2.1.

Setup of ICCD Camera-Based Imaging System

The imaging system (Fig. 1) allows for measurement of diffusely reflected light from a tissue in reflectance geometry. The tips of nine emitting fibers of a diameter of 200 μm (NA = 0.22) are located on the ring of a diameter of 5 cm around the imaged area. The fibers are independently spring mounted; therefore, they can be easily located on the rough structure of a human head (see Fig. 2).

Fig. 1

Setup of the imaging system. MaiTai—Ti:sapphire pulsed laser source, ICCD—time-gated, intensified CCD camera.

Fig. 2

Emitting fibers located on the ring, spring mounted independently.

The fibers were used to deliver femtosecond laser pulses ([full width at half maximum FWHM < 100 fs]) generated by MaiTai laser (SpectraPhysics, Irvine CA, USA) to the studied object (phantom or tissue). The laser was tuned to a wavelength of 780 nm. The wavelength was selected considering our previous studies carried out with indocyanine green as a contrast agent.²³ It should be noted that because of using only one wavelength, the measurement does not allow for estimation of changes in concentrations of oxy- and deoxyhemoglobin. However, the measured signals are sensitive to changes in oxygenation because the chosen wavelength did not match with isosbestic point.²⁴ Additionally, the availability of stable laser pulses was considered together with the optimal sensitivity of the ICCD camera. The laser operated at an 80-MHz repetition rate, however the delay line DEL350 (Becker&Hickl, Berlin, Germany) was limited to 40 MHz; thus, the system was triggered with every second laser pulse. One of two standard MaiTai trigger outputs, electronically divided into 40 MHz was used. The power of the laser was 0.85 W; however, power at the tips of the fibers was limited to 20 mW. During the measurements, the source position was sequentially switched between nine locations on the phantom or tissue using a fast optomechanical switch (Piezosystem Jena, Jena, Germany) with switching time of <5 ms. For each position of the source, images of diffusely reflected light were recorded using time-gated ICCD camera (PicoStar HR, LaVision, Goettingen, Germany). The measured FWHM of the instrumental response function (IRF) of the setup is ∼350 ps. The IRF includes dispersion of light in the emitting fibers, jitters of the electronics, and width of the time window of the camera. This width of IRF is better than reported by Selb et al. in a similar optical setup,²⁵ but it also does not diverge significantly from time-resolved brain-monitoring systems reported by various authors.^{26, 27, 28}

The camera is equipped with an image intensifier containing a microchannel plate (MCP) and electronic shutter between the photocathode and the MCP (Fig. 3). During the whole experiment, a high potential at the MCP is applied. Time gating is realized by electronic shutter and high-rate imager (Kentech Instruments Ltd., Wallingford, UK). The camera collects diffusely reflected photons within a 200-ps-long time window at defined delays in respect to the laser pulse. Late time windows are of special interest because the photons collected at these time windows have higher probability of penetrating into deeper tissue compartments. Because the number of photons collected at late time windows is low, a high gain on MCP must be applied. However, the massive number of early photons arriving at early time windows should not be detected.

Fig. 3

Scheme of the image intensifier with electronic shutter.

The key element of the ICCD camera is the electronic shutter, which provides acceleration of the electrons from the photocathode into the MCP only when the trigger is applied (when the potential between U _cathode and Gnd is negative). Therefore, one can “filter” a dangerously high number of early photons and amplify only late ones. Thus, the measurement procedure does not lead to damage of the MCP. However, it should be noted that the photocathode is deteriorated during the experiment and the large number of early photons must be converted to electrons, which limits the lifetime of the photocathode.

2.2.

Instrument Calibration

In the optomechanical switch (Fig. 1), different optical pathlengths for each emitting fiber were observed. This phenomenon was very critical, especially considering the measurements for a defined time window. It was necessary to provide that time windows for different emitting fibers were defined at exactly the same temporal position in respect to the laser pulse. Therefore, the calibration of the optomechanical switch was necessary.

We measured IRF for each of the emitting fibers and evaluated the temporal position of the maximum of probability. Thus, the time delays characterizing selected fibers were obtained. They varied up to 0.9 ns. The controlling software was enhanced by adding a calibration procedure based on adjustments of the delay line for every fiber separately. During changes of the switch position, the delay is changed according to an obtained calibration matrix containing measured temporal delays in order to adjust the position of the time window.

Relatively low dynamic range of the ICCD causes that the acquisition time needs to be adjusted for different delays of time windows with respect to the laser pulse. In order to achieve high signal-to-noise ratio it is impossible to measure different time windows at the same photon collection time. Images for the whole measured time range for a homogenous medium with optical properties similar to these that can be obtained in human tissue were acquired. For each time window, the number of photons was calculated. These numbers of photons acquired at different time windows were used for determination of acquisition time for different time windows.

2.3.

Image Acquisition and Processing

The measurement was controlled with in-house software developed in the DaVis environment (LaVision, Germany). The software controls the whole process of image acquisition: defining parameters of the acquisition, defining the time window's position with the use of a delay line, and controlling the process of switching the positions of the source. To control the optomechanical switch, the script, written in C++, must have been executed in DaVis, which allowed us to send switching commands via the Line Print Terminal (LPT) port. Images are collected with a resolution of 640 × 480 pixels and stored on the hard disk of a PC in 12-bit RAW format. The CCD's pixel size on the sample is ∼0.15 mm.

As shown in Fig. 4, images of light diffusely reflected from the phantom were collected for each of nine emitting fibers. Sequential switching of the laser light to the single source positions was carried out, and corresponding images of the area inside the ring formed by source fibers were acquired at defined time window. In the physical phantom, studies for each position of the emitting fiber of the background image was subtracted (see Fig. 5). The background images were obtained during measurement carried out in a homogenous medium. Images obtained for each of nine source positions were summed up. Resulting images were normalized from 0 to 256 against the minimum and maximum number of photons of images obtained at a certain delay of time window.

Fig. 4

Images of diffusely reflected light obtained for each of nine emitting fibers located on a ring surrounding imaged area.

Fig. 5

Images of diffusely reflected light for nine emitting fibers after background subtraction. Black spot correspondes to an increase in absorption in a place of absorbing inclusion.

2.4.

Phantom Experiments

The plexiglas aquarium with a front side made of 50-μm thick Mylar film (DuPont Polska, Warsaw, Poland) was used. It was filled with water, milk, and ink solution (3:1 water to milk with fat content of 3.2%) of the optical properties μ _a = 0.12 cm⁻¹ and μ’ _s = 15 cm⁻¹. Absorbing inclusions located at various positions in the phantom were applied in order to show depth-discrimination of the time-gated measurements. Small balls (6 mm diam) made of gelatin were used as absorbing inclusions. The balls were manufactured of mixture of 30 g of gelatin with 400 ml of milk, water, and ink solution by filling a homemade metallic form. Inclusions were fixed in the phantom with the use of a transparent fishing line (thickness, 0.2 mm). The inclusions were fixed at four different lateral positions (Fig. 6) and five different depths from d = 8 to d = 20 mm. The absorption coefficient of the inclusion was ∼50 times higher than the absorption coefficient of the surrounding medium. The differences in ink concentrations were similar to the ones reported by Selb .²⁵ Measurements for three different time windows delayed in respect to the maximum of the IRF were carried out.

Fig. 6

Positions of inclusion with respect to the imaging area (black circles). Fiber positions are marked with black rings.

2.5.

In Vivo Measurements

A series of measurements during sensomotoric cortex stimulation were carried out. The monitoring during left-hand finger-tapping stimulation was performed on two volunteers, 30 and 27 years old. They were instructed to touch their thumb with their index and middle fingers, self-controlled, with frequency of ∼3 Hz. The measurements were carried out in sitting position. The optode setup was positioned on the right side of the head in such a way that its center was located on the C4 position (according to the 10–20 EEG standard²⁹). The stimulation procedure consisted of five cycles of 30-s finger tapping, followed by 30 s of rest. The surface of the tissue was imaged at a single time window delayed by 2 ns with respect to the maximum of the IRF. Acquisition of nine images for all source positions, resulting in single integrated image of distribution of reflectance at defined time window took ∼4 s. The resulting image was smoothed with the use of a 2-D digital filter under Matlab environment.

3.1.

Phantom Measurements

The absorbing inclusion was located at 20 different positions in the homogeneous liquid phantom. For each position of the inclusion, the scanning procedure was performed in order to obtain images for three time windows delayed by 1.5, 2.5, and 3.5 ns with respect to the maximum of the IRF.

In Fig. 7, integrated images of absorbing inclusion positioned at different depths are presented. It can be observed that the local increase in the number of absorbed photons at a selected position of inclusion depends on the time window. Effective detection of the inclusion positioned deeper is possible only for later time windows. Inclusion positioned at a depth of 20 mm was not detected for the first (earliest) time window. It can also be noted that for early time window, the image of the inclusion is smaller than the same inclusion imaged for longer delays. This effect can be explained by the higher number of scattering events corresponding to the longer time of flight of the photons. Thus, the higher probability of lateral penetration of the photons can be noted. These results suggest that analysis of delayed time windows allow for selective evaluation of absorption changes located deeply in the medium.

Fig. 7

Integrated images of diffusely reflected light obtained by summing up the images recorded for each of nine source positions. Images were obtained for inclusion located at four different depths end for three different time windows delayed by 1.5, 2.5, and 3.5 ns, respectively, to the maximum of the IRF.

In Fig. 8, results of imaging of the absorbing inclusion, positioned at different depths and different lateral positions, are presented. These images were obtained for a single time window delayed by 3.5 ns in respect to the maximum of the IRF. It can be observed that inclusion can be localized at all lateral positions. However, the shapes and sizes of the images of absorbing inclusions are different. The shape of the image of the inclusion is most regular when the inclusion is located in the center of the imaged area.

Fig. 8

Images of diffusely reflected light recorded for single time window delayed by 3.5 ns with respect to the maximum of the IRF. The images were obtained for absorbing inclusion located at various depths and different lateral positions.

3.2.

In Vivo Measurements

In Fig. 9, spatial distribution of changes in the number of photons detected in the motor cortex area during the finger-tapping procedure is presented. As mentioned earlier, we performed five cycles of the 30-s stimulation period, followed by 30 s of rest. In the first step of analysis, we averaged all images obtained for stimulation periods and all images obtained for rest periods. As a result, two images were obtained, one corresponding to the stimulation period and one corresponding to the rest period. Furthermore, the rest image was subtracted from the stimulation image (see Fig. 9). Zero value represents the mean number of photons during the experiment. An increase in the number of photons at a wavelength of 780 nm during stimulation can be observed. This increase is connected with the decrease in tissue absorption caused by an increase in brain tissue oxygenation. The stimulated area can be observed in the lower-right quadrant of the image.

Fig. 9

Image representing a decrease in absorption in motor cortex caused by finger-tapping stimulation. The directions to the Nasion, Ear, Inion, and Vertex on human head are indicated. Middle of the image corresponds to the position of C4 point of 10–20 EEG system.