between the electrode and the skin. Secondly, the skin surface where the HD sEMG electrodes located was carefully cleaned with alcohol swabs to remove dry dermis and skin oils that could degrade the quality of the sEMG recordings. Finally, a 2D electrode array made up of 96 channels was evenly aligned and placed on the front region of the neck (covering submental and infrahyoid muscle complexes) (Fig. 1) for recording muscular signals during various swallowing tasks.
In order to obtain comprehensive electromyography information, sufficient electrode number and small inter-electrode distance is necessary. To avoid the collision of two adjacent electrodes during swallowing tasks, a certain distance should be kept between neighboring electrodes. Correspondingly, the 96-channel electrode array structured in a 6 × 16 grid equally spaced with an interval of 15 mm between two adjacent rows (A–F) and columns (1–16) was selected so that the front neck could be fully covered for all subjects (Fig. 1). Each electrode has a circular silver-plated recording surface with an external diameter of 10 mm. The first row (Row A) of sixteen electrodes was placed about 10 mm below the lower jaw to facilitate the positioning of the electrode array, and the electrodes of A8 and A9 were placed around the center of the submental triangle for locating purpose. Then the other five rows were placed downward relative to Row A. In addition, a reference electrode was placed on the right wrist of each subject to serve as the ground for all electrodes. After the electrode array was placed properly, HD sEMG signals were filtered with a band-pass filter (10–500 Hz) and recorded simultaneously from all the 96 channels by the REFA 128-channel system (TMSi International, the Netherlands) at a sampling rate of 1024 Hz.
Experimental procedures
In the experiment of HD sEMG acquisition, a subject was seated on a comfortable chair and instructed to complete different swallowing tasks. Four different swallowing tests (Fig. 2) were adopted in the study to systemically evaluate the muscle activities of an entire swallowing procedure by using HD sEMG signals. For all the subject, sEMG were firstly acquired when they did not perform any swallowing action, which were used as a base line for sEMG normalization. Secondly, the subjects were asked to drink different volumes of water with a normal swallowing speed. The voluntary swallowing test had four successive tasks: swallowing saliva (dry swallowing) and drinking 5, 10, and 20 ml of water at a time, respectively. Thirdly, a swallowing test to drink different viscous materials was conducted, where each subject was asked to sequentially swallow 5 ml of bolus with different viscosities: water, thin sesame paste, and thick sesame paste. The thin sesame paste was made of 100 ml of water and 20 g of sesame powder, while the thick sesame paste was mixed by 100 ml of water and 35 g of sesame powder. During the first three swallowing tests, the subjects were required to keep their face forward and avoid any body movements. Finally, in order to examine the effect of head posture changes during swallowing on the sEMG recordings, the subjects were asked to retain their head at middle position, to right side, and to left side when drinking 5 ml of water, respectively. For all the four swallowing tests, each swallowing task was repeated for three times. When performing a swallowing task, subjects were required to hold the water or sesame bolus in their mouth for a while and then to get started swallowing normally. HD sEMG signals were recorded from 96 channels during the entire process of all the swallowing tasks and stored locally for further offline analysis.
Data analysis
sEMG energy maps
The HD sEMG signals acquired during different swallowing tasks were properly filtered by using digital filters to reduce the impacts of ECG artifacts picked up by the electrodes and other baseline variations caused by unintentional body movements of the subjects. Then the sEMG recordings from an entire swallowing process were segmented into a series of sEMG analysis windows with a window length of 100 ms for each channel. For each sEMG analysis window, its root-mean square (RMS) value was computed for all the 96 channels as follows, which represented the averaged power (energy) of EMG signals over the analysis window.
$$I_{i,j}^{a} = \sqrt {\frac{{\sum\nolimits_{n = 1}^{N} {X_{i,j}^{2} (a,n)} }}{N}} \;$$
(1)
where \(I_{i,j}^{a}\) is the RMS value of sEMG from the channel located at row i and column j of the 2D electrode array in a th analysis window, X
i,j
(a, n)is the sEMG amplitude of the channel (i, j) in a th analysis window, and N is the total number of sEMG samples N is the total number of samples in the analysis window. Therefore, a 2D RMS matrix with a dimension of 16 × 6 would be obtained for each of the analysis window. Each 16 × 6 RMS matrix was extended by using a spline interpolation algorithm and then the extended matrix was used to build a 2D sEMG energy color map, where the intensity values were represented by color tones scaled linearly between maximum (hot) and minimum (cold). By displaying the map slide by slide over time, these topographical maps would graphically demonstrate the dynamic electrophysiological activities of muscles in an entire swallowing process. The dynamic sEMG maps of the entire swallowing process were divided (by time) into five consecutive non-overlapped intervals, in accordance with the five physiological phases by other studies, so that the motions of swallowing structures of each time interval could be known according to the literature [38]. Then the RMS values within each interval were calculated and the energy map of each interval was projected onto the human neck area to exhibit the energy distribution of each swallowing phase.
Quantitative analysis
In the study, some quantitative indices were proposed to evaluate the electrophysiological characteristics of swallowing muscles. The 2D electrode array were equally divided into two sections from the center line of the neck: left-lateral electrodes and right-lateral electrodes, as shown in Fig. 3. Each section contained a total of 48 electrodes (8 × 6 array). The mean values of sEMG RMS were calculated by averaging the RMS values over all the electrodes (48 electrodes) of the left and right sections and over all the analysis windows in an entire swallowing process, respectively. Two-tailed t tests were used to test whether the mean LRER values averaged across all subjects were significantly different from 1.0 for different tasks, with a p > 0.05 indicating that there was no significant difference between the left and right sides in muscle activities. Two-tailed t-tests were also used to examine whether the averaged LRED values were significantly different from 0, with a p > 0.05 meaning nearly equivalent energy between the two sides. The mean RMS values were designated as mRMS
left
for the left electrode section and mRMS
right
for right electrode section. A quantitative index, named as the Left/Right Energy Ratio (LRER), was defined to measure the symmetric features of left and right neck muscles in deglutition as follows:
$$LRER = \;\frac{{mRMS_{left} }}{{mRMS_{right} }}$$
(2)
In addition, another quantitative index, called as Left/Right Energy Difference (LRED), was proposed and used to assess the symmetry of muscle activities on the left/right sides. LRED was defined as the absolute difference between the sum of left and right RMS values as follows.
$$LRED = \frac{{sum(RMS_{left} ) - sum(RMS_{right} )}}{{\frac{1}{2}sum(RMS_{N} )}} \times 100\%$$
(3)
where sum(RMS
left
) and sum(RMS
right
) was the summation of the RMS values obtained from the left and right EMG electrode section, respectively, and sum(RMS
N
) represented the summation of the RMS values of all the EMG electrodes.
