Baby fOLD: An Age-specific Toolbox for Optimizing fNIRS Channel Arrangement in Developmental Samples

• In Event: Measurement Issues in Developmental Research

Abstract

Background:
Functional near-infrared spectroscopy (fNIRS) is a non-invasive technique that uses scalp-placed light sensors to measure event-evoked changes in cerebral hemoglobin concentration. fNIRS is an alternative neuroimaging tool for infants and young children who cannot undergo fMRI procedures. Commercial fNIRS instruments do not allow for whole-head coverage. Hence, the challenge is to design source-detector channel arrangement that maximizes sensitivity to a given brain region of interest (ROI) before data collection. The study aimed to extend an existing toolbox (fOLD: Zimeo Morais, Balardin, & Sato, 2018) for optimizing channel arrangement for adults to infant samples.

Method:
The present study estimated the mapping between ROIs and channel locations in infants and adults. Individual realistic head models were constructed for 3-month-olds (N=38), 6-month-olds (N=74), and 20-to-24-year-olds (N=134). We constructed 130 source-detector channels from neighboring 10-10 electrode/optode locations defined in the fOLD toolbox. The LPBA40 atlas that contained 56 regions (Shattuck et al., 2008) was constructed for individual MRIs. We simulated photon migration through brain tissues using the MCX program (Fang & Boas, 2009). The sensitivity for each channel was calculated by multiplying the source electrode/optode fluence distribution by the detector electrode/optode fluence distribution. The specificity percentage for each channel was defined as the channel’s sensitivity to a given ROI in respect to the whole brain.

Results:
The toolbox allows users to set an ROI and a specificity threshold. It outputs the channels that can measure the ROI with specificity exceeding the threshold. We set a specificity threshold of 15% to identify channels with sufficient sensitivity to an ROI. Figure 1 shows that our estimation of the channel configuration sensitive to the left inferior frontal gyrus (LIFG) overlapped with the published fOLD toolbox using adult head models. However, the specificity for channel F3-FC3 was less than 15% in the original estimation.
There was some between-group consistency in channel-to-ROI correspondence. Of the 56 ROIs, 27 ROIs were sampled by at least one channel for 3-month-olds, 31 for 6-month-olds, and 29 for 20-to-24-year-olds. The percentage of overlapping channels across age groups ranged from 40% to 100% for the 27 ROIs that were measured for all ages. Figure 1 and Table 1 shows that the channel locations for the LIFG were relatively consistent across age groups. However, channel F3-FC3 was sensitive to the LIFG for 6-month infants and adults but not for the 3-month-olds. Channel FC5-C5 was sensitive to the LIFG only for the 6-month-olds. And channel AF7-FP1 was sensitive to the region only for the adults. Examples of ROIs that could not be sampled with sufficient sensitivity in neither group included the orbitofrontal gyrus, fusiform gyrus, and subcortical ROIs. Table 1 displays that the superior occipital gyrus and cuneus could not be reliably sampled across three ages.

Conclusions:
We present a promising toolbox that assists fNIRS researchers to design age-specific source-detector channel configurations that maximize the sensitivity to user-defined ROIs. The toolbox will be further developed to include ROIs from multiple developmentally-appropriate atlases and include channel-to-ROI mapping data for age groups across infancy and childhood.

Authors

• Xiaoxue Fu, University of South Carolina
  Presenting Author

• John E Richards, University of South Carolina
  Non-Presenting Author