Scientific Papers

Decoding force production of skeletal muscle from the female brain using functional near-infrared spectroscopy | BMC Research Notes

Description of Image

DGIST Ethics Committee approved this human study (DGIST_180202_HR_-001-01). We conducted human experiments according to the Declaration of Helsinki. Eight right-handed healthy female adults (age 21 ± 1.7 years; weight 53.5 ± 5.4 kg; height 160.5 ± 3.3 cm) were recruited who did not have any neurological, physical, or psychiatric disease history. The sample size was estimated using the IBM SPSS Statistics version 27 (IBM Corp., Armonk, N.Y., USA) with the following conditions: one-sample t-test, power of 0.8, population mean of 5, population standard deviation of 5, and one-side analysis with significance level of 0.05. The subject group was relatively homogeneous and selected with no bias. We gave all subjects written informed consent before participating in the experiment. The experimental procedures and data analysis used for this study have been fully addressed in our previous study for males [13].

Briefly, the subjects lay on a customized bed that allowed their body to be fully supported except for their wrist and hand, isolating the wrist flexor muscles. A block design was applied to each trial, comprising three stages of rest (30 s), task (30 s), and rest (30 s). The subject conducted the same trial three times in a row over one session. A session was begun after 20 s rest to stabilize the hemodynamic signals. During the stage of motor task, the subject holds a specific load on her right palm while keeping the wrist straight, ensuring the isometric condition. The motor task was started and ended by having their right forearm touched. We determined the wrist flexor muscles’ maximal voluntary contraction (MVC) by increasing the load until the subject could not maintain it. Three sessions were carried out for each subject at 0%, 50%, and 100% MVC. At least 5 min was taken for muscle relaxation between sessions to avoid influences of fatigue and injury. During the experiment, the subjects were blindfolded and earplugged to minimize reactions evoked by visual and auditory stimuli.

We used a commercially available fNIRS system (FOIRE-3000, Shimadzu Co., Kyoto, Japan) with the default settings for wavelength (780, 805, and 830 nm) and sampling (30.303 Hz) over the left-brain hemisphere. The oxyhemoglobin concentration change (∆Hboxy) was estimated by the modified Beer-Lambert law with the default coefficient values set in the fNIRS system. The left hemisphere’s primary somatosensory and motor cortex were covered by a 5-by-4 array of twenty optodes (ten transmitters and ten receivers) (see Fig. 1A for graphical illustration). Individual optodes were placed in respective holders apart by 3 cm on a commercially available elastic cap, including a chin strap. The 3-dimensional coordinates of individual optodes were measured after the experiments using a 3-dimensional digitizing system (FASTRAK, Polhemus, VT, USA) and projected over the brain image rendered in the 3-dimensional space using the NIRS-SPM software (version NIRS-SPM_V4_r1 and spm8) [18]. The Brodmann areas (MRIcro) optically measured over the cerebral cortex were statistically estimated using the NIRS-SPM software. All channels related to the primary somatosensory and motor cortex were analyzed.

Fig. 1
figure 1

Identification of the most activated area on the primary sensorimotor cortex during maximal contraction of the wrist flexor muscles under isometric conditions. (A) Arrangement of optodes and channels over the head. Nz, Cz, AL, AR, and Iz indicate the nasion, central point, left preauricular point, right preauricular point, and inion, respectively. (B) Anatomical locations of channels in the left hemisphere. The green and red areas indicate the primary motor and somatosensory regions. Circled numbers indicate channels correlated with the primary somatosensory and motor areas. (C) Statistical group analysis for cortical activation across all subjects. A color code indicating a higher T value with a brighter color was applied to represent the relative cortical activation during the motor task. The most activated cortical area was identified during the motor task at 100% MVC (uncorrected p-value < 0.05). The dashed circle indicates the cortical area maximally activated during the motor task in males. This data was adopted from the previous study (Fig. 4 in [13])

Raw hemodynamic signal data were measured three times in a session and averaged using the LABNIRS system. The averaged data were preprocessed, including detrending with a discrete cosine transformation and bandpass filtering with a cutoff frequency of 1/128 Hz for high-pass filtering and a hemodynamic response function for low-pass filtering using the NIRS-SPM software. A linear baseline correction was made to remove longitudinal signal drift. Consequently, individual time series of filtered data were shifted such that their values at task onset were set to zero. We further normalized the corrected signal for unbiased comparison between subjects and channels. The normalization was conducted by dividing the corrected signal by the standard deviation calculated for 10 s before the initiation of motor task.

The cortical areas activated during the motor task were statistically identified using the functions (i.e., general linear model and continuous random field) built in the NIRS-SPM software. The changes in hemodynamic signal from the maximally activated cortical area were analyzed to predict the level of voluntary muscle contraction. As reported in the previous study [13], the hemodynamic signal trajectory was characterized by eight predictors. Four predictors represented the magnitude of oxygenated hemoglobin concentration (i.e., P1-P4 in Fig. 2A and D, respectively). Others described the variation in oxygenated hemoglobin concentration over time (i.e., P5-P8 in Fig. 2E H, respectively). The predictor values obtained for each subject were averaged across all subjects for 0%, 50%, and 100% MVC, respectively.

Fig. 2
figure 2

Correlation between the trajectory predictor and muscle force production. A-H. Each trajectory predictor for the ∆Hboxy signal from channel 21 was plotted against the force production varying from 0 to 100% MVC under isometric conditions. The trajectory predictor values measured from individual subjects were averaged for all subjects. The mean values of individual predictors were indicated with solid dots (black) and fitted to the linear regression line indicated by dotted lines (black). The regression equation and r2 value for each predictor are 0.1*x + 3.34 and 0.99 for P1, 2.52*x-23.53 and 0.99 for P2, 0.08*x-0.78 and 0.99 for P3, 0.11*x-0.92 and 0.99 for P4, 0.1*x + 41.76 and 0.61 for P5, 0.05*x + 35.77 and 0.7 for P6, 0.06*x + 47.69 and 0.75 for P7, and 0.01*x + 0.16 and 0.77 for P8, respectively, where x indicates the percentage of maximal muscle force

We assessed how the predictor’s grand mean represented the voluntary contraction force by calculating the Pearson coefficient (r) and p-value (two-tailed test) using the IBM SPSS Statistics version 27. The trajectory predictor-voluntary contraction relationship was fitted with a linear regression line, and the r2 value was used to compare the goodness of fit under the SPSS software environment. The statistical significance was determined by p-value < 0.05. The data from our previous study for males [13] were adapted for the purpose of comparison in this study.

Description of Image

Source link