Setting up and Analyzing High-Density fNIRS Measurements of Finger Tapping

We presented a two-part webinar series with NIRx on this topic of setting up a high-density fNIRS experiment and analyzing the experimental results with Homer and AtlasViewer.

Part 1 presents the rationale for performing high density fNIRS measurements to permit a higher spatial resolution imaging analysis compared with more traditional low-density channel-based analysis, and then shows how using AtlasViewer one can design the high density fNIRS montage of sources and detectors.

Part 2 then reviews how to analyze that data using Homer to do the short separation regression and estimate of the evoked hemodynamic response using the GLM, and then shows how bring the output of Homer into AtlasViewer to produce images of the brain activation. Importantly, these data also show that the addition of tomographic image reconstruction of brain and scalp signals permits further improvement in the estimate of the brain activation than can be achieved by short-separation regression alone.

You can access the video recordings of these webinars through NIRx at or directly at

Part 1: watch the recording here

Part 2: watch the recording here

You can download the data set used in Part 2 of the webinar here. Note that this is a large zip file as it also includes all of the results from the Monte Carlo run in AtlasViewer preparing for the image reconstruction. Thus, you will not need to run the Monte Carlo yourself to do the brain and scalp image reconstruction.

We provide information about the processing of these data below, but please feel free to post any discussion about these data in our data forum.

Finger tapping was alternating between left (condition 1) and right (condition 2) hands with the stimulus lasting for 5 sec with a stimulus starting every 15 sec.


The processing script and parameters used for the analysis in Homer were:


Sample group results from Homer with and without short-separation regression are shown below along with the imaging results from AtlasViewer reconstructing only in the brain versus reconstructing brain and scalp.