Test-retest reliability of brain activation induced by robotic passive hand movement synchronized with a Video game: A randomized crossover functional NIRS study

Published: 17 February 2021| Version 1 | DOI: 10.17632/vmg9jp4hgc.1
Contributors:
Sungjin Bae,
Jincheol Park,
Yong Seob Lim,
Pyung-Hun Chang

Description

Test-retest reliability (TRR) of brain activation, the likelihood that a certain brain area is repeatedly activated, is deeply related to brain plasticity, the basic paradigm of rehabilitation. In this study, we opted for a video game (VG) to promote TRR of robotic passive hand movement (RPHM), since our previous study has shown that RPHM has a low TRR. The two elements of VG, proprioception and real-time reaction, which we believe contribute to the interesting and immersive nature of VG, thereby enhancing the user engagement, a potent cause of contrasting TRR between the active and passive movements. To combine the VG with the RPHM, the pre-recorded video of VG is synchronized with the RPHM. The 40 subjects are being offered both RPHM and RPHM+VG, according to the crossover design, corresponding brain activations are measured by using fNIRS. Subsequently, TRR indices and the results of statistical analysis of crossover design are used to evaluate TRR. As a result, the introduction of VG to RPHM has shown enhanced activation in the primary sensorimotor area and premotor cortex as well as increasing TRR in those regions compared to RPHM. Current results suggest that RPHM with VG surpasses passive movement and is similar to active movement in terms of functional elements and the region of activation. Also, it can be inferred that VG is an appropriate modality of increasing TRR by enhancing user engagement when combined with passive movement. The 40 subjects were equally divided into and randomly assigned to two groups, Group A and Group B. The overall experiments take three months: one month for Period 1, followed by another month as a washout period, and still another month for Period 2. At the beginning and end of each period, a respective session of the experiment is conducted to determine the activation, and TRR is evaluated after the end-period session. The one month period was selected out of our previous research on TRR of brain activity. It is a part of the crossover design for one group to experiment RPHM and RPHM+VG in an order reverse to the other group does: While Group A carried out RPHM+VG in Period 1 (session1 and session2) before RPHM in Period 2 (session3 and session4), Group B RPHM in Period 1 before RPHM + VG in Period 2. There are two types of uploaded data as follows: first the relative changes in concentration of oxygenated hemoglobin, deoxygenated hemoglobin, and total hemoglobin; second, 3d coordinates of anatomical landmark and the placement of optodes. The file name is written in 5 characters consisted of one letter and four digits according to a set of rules. The first character is 'A' or 'B' which means the name of the group. And consecutive second and third digits means subject's number. The forth digit stands for the nubmer of sessions. The fifth digit refers to the kind of data that can be 0 or 1, the former refers to the 3d coordinates optodes and the latter refers to the the measured hemoglobin data.

Files

Steps to reproduce

The fNIRS data processing aims at obtaining t statistics of β, a brain activation index for verifying the significance of brain activation at the level of p < 0.05. The determination of that index is straightforward by following a series of standard procedures presented in NIRS-SPM, a MATLAB-based software package, which incorporates the following stages: first, mapping the measured channels to the standard brain; second, applying both a low pass filter and a high pass filter to eliminate noises; third, obtaining regression coefficient "β" based on the general linear model (GLM); finally obtaining t statistics of β]. When employing the filter, Gaussian smoothing with a full width at half maximum of two seconds as a low pass filter and wavelet-minimum description length detrending algorithm as a high pass filter. Besides, we incorporated the Lipschitz-Killing curvature-based Euler characteristic (EC) approach to control the familywise error rate resulting from multiple statistical hypothesis tests. With t statistics of β obtained through fNIRS data processing described above, we evaluated TRR in terms of Rsize/Roverlap and ICC, and statistical analysis of crossover design. As an index for evaluating TRR based on the activation map, these two were proposed by Rombouts et al. 1997; for calculation, it can be conducted in Matlab with a simple calculation code. For calculation of ICC, we used the ICC function of the psych package, ver. 1.8.12, 2019, provided by the R system for statistical computing, ver. 3.6.0; R Development Core Team, 2009. And for the statistical analysis of crossover design, we have adopted crossover model poposed by Grizzle et al. 1965, one of the most widely referenced method. Also, it can be conducted in Matlab.