|
1.IntroductionFunctional near-infrared spectroscopy (fNIRS) is an optical technique for the noninvasive measurement of human brain activity.1,2 Near-infrared (NIR) light, emitted from light sources placed on the scalp, travels through the tissues (i.e., scalp, skull, and brain) and is measured as diffusely reflected light by detectors. Based on the attenuation changes of the detected light, relative changes of oxy- () and deoxyhemoglobin (HHb) concentration in the brain region between source and detector can be reconstructed. When a person is mentally active (e.g., during the execution of a motor task), the concentrations of () and HHb ([HHb]) in the brain vary (functional hyperemia due to neurovascular coupling), thereby providing an estimate for local brain activity detectable by fNIRS. Through the development of fiberless and portable instruments, fNIRS opens new avenues for wearable brain imaging applications.3–10 In particular, wearable fNIRS devices may find application in brain–computer interfaces (BCI) for “out of the lab” applications, e.g., to trigger robotic devices for assistance or rehabilitation of neurological patients in the home environment,11,12 for the communication of locked-in patients,13,14 or for neuroergonomic investigations (i.e., investigating brain behavior in workplace environments).7,15 In comparison with other BCI technologies, fNIRS has the advantage of being small and inexpensive (e.g., compared to magnetic resonance imaging and magnetoencephalography), noninvasive (e.g., compared with electrocorticography), and robust to electrical noise (e.g., compared with electroencephalography). Recently, Ferrari et al.16 highlighted that fNIRS hardware development is still in an experimental stage and that there is a need for optimizing and engineering real-world fNIRS instrumentation. By employing commercially available light sources and sensors, fNIRS instruments can be miniaturized relatively easily and manufactured inexpensively.1 However, high requirements for signal quality and system reliability make the development of wearable fNIRS instruments challenging. Several approaches to improve the signal quality of fNIRS instruments have been proposed. Tachtsidis and Scholkmann17 recently highlighted the importance of using short-channel regression methods18,19 to remove the influence of systemic physiological changes from the fNIRS signal. Physiological signals are acquired by means of a short-separation (SS) channel [ideally with a source–detector separation (SDS) below 10 mm, as suggested by Refs. 20 and 21], which is then used for filtering the physiological “noise” from long-separation (LS, with SDS typically above 30 mm) measurements, for example, using the filtering approach developed by Saager and Berger.19 To the best of our knowledge, no wearable commercial fNIRS instrument currently implements SS channels below 20 mm, precluding short-channel regression approaches for the reduction of physiological noise. Another approach to maximizing fNIRS signal quality consists of measurements at multiple wavelengths (i.e., more than two) to obtain additional tissue information. Multiwavelength measurements enable the more accurate determination of and [HHb] in the modified Beer–Lambert law (MBLL), since the effect of electronic noise (e.g., switching noise from electronic components, thermal noise, electronic shot noise) is reduced thanks to more observations available for the calculation of the matrix inversion. Furthermore, by including additional wavelengths, the possibility of determining concentration changes of other chromophores is provided. In particular, concentration changes of oxidized cytochrome-c-oxidase (oxCCO)22 can be calculated. Since oxCCO is the terminal enzyme in the electron transport chain, it is a direct marker of mitochondrial oxygen consumption, and the amount of oxidation directly reflects the mitochondrial metabolism. Most commercial fNIRS devices employ only two wavelengths,1 and only a few prototypes offer the possibility of measuring at additional wavelengths. To reliably capture signals from the cerebral cortex of the human brain, a large SDS is needed as it allows comparably more light to travel through deeper tissue regions before reaching the detector.21 Since at larger distances the light is strongly attenuated (e.g., attenuation of to at 30 mm SDS23), highly sensitive hardware to detect small light intensities is needed.24 Many stationary fNIRS devices use avalanche photodiodes or photomultiplier tubes (see Ref. 1 for a review), which exhibit the highest photosensitivity.25 However, due to their large size, high cost, and high operating voltages, these technologies are not well suited for wearable instruments. Photodiodes are the most commonly used detector technology for wearable fNIRS instruments1 because they operate at low voltages (), are inexpensive components, and are small in size. In recent years, the use of silicon photomultipliers (SiPMs) has been exploited for fNIRS applications5,24,26 due to their high photosensitivity for a small component size and their relatively low cost. In our previous work,24 we developed a first prototype of an SiPM-based fNIRS instrument demonstrating a high signal-to-noise ratio (SNR) of more than 70 dB for SDS below 30 mm, as well as a high photosensitivity for small light intensities at larger SDS. While this initial prototype confirmed that SiPMs are highly suitable for wearable fNIRS instruments,24,26,27 it suffered from different limitations (e.g., size, data readout, lack of modularity, optical drift) that required further improvements and motivated the development of another instrument. Building on our previous work,24 the aim of this project is to design and implement a highly sensitive fNIRS instrument that unites three main features: (i) simultaneous measurements with SS and LS channels, (ii) measurements with four-wavelength light sources, and (iii) modular optode design to accommodate for fNIRS recordings over various brain regions of interest. We expect that an instrument combining these three key features will provide a high-quality fNIRS device that could help bring fNIRS technology “out of the lab” and make it available for a wide range of applications. In the subsequent sections, we present the concept and technical overview of the fNIRS instrument and further report on its preliminary performance evaluation through phantom and in vivo measurements. 2.Methods2.1.Instrument RequirementsThe proposed continuous-wave instrument was developed to fulfill four main aspects, adapted from von Lühmann et al.6
2.2.System DescriptionThe concept of the proposed fNIRS instrument (Fig. 1) consists of a reconfigurable network of small hexagonal modules with on-board electronics, as well as emitters (four LEDs) and a detector (SiPM). Each individual module can be operated separately and can be combined with other modules, allowing the realization of both SS (7.5 mm) and various LS (20 mm or more) measurement channels. Optode modules are controlled by a control unit that can be placed remotely and that is connected to a personal computer (PC). A freely selectable number of optode modules can be used simultaneously and in a way to optimally cover the area of interest on the subject’s head. A GUI programmed in LabVIEW [National Instruments (NI), Texas] operates the measurement from any computer containing a wireless receiver. The control unit consists of a power supply and a NI myRIO embedded controller. The NI myRIO performs all data processing steps and simultaneously acts as the interface between the computer and the optode modules, including Wi-Fi communication with the computer and (inter-integrated circuit) data transfer with the modules. A battery provides electrical power for the NI myRIO (12 V) and the optode modules (3.6 V). The optode modules and the control unit are connected by one cable with five lines: power (3.6 V, GND), communication (), and timing (SYNC; used for timed measurements). Each optode module contains four LEDs and one SiPM. Alternatively, one module of the connected modules operates as source and all the others as detector—including the source module itself— enabling simultaneous measurements with SS [7.5 mm, representing the physical distance between LEDs and SiPM on the printed circuit board (PCB)] and LS channels. Modularity is achieved by being able to add and remove modules between measurement sessions. 2.3.Hardware Design of SiPM-Based Optode ModulesThe PCBs of the optode modules contain all electrical components for individual fNIRS data acquisition (light emission and detection; see Fig. 2, bottom), including power regulators, microcontroller (), digital-analog convertor (DAC), and analog-digital convertor (ADC). The PCBs were manufactured based on the rigid-flex PCB technology, where specific parts of the PCBs have flexible properties that can be bent, allowing the omission of delicate board-to-board connectors. One PCB consists of two rigid hexagons that are connected by one flexible connecting part.
Table 1Measured LED properties at IF = 30 mA.
Note: IF, LED forward current; FWHM, full-width at half-maximum. The microcontroller communicates with the ADC and DAC via SPI and with the control unit via . For communication, each module is given an individual address and communication is performed sequentially for every module. To keep communication time low, only 10 bytes are transmitted per sample and module—two for every wavelength and backlight—resulting in a communication time of per module (Fig. 3). In total, one entire measurement cycle lasts less than 10 ms, including light measurement (6.6 ms), on-board data processing (1.2 ms), and data communication with the control unit (), where is the number of modules that have to be read-out. A maximal overall sampling frequency of 100 Hz can be achieved. In addition to the hardware safety mechanisms, software safety features are implemented at the level of the microcontroller. If the maximal current through the SiPM exceeds a predefined limit (i.e., 1.5 mA), the sensor is turned off by reducing the operating voltage below breakdown voltage, resulting in the deactivation of the channel. At the same time, all other channels can continue operating as before. 2.4.Calculation of Concentration ChangesConcentration changes of oxyhemoglobin, deoxyhemoglobin, and oxCCO (i.e., , [HHb], and [oxCCO], respectively) were obtained by applying MBLL for four wavelengths with 774, 817, 865, and 892 nm (UCL4 algorithm37). The algorithm was implemented for the reconstruction of two (, [HHb]), and three chromophores (, [HHb], and [oxCCO]) Extinction coefficients for the chromophore C and wavelength were selected according to Matcher et al.37 Normalized values for the differential pathlength factor (DPF) were obtained from the Biomedical Optics Research Group at University College London, renormalized according to Ref. 38, and multiplied with the physical channel length . Since the concentration of oxCCO in the cells is assumed to be constant, the difference spectrum between oxidized and reduced cytochrome-c-oxidize was used (i.e., oxCCO and redCCO). The inverse of the nonsymmetric extinction matrix was calculated based on the Moore–Penrose pseudoinverse . The change of optical density is defined as , where and are the optical signal measured by the sensor for the current and the previous sample, respectively.Short-channel regression was implemented according to Saager and Berger.19 It is assumed that the SS channel contains only signal contributions of extracerebral physiological origin and that it can be subtracted with a scaling factor from the LS measurement, which contains a combination of extracerebral and cerebral signals. The scaling factors for and HHb are both obtained from the correlation between the long and short channels. 2.5.Experimental Protocol for Instrument EvaluationThe proposed fNIRS instrument was tested by means of phantom and in vivo measurements to evaluate the ability to reliably measure hemodynamic changes. First, measurements on silicone phantoms mimicking human tissue were performed to evaluate the technical performance of the instrument. Subsequently, in vivo measurements on muscle and brain tissue were performed to characterize the instrument’s performance in reconstructing , [HHb], and [oxCCO]. 2.5.1.Phantom measurementsA silicone phantom (ISS Inc., Champaign, Illinois) with similar optical properties as the tissues of the human forehead was selected for the measurements (Table 2).11 The optical properties of the silicone phantom were obtained with a frequency-domain fNIRS system (ISS OxiplexTS). The optical loss (OL) in the phantom denotes the inverse value of the reflectance (units: ) multiplied with a standardized circular aperture of 8 mm according to the IEC 80601-2-71:2015 standard. Reflectance was obtained by solving the diffusion equation for a semi-infinite medium,39 which was validated with Monte-Carlo simulations,40 as well as reference measurements performed with a power meter (ILX Lightwave OMM-6810B). In the phantom, the scattering behavior (quantified by the reduced scattering coefficient ) is slightly stronger than that of the human forehead tissue, whereas the absorption coefficient () is similar; this results in a slightly stronger overall light attenuation of the silicone phantom compared with the human head tissue. Thinking of future fNIRS applications investigating motor-related brain areas under hairy regions, the overall light attenuation of the phantom is in a realistic range. Table 2Optical properties of a human forehead and the optical silicone phantom used for the validation measurements.24
High-quality raw signals are crucial in fNIRS for optimal spectroscopic separation of the chromophores and to achieve a high repeatability and reproducibility of the measurement. For validation of signal quality, different performance features were evaluated to obtain an overview of the instrument’s performance:
2.5.2.Physiological measurementsFurther evaluation of the instrument was performed during measurements on two human subjects to confirm the ability of the sensor to measure hemodynamic changes with correct trend and magnitude. All subjects were informed about the procedure and gave consent prior to the measurements.
3.Results3.1.Hardware Characterization and Phantom MeasurementsAn overview of the technical performance of the fNIRS instrument with focus on the custom-built optode modules is provided in Table 3. A tradeoff between signal quality, sampling frequency, heating, safety restrictions, and component size was found. The modules were manufactured from rigid-flex PCBs, which allowed the miniaturization of the modules to a size of (PCB only without mechanical casing) or (including mechanical casing), and a corresponding weight of 3 and 5 g, respectively. The four wavelengths at 770, 810, 850, and 885 nm were selected to closely match the optimal wavelength obtained from Arifler et al.34 to minimize optical crosstalk when calculating , [HHb], and [oxCCO]. The power consumption of the modules depends on the emitted light intensity, the photocurrent, and the sampling frequency; in the condition where the maximal power consumption is expected (when measuring at 100 Hz), 0.4 W was not exceeded. By limiting the photocurrent to 1 mA, the temperature of the sensor—being the hottest part of the setup—remained below 41°C. Table 3Technical performance of the fNIRS instrument.
The optical sensitivity of the sensor circuitry, expressed by the NEP, was found to be 0.94, 1.3, 1.86, and 2.67 pW for the four wavelengths. By calculating the ratio of the maximal measurable optical power (1 mW) and the NEP, dynamic ranges larger than 160 dB for all wavelengths are obtained, enabling simultaneous measurements in the order of milliwatt to picowatt. For OL larger than , the SNR results were identical to our previous work,24 with values gradually dropping from 64 to 20 dB for SDS between 35 and 65 mm on the used silicone phantom. The threshold at which the SNR falls below 40 dB is at (OL ) for the investigated phantom, LED power, and sampling frequency. For shorter distances, the SNR is in the range of 64 dB. While at beginning of operation (after 60 s) a visible drift was observed with a negative slope of 0.3 ‰/s or smaller (dependent on the SDS), this value continuously decreased to 0.1 ‰/s after 10 min. For smaller overvoltages (short distances), temperature changes have a stronger influence on the PDE, leading to slightly higher drift coefficients. Measurements with one or multiple channels are enabled. The total number of channels scales with the number of modules —there is always a maximum of channels (e.g., channels when using 2 modules). The sampling frequency is shared across the source modules and detector modules by , where 0.06 is of the 0.6 ms communication time that is required for data readout (Fig. 3). For example, for a configuration with 10 modules, where every module alternately operates as a source module, 100 optical channels sampled at 6.25 Hz can be achieved. When two instead of four wavelengths are used, the sampling frequency could be further increased to 125 Hz—the frequency does not double as some steps (e.g., backlight measurement, voltage adjustment) are performed for every sample, independent of the number of wavelengths used. The influence of electronic crosstalk between modules is negligible thanks to specific on-board electronics for the LEDs and SiPM, which are placed on each optode module. Table 4 highlights the changes in chromophore concentration variance when more than two wavelengths are used for calculating and [HHb]. While the variance is reduced by 6% for [] and 1% for [HHb] when three wavelengths are used, the effect gets larger by including a fourth wavelength. shows a reduction of variance by 25% when using four instead of two wavelengths, while the variance of [HHb] decreases by a factor of 8%. Table 4Decrease of signal variance when additional wavelengths are used for the calculation of [O2Hb] and [HHb]. For both chromophores, two wavelength-combinations were investigated (left row), showing the variance decrease in the right row.
3.2.Physiological Validation3.2.1.Arterial occlusionFigure 5 shows the calculated concentration changes during an arm arterial occlusion. As expected, a small increase of and [HHb], originating from an incomplete occlusion (venous occlusion), is observed when the pressure cuff is inflated, which is followed by a simultaneous increase of and decrease of [HHb] (arterial occlusion). The concentration change of [oxCCO] is small in comparison with the hemoglobin changes; when blood flow is occluded, a small upward trend is visible with a change less than of [HHb]. After pressure release from the cuff, the typical hyperemic change with a overshoot and a [HHb] undershoot can be measured. At the same time, [oxCCO] reveals a small peak (about of [HHb] peak change). 3.2.2.Task-evoked brain activityFigure 6 shows an example of resting state brain measurement with one SS and one LS channel. During rest, only extracerebral signal changes are expected, which is best seen in the signals. In of both channels, nearly identical signal contributions with the expected low-frequency oscillations (Mayer waves) and the heartbeat are measured. For the motor execution task, group averages for SS and LS channels were performed over the 10 repetitions (Fig. 7). During the task, a clear hemodynamic response with an increase of and a simultaneous smaller decrease in [HHb] was measured in the LS channel. [oxCCO] remained unchanged during the entire measurement. In the LS channel, task-related physiological changes can be observed, but no distinct hemodynamic response was obtained. After application of the short-channel regression to remove hemodynamic changes occurring in the extracerebral tissue layer (scalp blood flow) from and [HHb], a filtered signal with removed peaks at task onset () and during rest () was obtained. 4.DiscussionIn this paper, the development and characterization of an fNIRS instrument was presented. It combines three innovative features, namely (i) the simultaneous measurement with SS and LS channels, (ii) the emission of NIR light at four wavelengths, and (iii) a modular optode arrangement to optimize the concentration signals that can be obtained from an fNIRS instrument. To the best of our knowledge, the proposed device is the first wearable fNIRS instrument implementing all these three key features. It further provides low-noise and fast data acquisition, thus maximizing the reliable determination of concentration changes at short and long SDS. Several improvements in comparison with our previous prototype24 were achieved, including minimization of temperature drift, implementation of readout for fast data communication, design of safety circuits, advanced processing steps for the same sampling frequency, and miniaturization of the PCBs. The instrument allows simultaneous measurements with a large optical dynamic range of 160 dB, outperforming other existing wearable fNIRS instruments (80 dB in Ref. 4, 55 dB in Ref. 6). Robust measurements (i.e., ) can be obtained at OL up to , corresponding to SDS of to 55 mm in human head tissue.30 This finding was confirmed by a functional measurement on a human subject where a long SDS of 40 mm over the primary motor cortex was used, which is the same range or larger than SDS typically used in other fNIRS instruments (25 to 42 mm).3,4,6 According to simulations, increasing the SDS from 30 to 40 mm improves the sensitivity to detecting hemodynamic changes in the brain by .21 In general, the obtained SNR values for longer SDS (47 dB at OL, 31 dB at OL) are comparable24,26,30 or slightly larger4,32 than those reported in the literature and are in the same range for shorter SDS with OL below (SNR of 64 dB). In comparison with our previous prototype,24 identical SNR values were obtained, except for short SDS (74 dB instead 64 dB) where the maximal photocurrent was decreased from 6 to 1 mA to reduce heating effects in the SiPM. The SNR at longer SDS could be further improved by increasing the LED emission power (e.g., up to 20 mW such as in Piper et al.3). Optical drift in the raw signal could be maintained at a small level with values below 0.1 ‰/s, which does not affect concentration calculation and is in the same magnitude as that reported in other work.47 The overall sampling frequency of 100 Hz is comparably high, reducing electronic noise due to oversampling and making measurements with multiple modules (e.g., up to 10 modules) possible. The design of the presented prototype enables wearable applications thanks to the highly miniaturized hexagonal modules and a compact control unit supporting real-time wireless communication. The structure based on small size and low weight optodes that are connected to a control unit that can be placed in a backpack is similar to the approach presented by Piper et al.3 In comparison with other approaches,4,6,48 the modules placed on the head are distinctly smaller in our instrument, thereby decreasing the risk of uncomfortable optode placement and motion artefacts. Measurements with only one, but also with several modules—depending on the individual user needs and the application—are possible. When 10 modules are connected, a measurement with 100 channels and 6.25 Hz is achievable. By placing multiple modules next to each other, high-density optical channel arrangements are achievable, with a channel network similar to diffuse optical tomography (DOT) approaches.49,10 The DOT approach by Chitnis et al.10 provides the possibility to arrange four modules in such a way that a simultaneous 128-channel measurement with two wavelengths can be realized. When additional wavelengths are included in the calculation of concentration changes, the influence of electronic noise can be reduced significantly, which was demonstrated on phantom measurements by a decrease in the variance of and [HHb] by 25% and 8%, respectively, when four instead of the typical two wavelengths were used. These results suggest that more accurate and robust chromophore calculations can be realized thanks to additional wavelengths, leading to more information content in the MBLL. Furthermore, when using four wavelengths, it is possible to calculate [oxCCO], which is known as a marker of mitochondrial oxygen consumption. This information might allow a more robust determination of brain activity since oxCCO should be independent of systemic signals,22 which bears high potential for increasing the accuracy of classifiers in BCI applications. Nevertheless, the use of oxCCO is controversially discussed, as no golden standard for in vivo validation measurements exists at the moment and its accurate calculation is influenced by several factors (e.g., LED emission spectra, the DPF values, temperature effects, reconstruction method, extinction spectra).37,50 A distinct and expected hemodynamic response during a finger-pinching task was measured in the LS channel over the primary motor cortex. The hemodynamic response showed the same behavior as reported in Refs. 46 and 51 (i.e., an increase in and a decrease in [HHb]). In the SS channel, no distinct hemodynamic response was observed (i.e., the decrease in [HHb] was absent), which goes along with expectations since we assume that little to no signal contributions from cerebral regions were measured. However, perfusion changes in the extracerebral tissue in response to the task17,47 were observed (i.e., a strong increase in ). The double-peak in [O2Hb] in the SS channel is also an indication for scalp blood flow changes in response to a mental task (own observation based on various experiments and measurement devices). By performing a short-channel regression according to Saager and Berger,19 a cleaner hemodynamic response was obtained, highlighting the feasibility and benefit of including a short SDS for filtering and demasking the LS channel from physiological noise. While other approaches for removing physiological noise exist, e.g., through software filtering applied to the LS channel only to reduce confounding effects (see Ref. 12 for a review), these algorithms are sophisticated, computationally demanding, and cannot make up for the additional information obtained when measuring systemic signals directly. SS channels deliver information that is important for a correct interpretation of fNIRS measurements. Zimmermann et al.52 proposed a different approach in acquiring biosignals in parallel (i.e., heart rate, breathing rate, blood pressure, and skin conductance). However, this comes at the disadvantage of requiring additional experimental setup that is complicated to use, time-consuming to set up, and, therefore, limited to laboratory use. To further improve the proposed instrument, the replacement of the selected LED types should be considered to measure with more suitable optical properties (smaller FHWM, more accurate peak wavelengths32) and to minimize the distances between the four LEDs (e.g., using multiwavelength LEDs), thereby better satisfying the assumptions of the MBLL (i.e., point source, single emission wavelength). A mechanical casing that contains optical fibers for simpler guidance of the light through the hair (which is a common issue in all fNIRS instruments) should be developed. Careful selection of optical fibers should allow improvement of the robustness of the instrument for use in different experimental paradigms. In combination with a head fixation, for example, similar to available electroencephalography-setups such as the Emotiv EPOC (Emotiv, California), wearability and simple use of the device should be guaranteed. By providing a robust and unobtrusive module fixation, we hope to make the step out of the laboratory environment toward everyday environments with the proposed fNIRS instrument. In this work, no compensation for the nonlinear SiPM behavior at short SDS was performed. By acquiring calibration measurements at various short SDS, the nonlinearity could be compensated for, with potential to further improve estimations of concentration changes. In future applications, it is desirable to deploy and test a configuration with a larger number of modules to obtain simultaneous information from multiple brain areas and to generate two-dimensional images based on the measured signals. 5.ConclusionAn fNIRS instrument with three important features (i.e., short and long SDS, four-wavelength light sources, modular placement and configuration) was proposed and implemented. High modularity is achieved through miniaturized hardware design of optode modules that contain sources and detector and that can be individually connected to a central unit. The inclusion of SS and LS channels help to detect and compensate for physiological signals, a crucial feature for many fNIRS applications, in particular BCIs. By including four wavelengths, more robust estimates of concentration changes could be achieved, while allowing further investigation of the use of [oxCCO] as an additional marker for brain activity. High-quality test bench measurements that outperform existing fNIRS instruments in many aspects were obtained, and in vivo tests confirmed the sensible chromophore calculation of , HHb, and oxCCO. The proposed prototype could pave the way for robust fNIRS measurements in real-life applications. DisclosuresThe authors declare that there is no conflict of interest, financial or of any other nature, regarding the publication of this article. AcknowledgmentsThis work was supported by the Strategic Japanese-Swiss Cooperative Research Program on “Medicine for an Aging Society” and the ETH Foundation in collaboration with Hocoma AG. We thank Camila Shirota for her help in reviewing the manuscript and Thomas Ganka from KETEK GmbH for providing valuable input on the use of the SiPM. We also thank Jumpei Arata and Rob Labruyère for allowing us to perform test measurements with their commercial fNIRS instruments. ReferencesF. Scholkmann et al.,
“A review on continuous wave functional near-infrared spectroscopy and imaging instrumentation and methodology,”
Neuroimage, 85 6
–27
(2014). http://dx.doi.org/10.1016/j.neuroimage.2013.05.004 NEIMEF 1053-8119 Google Scholar
F. F. Jobsis,
“Noninvasive, infrared monitoring of cerebral and myocardial oxygen sufficiency and circulatory parameters,”
Science, 198 1264
–1267
(1977). http://dx.doi.org/10.1126/science.929199 Google Scholar
S. K. Piper et al.,
“A wearable multi-channel fNIRS system for brain imaging in freely moving subjects,”
Neuroimage, 85 64
–71
(2014). http://dx.doi.org/10.1016/j.neuroimage.2013.06.062 NEIMEF 1053-8119 Google Scholar
D. Chitnis et al.,
“Towards a wearable near infrared spectroscopic probe for monitoring concentrations of multiple chromophores in biological tissue in vivo,”
Rev. Sci. Instrum., 87
(6), 065112
(2016). http://dx.doi.org/10.1063/1.4954722 RSINAK 0034-6748 Google Scholar
D. Agro et al.,
“Design and realization of a portable multichannel continuous wave fNIRS,”
in AEIT Annual Conf.—From Research to Industry: The Need for a More Effective Technology Transfer (AEIT),
1
–6
(2015). http://dx.doi.org/10.1109/AEIT.2014.7002013 Google Scholar
A. von Lühmann et al.,
“Toward a wireless open source instrument: functional near-infrared spectroscopy in mobile neuroergonomics and BCI applications,”
Front. Hum. Neurosci., 9 617
(2015). http://dx.doi.org/10.3389/fnhum.2015.00617 Google Scholar
R. McKendrick, R. Parasuraman and H. Ayaz,
“Wearable functional near infrared spectroscopy (fNIRS) and transcranial direct current stimulation (tDCS): expanding vistas for neurocognitive augmentation,”
Front. Syst. Neurosci., 9 27
(2015). http://dx.doi.org/10.3389/fnsys.2015.00027 Google Scholar
J. Safaie et al.,
“Toward a fully integrated wireless wearable EEG-NIRS bimodal acquisition system,”
J. Neural Eng., 10
(5), 056001
(2013). http://dx.doi.org/10.1088/1741-2560/10/5/056001 1741-2560 Google Scholar
P. Pinti et al.,
“Using fiberless, wearable fNIRS to monitor brain activity in real-world cognitive tasks,”
J. Vis. Exp., 2015
(106), e53336
(2015). http://dx.doi.org/10.3791/53336 Google Scholar
D. Chitnis et al.,
“Functional imaging of the human brain using a modular, fibre-less, high-density diffuse optical tomography system,”
Biomed. Opt. Express, 7
(10), 4275
–4288
(2016). http://dx.doi.org/10.1364/BOE.7.004275 Google Scholar
R. Zimmermann et al.,
“Towards a BCI for sensorimotor training: initial results from simultaneous fNIRS and biosignal recordings,”
in Proc. Annual Int. Conf. IEEE Engineering in Medicine and Biology Society (EMBS),
6339
–6343
(2011). http://dx.doi.org/10.1109/IEMBS.2011.6091565 Google Scholar
N. Naseer and K.-S. Hong,
“fNIRS-based brain–computer interfaces: a review,”
Front. Hum. Neurosci., 9 3
(2015). http://dx.doi.org/10.3389/fnhum.2015.00003 Google Scholar
U. Chaudhary, N. Birbaumer and A. Ramos-Murguialday,
“Brain–computer interfaces for communication and rehabilitation,”
Nat. Rev. Neurol., 12
(9), 513
–525
(2016). http://dx.doi.org/10.1038/nrneurol.2016.113 Google Scholar
M. Naito et al.,
“A communication means for totally locked-in ALS patients based on changes in cerebral blood volume measured with near-infrared light,”
IEICE Trans. Inf. Syst., E90-D
(7), 1028
–1037
(2007). http://dx.doi.org/10.1093/ietisy/e90-d.7.1028 Google Scholar
S. H. Fairclough,
“Fundamentals of physiological computing,”
Interact. Comput., 21
(1–2), 133
–145
(2009). http://dx.doi.org/10.1016/j.intcom.2008.10.011 INTCEE 0953-5438 Google Scholar
M. Ferrari et al.,
“Special section guest editorial: clinical near-infrared spectroscopy and imaging of the brain,”
Neurophotonics, 3
(3), 031401
(2016). http://dx.doi.org/10.1117/1.NPh.3.3.031401 Google Scholar
I. Tachtsidis and F. Scholkmann,
“False positives and false negatives in functional near-infrared spectroscopy: issues, challenges, and the way forward,”
Neurophotonics, 3
(3), 039801
(2016). http://dx.doi.org/10.1117/1.NPh.3.3.039801 Google Scholar
Q. Zhang, E. N. Brown and G. E. Strangman,
“Adaptive filtering for global interference cancellation and real-time recovery of evoked brain activity: a Monte Carlo simulation study,”
J. Biomed. Opt., 12
(4), 044014
(2007). http://dx.doi.org/10.1117/1.2754714 JBOPFO 1083-3668 Google Scholar
R. B. Saager and A. J. Berger,
“Direct characterization and removal of interfering absorption trends in two-layer turbid media,”
J. Opt. Soc. Am. A, 22
(9), 1874
–1882
(2005). http://dx.doi.org/10.1364/JOSAA.22.001874 Google Scholar
S. Brigadoi and R. J. Cooper,
“How short is short? Optimum source–detector distance for short-separation channels in functional near-infrared spectroscopy,”
Neurophotonics, 2
(2), 025005
(2015). http://dx.doi.org/10.1117/1.NPh.2.2.025005 Google Scholar
G. E. Strangman, Z. Li and Q. Zhang,
“Depth sensitivity and source-detector separations for near infrared spectroscopy based on the Colin27 brain template,”
PLoS One, 8
(8), e66319
(2013). http://dx.doi.org/10.1371/journal.pone.0066319 POLNCL 1932-6203 Google Scholar
G. Bale, C. E. Elwell and I. Tachtsidis,
“From Jöbsis to the present day: a review of clinical near-infrared spectroscopy measurements of cerebral cytochrome-c-oxidase,”
J. Biomed. Opt., 21
(9), 091307
(2016). http://dx.doi.org/10.1117/1.JBO.21.9.091307 JBOPFO 1083-3668 Google Scholar
F. B. Haeussinger et al.,
“Simulation of near-infrared light absorption considering individual head and prefrontal cortex anatomy: implications for optical neuroimaging,”
PLoS One, 6
(10), e26377
(2011). http://dx.doi.org/10.1371/journal.pone.0026377 POLNCL 1932-6203 Google Scholar
R. Zimmermann et al.,
“Silicon photomultipliers for improved detection of low light levels in miniature near-infrared spectroscopy instruments,”
Biomed. Opt. Express, 4
(5), 659
–666
(2013). http://dx.doi.org/10.1364/BOE.4.000659 BOEICL 2156-7085 Google Scholar
G. Strangman, D. A. Boas and J. P. Sutton,
“Non-invasive neuroimaging using near-infrared light,”
Biol. Psychiatry, 52
(7), 679
–693
(2002). http://dx.doi.org/10.1016/S0006-3223(02)01550-0 BIPCBF 0006-3223 Google Scholar
D. Sanfilippo et al.,
“Design and development of a fNIRS system prototype based on SiPM detectors,”
Proc. SPIE, 8990 899016
(2014). http://dx.doi.org/10.1117/12.2037559 Google Scholar
M. Mazzillo et al.,
“Noise reduction in silicon photomultipliers for use in functional near-infrared spectroscopy,”
IEEE Trans. Radiat. Plasma Med. Sci., 1
(3), 212
–220
(2017). http://dx.doi.org/10.1109/TRPMS.2017.2679003 Google Scholar
F. Gao et al.,
“Optical tomographic mapping of cerebral haemodynamics by means of time-domain detection: methodology and phantom validation,”
Phys. Med. Biol., 49
(6), 1055
–1078
(2004). http://dx.doi.org/10.1088/0031-9155/49/6/014 PHMBA7 0031-9155 Google Scholar
F. Normandin, M. Sawan and J. Faubert,
“A new integrated front-end for a noninvasive brain imaging system based on near-infrared spectroreflectometry,”
IEEE Trans. Circuits Syst. I, 52
(12), 2663
–2671
(2005). http://dx.doi.org/10.1109/TCSI.2005.857920 Google Scholar
R. Pagano et al.,
“Improvement of sensitivity in continuous wave near infrared spectroscopy systems by using silicon photomultipliers,”
Biomed. Opt. Express, 7
(4), 1183
–1192
(2016). http://dx.doi.org/10.1364/BOE.7.001183 BOEICL 2156-7085 Google Scholar
D. G. Greenhalgh et al.,
“Temperature threshold for burn injury: an oximeter safety study,”
J. Burn Care Rehabil., 25
(5), 411
–415
(2004). http://dx.doi.org/10.1097/01.BCR.0000138295.63830.90 Google Scholar
A. Von Lühmann et al.,
“M3BA: a mobile, modular, multimodal biosignal acquisition architecture for miniaturized EEG-NIRS based hybrid BCI and monitoring,”
IEEE Trans. Biomed. Eng., 64
(6), 1199
–1210
(2017). http://dx.doi.org/10.1109/TBME.2016.2594127 Google Scholar
A. Kassab et al.,
“Functional near-infrared spectroscopy caps for brain activity monitoring: a review,”
Appl. Opt., 54
(3), 576
–586
(2015). http://dx.doi.org/10.1364/AO.54.000576 Google Scholar
D. Arifler et al.,
“Optimal wavelength combinations for near-infrared spectroscopic monitoring of changes in brain tissue hemoglobin and cytochrome c oxidase concentrations,”
Biomed. Opt. Express, 6
(3), 933
–947
(2015). http://dx.doi.org/10.1364/BOE.6.000933 BOEICL 2156-7085 Google Scholar
K. Uludaǧ et al.,
“Cross talk in the Lambert–Beer calculation for near-infrared wavelengths estimated by Monte Carlo simulations,”
J. Biomed. Opt., 7
(1), 51
(2002). http://dx.doi.org/10.1117/1.1427048 JBOPFO 1083-3668 Google Scholar
K. Uludaǧ et al.,
“Separability and cross talk: optimizing dual wavelength combinations for near-infrared spectroscopy of the adult head,”
Neuroimage, 22
(2), 583
–589
(2004). http://dx.doi.org/10.1016/j.neuroimage.2004.02.023 NEIMEF 1053-8119 Google Scholar
S. J. Matcher et al.,
“Performance comparison of several published tissue near-infrared spectroscopy algorithms,”
Anal. Biochem., 227 54
–68
(1995). http://dx.doi.org/10.1006/abio.1995.1252 ANBCA2 0003-2697 Google Scholar
A. Duncan et al.,
“Optical pathlength measurements on adult head, calf and forearm and the head of the newborn infant using phase resolved optical spectroscopy,”
Phys. Med. Biol., 40
(2), 295
–304
(1995). http://dx.doi.org/10.1088/0031-9155/40/2/007 PHMBA7 0031-9155 Google Scholar
A. Kienle and M. S. Patterson,
“Time-resolved diffusion equations for reflectance from a semi-infinite turbid medium,”
J. Opt. Soc. Am. A, 14
(1), 246
–254
(1997). http://dx.doi.org/10.1364/JOSAA.14.000246 JOAOD6 0740-3232 Google Scholar
L. Wang, S. L. Jacques and Z. Liqiong,
“MCML—Monte Carlo modeling of light transport in multi-layered tissues,”
Comput. Meth. Programs Biomed., 47
(2), 131
–146
(1995). http://dx.doi.org/10.1016/0169-2607(95)01640-F CMPBEK 0169-2607 Google Scholar
F. Scholkmann et al.,
“The effect of venous and arterial occlusion of the arm on changes in tissue hemodynamics, oxygenation, and ultra-weak photon emission,”
Adv. Exp. Med. Biol., 765 257
–264
(2013). http://dx.doi.org/10.1007/978-1-4614-4989-8_36 AEMBAP 0065-2598 Google Scholar
S. Lacroix et al.,
“Reproducibility of near-infrared spectroscopy parameters measured during brachial artery occlusion and reactive hyperemia in healthy men,”
J. Biomed. Opt., 17
(7), 077010
(2012). http://dx.doi.org/10.1117/1.JBO.17.7.077010 JBOPFO 1083-3668 Google Scholar
F. Al-Shargie et al.,
“Simultaneous measurement of EEG-fNIRS in classifying and localizing brain activation to mental stress,”
in IEEE Int. Conf. Signal Image Processing Applications (ICSIPA 2015),
282
–286
(2016). http://dx.doi.org/10.1109/ICSIPA.2015.7412205 Google Scholar
C. Herff et al.,
“Classification of mental tasks in the prefrontal cortex using fNIRS,”
in 35th Annual Int. Conf. of the IEEE Engineering in Medicine and Biology Society (EMBC),
2160
–2163
(2013). http://dx.doi.org/10.1109/EMBC.2013.6609962 Google Scholar
N. H. Kashou et al.,
“Hand-grasping and finger tapping induced similar functional near-infrared spectroscopy cortical responses,”
Neurophotonics, 3
(2), 025006
(2016). http://dx.doi.org/10.1117/1.NPh.3.2.025006 Google Scholar
V. Toronov et al.,
“Near-infrared study of fluctuations in cerebral hemodynamics during rest and motor stimulation: temporal analysis and spatial mapping,”
Med. Phys., 27
(4), 801
–815
(2000). http://dx.doi.org/10.1118/1.598943 Google Scholar
M. Caldwell et al.,
“Modelling confounding effects from extracerebral contamination and systemic factors on functional near-infrared spectroscopy,”
Neuroimage, 143 91
–105
(2016). http://dx.doi.org/10.1016/j.neuroimage.2016.08.058 NEIMEF 1053-8119 Google Scholar
H. Atsumori et al.,
“Development of wearable optical topography system for mapping the prefrontal cortex activation,”
Rev. Sci. Instrum., 80
(4), 043704
(2009). http://dx.doi.org/10.1063/1.3115207 RSINAK 0034-6748 Google Scholar
B. W. Zeff et al.,
“Retinotopic mapping of adult human visual cortex with high-density diffuse optical tomography,”
Proc. Natl. Acad. Sci. U. S. A., 104
(29), 12169
–12174
(2007). http://dx.doi.org/10.1073/pnas.0611266104 Google Scholar
D. A. Boas and M. A. Franceschini,
“Haemoglobin oxygen saturation as a biomarkerthe problem and a solution,”
Philos. Trans. R. Soc. A, 369
(1955), 4407
–4424
(2011). http://dx.doi.org/10.1098/rsta.2011.0250 Google Scholar
L. Gagnon et al.,
“Quantification of the cortical contribution to the NIRS signal over the motor cortex using concurrent NIRS-fMRI measurements,”
Neuroimage, 59
(4), 3933
–3940
(2012). http://dx.doi.org/10.1016/j.neuroimage.2011.10.054 NEIMEF 1053-8119 Google Scholar
R. Zimmermann et al.,
“Detection of motor execution using a hybrid fNIRS-biosignal BCI: a feasibility study,”
J. Neuroeng. Rehabil., 10
(1), 4
(2013). http://dx.doi.org/10.1186/1743-0003-10-4 Google Scholar
BiographyDominik Wyser received his BSc and MSc degrees in mechanical engineering from ETH Zurich, Switzerland, in 2012 and 2014, respectively. In 2014, he joined the Rehabilitation Engineering Lab at ETH Zurich, where he is currently pursuing his PhD on the topic of brain–computer interfaces, in collaboration with the Biomedical Optics Research Lab at the University Hospital Zurich. He is working on the development of an fNIRS instrument for the real-time detection of motion intention. Olivier Lambercy received his MSc degree in microengineering from the Ecole Polytechnique Fédérale de Lausanne, Switzerland, in 2005 and his PhD in mechanical engineering from the National University of Singapore in 2009. In 2009, he joined the Rehabilitation Engineering Laboratory at ETH Zurich as a senior research associate. He is associate editor of the Journal of NeuroEngineering and Rehabilitation. His research interests are in medical and rehabilitation robotics, human motor control and human-machine interaction. Felix Scholkmann received his PhD from the University of Zurich, Switzerland, in 2014. A research associate at the University Hospital Zurich (Biomedical Optics Research Laboratory, Department of Neonatology) and University of Bern, his research focuses on biomedical signal processing, biophotonics (development and application of NIRS), neuroscience, integrative physiology and biophysics. Martin Wolf is a professor of biomedical optics at the University of Zurich. He received his PhD from ETH Zurich. He heads the Biomedical Optics Research Laboratory, which specializes in developing techniques to measure and quantitatively image oxygenation of brain, muscle, tumors and other tissues. His aim is to translate these techniques to clinical application for the benefit of adult patients and preterm infants. Roger Gassert is an associate professor of rehabilitation engineering in the Department of Health Sciences and Technology at ETH Zurich. He received his MSc degree in microengineering and his PhD in neuroscience robotics from the Ecole Polytechnique Fédérale de Lausanne in 2002 and 2006, respectively. His research interests are in physical human-robot interaction, rehabilitation and neuroscience robotics, noninvasive brain–robot interfaces and assistive technology. |