Functional near-infrared spectroscopy (fNIRS) is a noninvasive cortical neuroimaging method that measures the task-dependent blood-oxygen level concentration changes in time.
The relatively young but quickly spreading technology does not have sufficient software support. The two most promising MATLAB-based integrated software packages are suffering from some early problems that make data processing inconvenient, and some specific steps impossible.
In my thesis I review the theoretical basis for near-infrared spectroscopy and the physical and physiological origins of the measured signals. I examine the processing options of the signals with the two available software packages in depth, paying special attention to their problems and insufficiencies.
Based on the problems with the available software, as well as the special requests from Hungarian Academy of Sciences Institute of Cognitive Neuroscience and Psychology, I developed an open-source, MATLAB-based software. The software, named gNIRS, is flexibly configurable and allows fully automated processing. After presenting the user interface, I will analyze the source code and algorithms used in detail.
In order to fully understand the operation of the software and the algorithms used it is essential to understand the experiment where the measured signals originate. The goal of the experiment is to better understand the “auditory stream segregation” phenomenon, which I present in detail. The physical and physiological origins of the signals are also important for the processing so I review them as well, and I also illustrate the physiological effects and processing steps using a pilot experiment.
In order to verify the efficient and reliable operation of gNIRS, I carried out tests in which I compared the waveforms obtained from gNIRS with the ones from nirsLAB software, that is regarded as the standard. The comparisons yielded good results, which verifies that gNIRS brings in fact the expected results.