Open Access
Add to BibSonomyAbstract
We have designed a versatile, multi-channel near-infrared spectrophotometry (NIRS) instrument for the purpose of mapping neuronal activation in the neonatal and adult brain in response to motor, tactile, and visual stimulation. The optical linearity, stability, and high signal to noise ratio (>70 dB) of the instrument were demonstrated using an in vitro validation procedure. In vivo measurements on the adult forearm were also performed. Changes in oxygenation, induced by arterial occlusion of the forearm, were recorded and were shown to compare well with measurements acquired using a conventional NIRS instrument. To demonstrate the capabilities of the instrument, functional measurements in adults and neonates were performed. The instrument exhibited the capability to differentiate with a spatial resolution in the order of cm, local activation patterns associated with a finger tapping sequence.
©2005 Optical Society of America
1. Introduction
Brain activity can be assessed by near-infrared spectrophotometry (NIRS). Two major types of signals can be discriminated.
• The fast neuronal signal, which relies on optical changes directly associated with neuronal activity. This signal arises within milliseconds after the onset of the brain stimulation [1–5]. Neuronal activity leads to small changes presumably in the light scattering properties of the neurons [6]. A typical response to a functional stimulus is expected within 200 ms.
• The slow hemodynamic signal, which relies on an increased blood perfusion of an area of activated neurons due to neurovascular coupling. Brain activity leads to an increase in the local oxygen consumption followed closely by an increase in blood flow, which in turn changes the hemoglobin concentrations and oxygenation. These changes occur within a few seconds after the onset of the stimulation [7–14].
NIRS currently offers the only available means of detecting both neuronal activation and concurrent changes in blood flow. EEG and MEG can only measure the fast neuronal signal, while PET and fMRI assess only the slow hemodynamic signal.
The aim of this study was to build a novel NIRS instrument capable of measuring both the fast neuronal and slow hemodynamic signals simultaneously. For this purpose we have defined the following requirements:
1. Low instrument noise and high time resolution (100Hz). This requirement stems from the observation that activation induced changes in optical properties are both small, relative to normal bulk fluctuations in optical properties, and extremely short lived. The instrument must therefore introduce as little noise as possible and must also provide measurements at extremely high temporal resolution.
2. Spatial sampling at numerous locations (imaging). In adults, functional activation is known to occur in very localized areas [2]. In order to avoid missing the activation signal, due to anatomical differences between individuals, the instrument must therefore be capable of simultaneously measuring, in a spatially resolved manner, a large tissue area.
3. Good clinical usability and ease of transportation. In a clinical environment it is essential to have an instrument which does not require darkening the room or excessive light shielding, and which has practical and easily attachable sensors. Furthermore bedside applicability is one of the key features of NIRS. E.g. fMRI and PET require transport of the patient, which, in critically ill patients is not always feasible.
This paper describes the novel NIRS system with multiple channels and a high sensitivity, which allows for monitoring of small and spatially resolved changes in optical properties of the cerebral tissue due to cortical activation in response to a sensory stimulus. The novelty of this instrument is that it fulfills all three requirements. Furthermore, although it is well suited for measurements in adults, it is also particularly appropriate for neonates.
We envision a broad area of clinical application, e.g. in adult and neonatal intensive care, where transport of the patient is risky and an assessment of the function of the brain particularly important, but also in psychiatry, neuroscience, neurology and many other fields.
2. Instrumentation
2.1 Overview
A block diagram of the functional NIRS system is shown in Fig. 1. The system consists of sensors, which are attached to the tissue of interest. The data acquisition unit controls the sensors and measurement. The stimulation unit generates the stimuli, which are highly synchronized with the data acquisition. The PC records, visualizes, and processes the acquired data.
2.2 Sensors
Special attention was paid to the design and development of the sensor to ensure safe and easy application on the soft and sensitive neonatal head. In place of glass fibers and bulky photo multiplier tubes, often employed to transmit and detect NIR light, we have incorporated emission and detection of NIR light on the sensor. The sensor uses light emitting diodes (LED) as light sources with wavelengths of 730, 770 and 805nm. Although LEDs have a broader spectrum (50% spectral width ≈25nm) than laser diodes, they have the following advantages: LEDs can be built into a sensor (distance between LEDs 1.5mm), they are less dangerous for the eye, which is important in a clinical environment, they have a higher dynamic range of the driving current (no threshold current) and are inexpensive. The output power is adjusted to the optical density of the tissue, the peak optical power of a pulse is 25mW, which leads to an average power of <1mW. To increase the sensitivity, a silicon PIN photodiode with a large active area of 7.5mm2 was chosen as a detector. Such a detector has the advantage that it cannot be damaged by excessive light, as is the case for e.g. photomultiplier tubes.
A major concern was ambient electromagnetic noise coupling into the sensor signal. To reduce this effect, a transimpedance amplifier (first amplifier stage) and a differential amplifier (second amplifier stage) were mounted close to the PIN photodiode on a fingernail sized printed circuit board.
The sensor was molded with a curvature which corresponds to the average curvature of a neonatal head. The use of medical grade soft silicon and of flexible cables avoids pressure on the skin, which prevents discomfort during patient movement. Since the sensor is flexible, it can also be used for the adult head. The interoptode distance was 25mm. A distance of 25mm is also appropriate for the adult head. To prevent direct leakage between the emitter and detector, the silicon and several surface mountable devices, of the PCB design, were employed as optical barriers. Transparent silicon was molded around the diodes to provide two optical windows. Eight sensors can be simultaneously placed on the head to allow regional mapping. The light of one LED can be measured by two detectors simultaneously. The instrument enables different interoptode distances and geometrical arrangements and we plan to build a variety of sensors for specific purposes.
The maximum voltage on the sensor is +/- 12Volt. The maximum driving current is 400mA. The electronics are fully isolated by the sensor mold and the instrument is electrically separated by an isolation transformer which conforms to the CE standards.
Fig. 1. The block diagram shows the data acquisition unit, the stimulation unit, and the brain mapping sensors. The amplified analog signals from the brain mapping sensors are filtered and digitized. The data acquisition processor controls the measurement by switching the stimuli devices and selecting the photodiode — LED pairs of each sensor in a time multiplexed fashion. The communication processor transfers the acquired data to the PC-notebook. Both devices use Linux as an operating system.
2.3 Data acquisition unit
A dual processor architecture controls the measuring system. One processor performs the sampling process by switching the different wavelengths in a time multiplexed fashion shown in Fig. 2, while performing a 16 x oversampling of the currently selected detector - wavelength time slot. Within a given time slot the sampling process starts after a predefined delay, allowing the LEDs to stabilize on a programmable constant emission intensity. In order to allow an optimal signal level at the analog to digital converter (ADC), a 12 bit digital to analog converter (DAC) is used to adjust the LED intensity by an LED current control loop.
The instrument is capable of acquiring two data channels at the same time through its two incorporated ADCs. The ADCs have a 16 bit resolution with an integral nonlinearity error of +/- 2 least significant bits. In principle the dynamic range of the system is 21 bits, because of the oversampling and programmable amplifiers. The time resolution is t3-t1=10ms, i.e. 100Hz (Fig. 2), i.e. all wavelengths and positions are sampled every 10ms. All of the data sampling is fully synchronized to a 2.4kHz clock signal derived from the 50Hz power line signal, allowing a 2×24 channel data acquisition. Four data channels record the current state of the various stimulation devices.
The second processor, running with uClinux (a Linux based embedded operating system), provides network communication capability to the system. The raw data are transferred by a 10-Base-T network to a PC Notebook for further data processing, analysis, and storage.
Fig. 2. Data are sampled in a time multiplexed fashion. The LEDs are switched on and off in a sequence which is predefined in a device initialization script.
2.4 Stimulation unit
The stimulation unit consists of two ultra miniature vibration motors for tactile stimulation, a pair of goggles with integrated LEDs for visual stimulation, arranged as four bars of three LEDs each, and a pair of headphones connected to a compact disk player for auditory stimulation. All stimuli can be switched on or off with a time resolution of 100Hz and synchronized by the data acquisition unit, which allows correlation of the fast stimuli pattern with physiological responses.
2.5 Graphical User Interface application
A JAVA application facilitates the set-up of the data acquisition and stimulation unit. It initializes the instrumentation, records the data, and visualizes the data in real time. Several postprocessing applications are implemented to average and filter the data. The application is platform independent.
3. Theory
3.1 Calibration model
A continuous wave light emitting technique in conjunction with a two compound model, derived from the modified Beer-Lambert law Eq. (1), was used [15,16]. With this method, changes in O2Hb and HHb concentration were calculated from changes in optical density (OD).
The spectrum of all LEDs was measured with a spectrophotometer (USB2000, Ocean Optics Inc., Dunedin, USA). We accounted for the spectrum of the LEDs and for the spectral sensitivity of the PIN photodiode in our physical model.
The emitted intensity, I0, from the LED is attenuated by tissue and the transmitted intensity, I, detected at the photodiode. The process is dependent on the geometric factor G, the specific absorption coefficient α, the concentration c of the compound, and the interoptode distance d. The interoptode distance d is corrected by the differential pathlength factor DPF which takes into account the effect of multiple scattering in tissue, a phenomenon that leads to an increased mean optical pathlength.
The characteristic spectrum of the LED intensity is included by introducing a wavelength dependence.
The intensity integrated over the whole spectrum of interest is expressed in Eq. (3).
The raw ADC values reflect the intensity proportional to the photocurrent flowing through the photodiode as well as the integral over the spectral sensitivity of the photodiode, which is also wavelength dependent.
Equation (4) can be derived from the spectral expansion of the law (2) by (3). The measuring system function SDet depends on the efficiency of the detector at a given wavelength, gain of the amplifier, and the resolution of the ADC. SDet is constant during the entire measurement. By subtracting the ADCLedOff ambient light from the ADCLedOn intensity, only the light incident on the detector, emitted by the LED, is taken into account. Thus measurements are carried out without darkening the room or excessive light shielding, which is essential in the clinical environment. The product of the terms SDet(λ) and I0(λ) can be substituted in good approximation by one Gaussian function using the experimentally measured spectral characteristics of the LEDs and sensitivity profile of the photodiode manufacturer specification datasheet.
The peak intensity amplitude Î0, center wavelength λ0 and spectral half width, σ, were fitted over the characteristics of the LED emission spectrum and shaped by the spectral sensitivity of the PIN photodiode.
For simplification of the terms, the following definitions are introduced in Eq. (7) for the transfer function hα,λ0(c) and (8) for the light attenuation ADCld.
In Eq. (9) the logarithm of Eq. (6) is rewritten by introducing substitutes (7) and (8).
The evaluation of the transfer function does not depend on the absolute intensity, I0, of the LED but on the shape of the Gauss function of the LED. Equations (1–8) correspond to the standard approach used in NIRS [17]. Due to the nonnegligible 50% spectral width of ≈25nm of the LEDs, the theory has to be adapted for our instrument.
The transfer function hα,λ0(c) can be extended to hα1,α2,λ0(c1,c2) to take into account multiple compounds such as O2Hb and HHb.
The partial derivative of hα1,α2(c1,c2) with respect to c1 and c2 gives the slope of the curve as a function of the concentration changes. The numerical evaluation of a two compound model, using O2Hb and HHb changes within the physiological range (±30µmol/l), shows a sufficiently constant slope (linear regression fit r>0.999).
Equation (11) represents the first order Taylor approximation starting at the baseline concentrations c10,c20.
Relative concentration changes can be calculated without the knowledge of Î0 and G by subtracting a baseline value ADCld(t0).
Measuring with two or three wavelengths enables one to obtain the concentration changes, taking absorption coefficients from the literature [17,18] to solve Eq. (14). A matrix can be defined with the partial derivative of the numerically evaluated integral.
The resulting simplified form is shown in Eq. (15).
The standard Eq. (16) can be solved, provided det(HTH)>0, to obtain a unique solution.
4. Experiments and results
The aim of the experiments was to assess the new system in vitro to ensure the technical performance and in vivo to demonstrate the accuracy and its capability to detect functional activation.
4.1 In vitro system performance
System performance was evaluated to determine the technical ability to detect the functional signals in the brain:
1. For the accurate detection of slow hemodynamic signal, the system requires a low drift within a period of one minute, which corresponds to a typical experimental period of stimulation and rest, low noise at a time resolution of approximately 10Hz (slow system noise), and a high linearity in order to quantify the concentration changes in O2Hb and HHb precisely.
2. For the detection of fast neuronal signal the most relevant parameter is the noise level at high time resolution (fast system noise).
To test for drift and slow and fast system noise, we placed the brain mapping sensor on a head phantom and fixed it with a bench vice to eliminate motion artifacts. The phantom (Johnson & Johnson Medical GmbH, Newport, UK) consisted of a transparent silicone rubber with a scatterer (TiO2) and an absorber (Zeneca, Manchester, UK): µs′≈1.7 mm-1, µa≈0.05 mm-1. Following a cold start of the entire system, data were acquired for 120 minutes. Immediately after a restart of the system, data were recorded for a further 120 minutes (warm start). A difference between cold start and warm start of the device is expected to be caused by the temperature drift of the LEDs and amplifier circuits.
Drift: Since the applied method calculates relative changes in O2Hb and HHb from raw signals, it is important that the zero drift and the fluctuation of the baseline are smaller than the expected changes due to the slow hemodynamic signal. The system drift over the first 120 minutes after warm and cold start are displayed in terms of concentration changes in HHb, O2Hb and tHb in Table 1. The drift is much lower than the expected slow hemodynamic signal (0.5µmol/l).
Table 1. Warm and Cold Start drift
The mean system drift over 1 min and the range of slow standard deviation (system noise) were measured after warm and cold start of the instrument over the full 120 minute interval (sampling rate 10Hz). The reference period of 1 min was chosen because a typical stimulation interval would be 1 min or less. The drift and SD is small compared to the expected functional changes of 0.5µmol/l.
Slow system noise: A time resolution of 10Hz is sufficient to assess the slow hemodynamic signal. Ten samples were averaged to determine the slow system noise. The standard deviation over one minute (600 samples) was calculated. For drift analysis each block of ten consecutive ADC values was averaged and transformed to obtain O2Hb and HHb concentration changes (10Hz resolution). Drift analysis was performed on the two recorded data sets. The range of the slow system noise is displayed in Table 1 and is much lower than the expected signal.
Fast system noise: The fast system noise was calculated as standard deviation within each block of 10 samples at the original sample rate of 100Hz. The results were expressed in percent of the mean light intensity. Table 2 shows the mean and the standard deviation of the fast system noise over the entire 120 minute warm start data set. These results demonstrate that the system drift and the noise are very low. In addition, the noise equivalent power was measured to be in the order of 100 pW.
Table 2. Instrument Noise
For the fast system noise analysis of the data acquisition hardware, the 120 min warm drift data were analyzed at full data sample speed (100Hz). The noise is given in % of the mean intensity.
Linearity: We tested the linearity of our instrument on a special phantom with adjustable optical density (OD) on which the brain mapping sensor was placed and fixed with a bench vice. The phantom consisted of two blocks of a medium with tissue-like scattering and absorption properties (50mm cube), which were separated by a black light blocking layer. In the middle of the layer was a slide with drill holes with increasing diameter, which allowed the light to penetrate from one side to the other. The amount of light could be adjusted by shifting the slide to four discrete positions and at each position only one optical hole was open. OD values were recorded for every diameter and averaged over a 10 second interval. Linear regression of all three wavelengths was performed against the reference OD values. The results demonstrate that the MCP II is linear to within ±3% over a large range of ~0.6OD.
4.2 In vivo arterial occlusion on the adult arm
Arterial occlusions were performed to validate the measuring system in vivo. The measurements were done on five healthy adult volunteers. Eight brain mapping sensors were attached in two rows of 4 sensors on the left forearm of the subject. In addition a NIRO-300 (Hamamatsu Photonics Deutschland GmbH, Germany) near infrared spectrophotometer was attached on the forearm for comparison of the in vivo measurements. A pneumatic pressure cuff (Kontante II, TRIMED AG, Bochum, Germany) was used to induce arterial occlusion.
The following procedure was repeated five to seven times: after a stable baseline for one minute the cuff was inflated to 300mmHg for one minute (arterial occlusion), followed by a two minute rest period with the cuff pressure released. The whole procedure was repeated after swapping positions of the brain mapping sensors and of the NIRO-300 sensor. Both instruments measured the concentration changes ΔO2Hb, ΔHHb, and ΔtHb during the procedure, which were recorded on a PC notebook. The protocol was approved by the ethical committee of our institution.
The slopes of O2Hb, HHb, and tHb during the arterial occlusion, recorded with both instruments, were analyzed and compared (mean and SD). The NIRO-300 was configured to measure one set of data every 2 s. For the MCP II, 200 samples were averaged to achieve the same time resolution.
A total of 70 arterial occlusions were performed with each instrument. The raw data was visually preprocessed and 7 occlusions were excluded from the analysis due to improper arterial occlusion (slope did not become constant while cuff was inflated) or motion artifacts (curve not continuous). A comparison of the measured concentration changes during arterial occlusion between the NIRO 300 and the MCP II (median of the 8 identical probes) is shown in Table 3. Figure 3 depicts the concentration changes in HHb, O2Hb, and tHb during four consecutive arterial occlusions in one subject.
4.3 In vivo functional measurement in an adult subject
Finger tapping exercises were performed by one right handed adult volunteer and recorded by the MCP II device at 100Hz. Four sensors where placed over the motor cortex of each hemisphere at the C3 and C4 positions according to the international EEG 10/20 system [19]. Twenty finger tapping exercises were performed. Each exercise consisted of a 20 second tapping period followed by a 10 second rest period. The two hands were alternated in performing the tapping exercise.
Fig. 3. Arterial occlusions were performed on the left arm of a subject at 1, 5, 9, and 13 minutes as indicated by the black bars. During the occlusion, the oxyhemoglobin (O2Hb) concentration decreases and the deoxyhemoglobin (HHb) concentration increases. The magnitude of the concentration changes is similar for both instruments, although the trace is much noisier for the NIRO-300. Both devices had a time resolution of 2s in this setting. (MCP II (top), NIRO-300 (bottom))
Table 3. Instrument Comparison
A total of 63 arterial occlusions were analyzed on 5 healthy volunteers with a. During the occlusion, oxyhemoglobin (O2Hb) concentration decreases and the deoxyhemoglobin (HHb) concentration increases in a ramp function (Fig. 4). The mean values and the standard deviations of the arterial occlusion slopes were calculated.
Data processing was performed with a Matlab script to remove heart beat, breathing or vasomotion artifacts, and trends not associated with brain activity (physiological trends and instrument drift). For this purpose, the recorded oxygenation changes were low pass filtered (finite duration impulse response filter of 644th order, cut-off frequency at 0.5Hz) and a symmetrical moving average over one minute was subtracted from the filtered data (to remove constant component and trend). To suppress signals not correlated to the functional stimulation, a time triggered average was performed over the repeated stimulation and resting periods [8].
Figure 4 shows the time triggered average of HHb and O2Hb concentration changes during left hand and right hand finger tapping exercises respectively. The left column in each figure shows the sensors 1–4 placed over the left hemisphere (C3 area). Sensors 5–8 are placed over the C4 area on the right hemisphere (right column). Typical [7] activation patterns were observed with a high degree of localization. These measurements were repeated and activations pattern were observed in 4 out of 5 subjects. Probably in the one subject, the sensor was not precisely located on the motor cortex.
4.4 In vivo functional measurement in neonates
Due to the sophisticated sensor design, our protocol of functional measurements does not require sedation of the infant. One infant with a gestational age of 37 6/7 weeks was measured at six days postnatal age. Four sensors where placed over the C3 area (international 10/20 system, 1st sensor frontal, 4th sensor occipital). A tactile stimulation was performed using two miniature vibration motors attached to each palm. Each motor was switched on for 20 seconds followed by a 10 second rest period. The two palms were stimulated alternately. Similar analysis techniques described above for functional measurements in adults were employed to remove heart beat, breathing or vasomotion artifacts, and trends not associated with neuronal activity.
Fig. 4. Finger tapping exercises were performed from second 10 to 30 (dotted box). The first (sensors 1–4) and the second column (sensor 5–8) measured over left and right hemisphere respectively before, during, and after left hand finger tapping. The third (sensors 1–4) and fourth column (sensor 5–8) show the left and right hemisphere during right hand finger tapping. An activation typically consists of an increase in oxyhemoglobin (O2Hb) concentration and of a decrease in deoxyhemoglobin (HHb) concentration. The higher O2 consumption in the activated area is immediately overcompensated by an increase in blood flow, which leads to the observed pattern. Both hands showed a stronger contralateral activation on the motor cortex hemisphere. The ordinates are scaled to µmol/l.
A 40×10 pixels matrix was used to linearly interpolate the contribution of each sensor’s location (at middle of each 10×10 pixels field). A second Matlab script was used to extract 5 Hz image frames of this 40×10 pixels image. The frames were compressed and stored in an AVI file.
Figure 5(a) shows a video of the time triggered average of the changes in HHb and O2Hb concentration during functional activation in a neonate. The corresponding time traces are displayed in Fig. 5(b).
In the clinical application the sensor was well tolerated by the neonates, as it can be applied and removed quickly and easily and no calibration is required.
5. Discussion
5.1 In vitro system drift and signal to noise analysis
The concentration changes in O2Hb, HHb, and tHb in the infant’s head during functional stimulation were expected to be approximately 0.5µmol/l [8, 20, 21]. We took this value as a reference to determine the relevance of the drift and noise.
In a typical stimulation interval of up to 1 minute, our results show that the system drift after a warm start was much smaller than 0.5µmol/l, and that the drift, therefore, does not need to be taken into account. As the system drift is a monotonic process, detrending by a moving average will further decrease the influence of the system drift.
The system drift after cold start, however, is relevant. In practice it is not a problem to warm up the device two hours before starting a measurement.
The noise was different in magnitude between the three LEDs types. We therefore conclude that the dominant source of noise in the system is the LED intensity fluctuation. With smaller ADC values, the quantification error of the linear ADC increases. This is especially true for low light conditions on a very dense object. Since the LED noise is found to be dominant over the quantification error, we conclude that the 16 bit ADC provides a satisfactory resolution at a high converter speed.
In addition and due to the superposition of slow vasomotion [22, 23] on functional hemodynamic changes, it is necessary to repeat a functional stimulation at least 10 times. The triggered averaging reduces instrumental and physiological noise by a factor of the square root of 10.
Consequently, the implemented steps to minimize electrical noise, such as integration of the transimpedance amplifier in the sensor and sophisticated PCB design for all functional blocks, power supply decoupling (switched current of LED drivers, digital and analog part), and power supply filtering, have been successful and have enabled visualization of an activation pattern with high signal to noise ratio.
5.2 In vitro system linearity
The measured deviation of the steepness would result in an error of less than 2.3% for O2Hb and 6.4% for HHb, which is negligible for physiological studies.
5.3 In vivo arterial occlusion on the adult arm
To achieve a sufficient arterial occlusion the arm had to be occluded with a cuff pressure of 300mmHg. Such pressures cause displacement of tissue, a phenomenon visible as artifacts in the measurement. As shown in Fig. 3, the slope of the HHb increase, with respect to the observed decrease in O2Hb, was not a perfectly constant slope at the beginning of the occlusion. As expected, the slope remained constant once the maximal pressure was established and the displaced tissue was steady. The results show a good agreement between the two instruments.
Fig. 5. Tactile stimulation was performed from second 5 to 25. An activation typically consists of an increase in oxyhemoglobin (O2Hb) concentration and of a decrease in deoxyhemoglobin (HHb) concentration. The higher O2 consumption in the activated area is immediately overcompensated by an increase in blood flow, which leads to the observed pattern. Only the contra-lateral hemisphere (right column) showed a strong activation on the somatosensory cortex. a) (Top) The color palettes are scaled to +/-0.55µmol/l (size of the videos is 1.1MB each) [Media 1, Media 2]. b) (Bottom) The time traces are shown below.
5.4 In vivo functional measurement in an adult subject
Consistent with results from the literature, the subject showed a strong contralateral activation [8, 24, 25]. During left hand finger tapping (Fig. 4, left half) the left hemisphere showed strong activation on sensor 3 and 4 while the other sensors showed only a weak activation.
The right hemisphere showed a strong activation on all sensors. A decrease in HHb with a simultaneous increase in O2Hb during the activation indicates an increased blood flow over the motor and somatosensory brain areas on the center over sensor 7.
During right hand finger tapping (Fig. 4, right half) a more differentiated activation pattern was observed. Sensor 3 over the left hemisphere showed a congruous pattern of response during left and right hand finger tapping, while the contralateral response was 1.5 times larger in magnitude. The pattern exhibits a high degree of localization, since it is only visible in one sensor. On sensors 2, 3, 4, 5, 6, 7, 8 both hemispheres showed an activation of about 4 seconds before the right hand finger tapping exercise was started. This premotor activation, which can be observed at approximatively the 10th second in the plot may be explained by a more functionally differentiated right hand.
5.5 In vivo functional measurement in neonates
Localized brain activation is visible on the contralateral side of the stimulation.
6. Conclusion
The present study demonstrates the operation principle and validity of a new multi-channel NIRS system. This new brain mapping device shows an excellent signal to noise ratio with a concomitant high time resolution. From a technical point of view, this device has the potential to provide the clinician with useful information regarding the extension and the progress of brain lesions in adults and neonates undergoing intensive care or in other clinical settings. The video best displays the potential of the new instrument.
Acknowledgments
The authors thank Thomas Schaerer, Dr. Thomas von Hoff, and Dr. Kurt von Siebenthal for their insightful comments, Tom Walsh and Dr. Alessandro Rubini for technical assistance, and Nikolaus Correll for general assistance.
References and links
1. G. Gratton, M. Fabiani, P. M. Corballis, D. C. Hood, M. R. Goodman-Wood, J. Hirsch, K. Kim, D. Friedman, and E. Gratton, “Fast and localized event-related optical signals (EROS) in the human occipital cortex: comparisons with the visual evoked potential and fMRI,” Neuroimage 6, 168–80 (1997). [CrossRef] [PubMed]
2. M. Wolf, U. Wolf, J. H. Choi, R. Gupta, L. P. Safonova, L. Paunescu, A. Michalos, and E. Gratton, “Functional frequency-domain near-infrared spectroscopy detects fast neuronal signal in motor cortex,” Neuroimage , 17, 1868–75 (2002). [CrossRef] [PubMed]
3. J. Steinbrink, M. Kohl, H. Obrig, G. Curio, F. Syre, F. Thomas, H. Wabnitz, H. Rinneberg, and A. Villringer. “Somatosensory evoked fast optical intensity changes detected non-invasively in the adult human head.” Neurosci Lett 15, 105–108 (2000). [CrossRef]
4. M. Wolf, U. Wolf U, J. H. Choi, V. Toronov, L. A. Paunescu, A. Michalos, and E. Gratton. “Fast cerebral functional signal in the 100-ms range detected in the visual cortex by frequency-domain near-infrared spectrophotometry.” Psychophysiology 40(4), 521–528 (2003). [CrossRef]
5. M. A. Franceschini and D. A. Boas. “Noninvasive measurement of neuronal activity with near-infrared optical imaging.” Neuroimage 21(1), 372–386 (2004). [CrossRef]
6. I. Tasaki. “Rapid structural changes in nerve fibers and cells associated with their excitation processes,” Jpn J Physiol 49, 125–38 (1999) [CrossRef] [PubMed]
7. A. Villringer and B. Chance, “Non-invasive optical spectroscopy and imaging of human brain function,” Trends Neurosci 20, 435–42 (1997). [CrossRef] [PubMed]
8. M. Wolf, U. Wolf, V. Toronov, A. Michalos, L. A. Paunescu, J. H. Choi, and E. Gratton, “Different time evolution of oxyhemoglobin and deoxyhemoglobin concentration changes in the visual and motor cortices during functional stimulation: a near-infrared spectroscopy study,” Neuroimage 16, 704–12 (2002). [CrossRef] [PubMed]
9. J. Ruben, R. Wenzel, H. Obrig, K. Villringer, J. Bernarding, C. Hirth, H. Heekeren, U. Dirnagl, and A. Villringer, “Haemoglobin oxygenation changes during visual stimulation in the occipital cortex,” Adv Exp Med Biol 428, 181–7 (1997). [CrossRef] [PubMed]
10. J. H. Meek, M. Firbank, C. E. Elwell, J. Atkinson, O. Braddick, and J. S. Wyatt, “Regional hemodynamic responses to visual stimulation in awake infants,” Pediatr Res 43, 840–3 (1998). [CrossRef] [PubMed]
11. K. Sakatani, S. Chen, W. Lichty, H. Zuo, and Y. P. Wang, “Cerebral blood oxygenation changes induced by auditory stimulation in newborn infants measured by near infrared spectroscopy,” Early Hum Dev 55, 229–36 (1999). [CrossRef] [PubMed]
12. W. N. Colier, V. Quaresima, R. Wenzel, M. C. van der Sluijs, B. Oeseburg, M. Ferrari, and A. Villringer, “Cortical blood oxygenation changes in the left and right occipital area induced by selective visual stimuli in humans,” Adv Exp Med Biol 471, 35–41 (1999). [CrossRef]
13. W. N. Colier, V. Quaresima, B. Oeseburg, and M. Ferrari, “Human motor-cortex oxygenation changes induced by cyclic coupled movements of hand and foot,” Exp Brain Res 129, 457–61 (1999). [CrossRef] [PubMed]
14. M. Bartocci, J. Winberg, C. Ruggiero, L. L. Bergqvist, G. Serra, and H. Lagercrantz, “Activation of olfactory cortex in newborn infants after odor stimulation: a functional near-infrared spectroscopy study,” Pediatr Res 48, 18–23 (2000). [CrossRef] [PubMed]
15. F. F. Jobsis, “Noninvasive, infrared monitoring of cerebral and myocardial oxygen sufficiency and circulatory parameters,” Science 198, 1264–7 (1977). [CrossRef] [PubMed]
16. S. J. Matcher, C. E. Elwell, C. E. Cooper, M. Cope, and D. T. Delpy, “Performance comparison of several published tissue near-infrared spectroscopy algorithms,” Anal Biochem , 227, 54–68 (1995). [CrossRef] [PubMed]
17. S. Wray, M. Cope, D. T. Delpy, J. S. Wyatt, and E. O. Reynolds, “Characterization of the near infrared absorption spectra of cytochrome aa3 and haemoglobin for the non-invasive monitoring of cerebral oxygenation,” Biochim Biophys Acta 933, 184–92 (1988). [CrossRef] [PubMed]
18. W. G. Zijlstra, A. Buursma, and W. P. Meeuwsen-van der Roest, “Absorption spectra of human fetal and adult oxyhemoglobin, de-oxyhemoglobin, carboxyhemoglobin, and methemoglobin,” Clin Chem , 37, 1633–8 (1991). [PubMed]
19. H. Jasper, “Report of Committee on the Methods of Clinical Examination in Electroencephalography,” Electroencephalogr Clin Neurophysiol , 10, 370–375 (1958). [CrossRef]
20. P. Zaramella, F. Freato, A. Amigoni, S. Salvadori, P. Marangoni, A. Suppjei, B. Schiavo, and L. Chiandetti, “Brain auditory activation measured by near-infrared spectroscopy (NIRS) in neonates,” Pediatr Res 49, 213–9 (2001). [CrossRef] [PubMed]
21. P. Wobst, R. Wenzel, M. Kohl, H. Obrig, and A. Villringer, “Linear aspects of changes in deoxygenated hemoglobin concentration and cytochrome oxidase oxidation during brain activation,” Neuroimage 13, 520–30 (2001). [CrossRef] [PubMed]
22. K. von Siebenthal, J. Beran, M. Wolf, M. Keel, V. Dietz, S. Kundu, and H. U. Bucher, “Cyclical fluctuations in blood pressure, heart rate and cerebral blood volume in preterm infants,” Brain Dev , 21, 529–34 (1999). [CrossRef] [PubMed]
23. H. Obrig, M. Neufang, R. Wenzel, M. Kohl, J. Steinbrink, K. Einhaupl, and A. Villringer, “Spontaneous low frequency oscillations of cerebral hemodynamics and metabolism in human adults,” Neuroimage 12, 623–39 (2000). [CrossRef] [PubMed]
24. H. Obrig, C. Hirth, J. G. Junge-Hulsing, C. Doge, T. Wolf, U. Dirnagl, and A. Villringer, “Cerebral oxygenation changes in response to motor stimulation,” J Appl Physiol 81, 1174–83 (1996). [PubMed]
25. C. Hirth, H. Obrig, K. Villringer, A. Thiel, J. Bernarding, W. Muhlnickel, H. Flor, U. Dirnagl, and A. Villringer, “Non-invasive functional mapping of the human motor cortex using near-infrared spectroscopy,” Neuroreport 7, 1977–81 (1996). [CrossRef] [PubMed]
