In recent years, fNIRS has been increasingly used for human cerebral functional studies. Similar to blood oxygenation level-dependent (BOLD) functional magnetic resonance imaging (fMRI), fNIRS observes regional hemodynamic responses associated with neuronal activation. However, the criteria for detecting true positive fNIRS and fMRI signals appear to be based on different understandings of cerebral hemodynamics. Some recent studies addressed to this issue by comparing fNIRS data with BOLD fMRI data. There, however, remains an argument which of oxy- and deoxygenated hemoglobin changes measured by fNIRS more highly correlates with BOLD fMRI signal. In many cases of such studies, multichannel fNIRS measurement with spatially sparse sampling (around 30 mm) was performed, which may often fail to detect true functional hemodynamics. Furthermore, fNIRS channel positions in these studies were not very precisely co-registered on fMRI data. Comparisons of such data with BOLD signals could confound the discussion on the issue above. To solve this problem, fNIRS with a higher spatial resolution at least less than 10-15 mm is very necessary. One solution is a double-latticed optode arrangement with a shift of 15 mm. This arrangement, however, required twice larger number of optodes than that required by the single-latticed normal fNIRS measurement, which needs longer time and more effort for setting up all optodes successfully; thus, is not very practical especially for the simultaneous measurement with fMRI. For another solution, we propose a novel method using fewer optodes of a bifurcated end with a triangular arrangement of 30 mm pitch. The bifurcated ends were separately connected to the source and detector in the NIRS main unit. By temporally switching the use of each optode as the source and detector, a high-density measurement of 15 mm spatial resolution was accomplished. The functional hemodynamics in the data was extracted by using the hemodynamics modality separation method proposed in our previous study. Functional experiments of single-sided motor task showed that this system could successfully detect lateralized and considerably localized functional hemodynamics.