Objective Functional near-infrared spectroscopy (fNIRS) is an effective neuroimaging tool used to directly determine changes in oxygenated hemoglobin (HbO) and deoxygenated hemoglobin (HbR). This technique offers portable and radiation-free alternatives to conventional neuroimaging modalities, such as functional magnetic resonance imaging, positron emission tomography, and electroencephalograph. However, current portable fNIRS systems have limited detection sensitivity due to the employment of photodiodes working in analog mode. To cope with the adversities, we propose a portable two-wavelength continuous-wave fNIRS-topography system based on lock-in photon-counting technology to be applied in cognitive neuroscience and brain-computer interface (BCI) studies in a daily environment. The proposed system has a good prospect of application in faint activation detection by integrating the parallel measurement capability derived from lock-in technology and the high sensitivity derived from photon-counting technology. Thus, the proposed system should provide a noninvasive route, reasonable temporal resolution, and high-sensitive information for studying fNIRS-BCI.
Methods In this study, the proposed portable fNIRS system possesses four pairs of source- and detection-optodes, which are matched with the configurations of a nonoverlapping source-detector array to conduct experiments. Source-optode bundles two single-mode fibers with a core diameter of 9 μm connected to two laser diodes (LDs) at 785 and 830 nm wavelengths. Besides, a modulator with eight square wave signals of different frequencies is designed based on a field-programmable gate array (FPGA) kit. The modulated light reaches the cerebral cortex and is subject to refraction, scattering, and absorption by HbO and HbR. A part of the light is redirected to the scalp, where HbO and HbR concentration changes are detected by the detection module in turn, which adopts a photomultiplier tube with a linear range of 5.0×10 6 s -1 and a dark count of 600 s -1 as the core device. A lock-in photon-counting demodulator is designed based on FPGA to achieve parallel signal demodulation and detection in all sampling channels. Besides, the intensity of each LD can be automatically adjusted using a digital potentiometer (AD5259) for adaption to different subjects. Wireless communication and rechargeable batteries are used to ensure portability. All the mentioned components and printed circuit boards are fixed in a portable case with a volume of 23.6 cm×11.0 cm×20.5 cm, which is lightweight and miniaturized.
Results and Discussions A series of phantom and in-vivo experiments have been implemented to examine the effectiveness of the proposed portable fNIRS-topography system. The phantom experiments showed that the proposed system has excellent stability (Fig.3), linearity (Fig.4), and anticrosstalk ability (Fig.5). For the in-vivo experiments (Fig.6), we adopt two stimulation paradigms: breath-holding (BH) and mental arithmetic (MA) to evaluate the system. The complete dataset is synchronously acquired at a sampling frequency of 4 Hz. Then, changes in the perturbed HbO and HbR concentrations are calculated according to the modified-Beer-Lambert-law. The frequency spectrum of the measurements is analyzed using the Fourier transform, which provides a theoretical basis for the bandpass filtering (0.018--0.3 Hz), and the spectrum peak (0.02333 Hz) is close to the stimulation frequency (Fig.7). By further analyses of these spectra [Fig.8(a) and Fig. 9(a)], we strengthen the actual activation frequency induced by the stimulation. The results show that the proposed system can accurately trace the time-course of the global activation perturbation by BH stimulation [Fig. 8(b)] and the partial activation perturbation by MA stimulation [Fig.9(b)]. Besides, the optical topography (OT) image of the measurements from the MA paradigm showed that the activation region is roughly located at the center of the left half of the prefrontal lobe [Fig.9(c) and Fig. 9(d)].
Conclusions In this study, we developed a portable fNIRS-topography system for the applications of BCI based on lock-in photon-counting technology, which has the advantages of high sensitivity and moderate time resolution. A two-layered brain phantom is employed to evaluate the system. The results showed that the proposed system has excellent stability, linearity, and anticrosstalk ability. Besides, the proposed system is used to measure the perturbed hemoglobin concentration in the prefrontal lobe during BH and MA stimulation paradigms. The results showed that the proposed system can track changes in hemoglobin concentration. The frequency spectrum of the measurements is analyzed, and the spectrum peak (0.02333 Hz) is close but not equal to the stimulation frequency (0.025 Hz). The probable cause of this situation is the hemodynamic response delay (about 2--3 s) in the brain. The image reconstructed by OT during the MA stimulation paradigm showed that the activation region is roughly located in the center of the left half of the prefrontal lobe. We analyzed the entire experimental process and confirmed that simple training before experiments enhances the system performance. In future studies, we will design more stimulation paradigms, e.g., sports and idea recognition, to further explore the potential of fNIRS-BCI in clinical application.
