The Development of the NIRS-based Neurofeedback System

If we can feed back the information of our brain activity such as brain waves and cerebral blood flow, we may be able to control our brain activation for ourselves. This technique is called neurofeedback and can be applied to the therapy and training in the neurological and psychiatric field. Near infrared spectroscopy (NIRS) is a non-invasive method to measure brain activity using changes in the degree of hemoglobin oxygenation. We focused on NIRS, whose features are high temporal resolution and flexibility for routine monitoring. In this study, we have developed the NIRS-based neurofeedback system. When multiple channel data from NIRS are fed back directly to subjects, they cannot do effective training because these data contain too much information. We therefore tried to localize emotion-responsive areas with NIRS (Hitachi ETG-4000). To make subjects' emotions evoked, the visual stimulus using pictures was presented to them. The neutral and negative pictures were prepared from International Affective Picture System (IAPS), which is a database of photographs for emotion research, and 28 pictures in each emotion category were selected. A block design was used, which is a typical procedure for the NIRS experiment. Every block consisted of a 4-second presentation of 2 pictures from the same category (2 seconds per picture) and an 8-second fixation baseline. First 14 blocks in neutral category were presented, and then 14 blocks in negative category. The relative changes of oxy-hemoglobin concentrations were measured at 52 measurement points on the subject's frontal region. In 6 healthy subjects, the brain activated areas were localized with highest response to negative pictures compared to neutral pictures. Two analysis methods were used for localization: t test and NIRS-SPM. Using the results of the t test, the localized area was dorsolateral prefrontal cortex (3 subjects). Computing the general linear model on NIRS raw data with the NIRS-SPM software, the localized areas were superior temporal sulcus (6 subjects) and ventrolateral prefrontal cortex (2 subjects). To realize neurofeedback, the NIRS unit was connected with a PC by TCP/IP network. The PC picked up the oxy-hemoglobin data of the emotion-responsive areas and displayed the picture of a speed meter whose speed reflected oxy-hemoglobin concentration from NIRS. When we fed back on trial to the same 6 subjects using the developed system, real-time neurofeedback could be done without any trouble. These results suggested the feasibility of NIRS-based neurofeedback.