We have built an optical topography system at UCL, shown in figure 1, which consists of 32 laser diode sources (16 at 780 nm and 16 at 850 nm) and 16 avalanche photodiode detectors (APDs). All the sources are illuminated simultaneously, but are modulated at different frequencies. By Fourier transforming the signal received by each detector in software, the signal corresponding to each source can be isolated. This novel approach to optical topography allows great flexibility in the positioning of sources and detectors, such that a variety of arrangements of sources and detectors can be employed on the tissue surface with only minor changes being needed in the software. A detailed description of the system is available elsewhere [2]. The lasers diodes, each emitting a power of about 2 mW, are driven by frequencies within a single octave (from 2 kHz to 4 kHz) to prevent interference from harmonics. The 16 APDs each have a 3 mm diameter collecting area and a measurement bandwidth of approximately 100 kHz.

Figure 1: The UCL optical topography system. .

Figure 2: The geometry of the optical fibre array used for initial testing of the system.

The system is designed to be able to produce an image at a maximum rate of 20 frames/second using the information from 16 detectors. Sources and detectors are coupled to the tissue via multimode PMMA optical fibres, of 1 mm diameter, or, more recently, by glass fibre bundles. These fibres can be arranged in various configurations, an example of which is illustrated in figure 2. This specific arrangement was selected for a preliminary evaluation of the system, employing 16 sources (two at each position) and 4 detectors. The maximum separation over which a signal can be detected through tissue has been found to be approximately 40 mm. For the configuration in figure 2 the useful separations are 15.8 mm, 25 mm and 38.1 mm. For the two detectors nearest the centre of the array a signal can be obtained from the six nearest sources, and in total there are twenty source-detector pairs with separations less than 40 mm. The fibre-holding probe consists of a thermoplastic pad which can be easily moulded to the shape of the part of the body under investigation. The interior surface of the pad is lined with soft, light-absorbing foam.

2. Image reconstruction

We reconstruct three-dimensional (3D) images of the optical properties of tissues within the volume immediately below the array using an algorithm based on the so-called Rytov approximation. Measured changes in log(signal amplitude), y, are assumed to be related to changes in the optical absorption, x, by the matrix equation y = Ax, where A is the so-called Jacobian, or sensitivity matrix. We can calculate A by solving the diffusion equation using the finite element method applied to a mesh representing the geometry and (estimated) average optical properties of the tissue. Images, representing x, are generated by Tikhonov regularisation of the Moore-Penrose generalised inverse [4]. While the computationally intensive inversion of A can be performed offline in advance, images can be generated in real time by a straightforward matrix multiplication.

3. Imaging a dynamic phantom

A dynamic phantom has been constructed, consisting of a plastic box full of scattering fluid, with one wall of the box made from a 5 mm thick epoxy resin slab. The slab has an absorption coefficient (at 800 nm) of 0.01 mm-1, a transport scattering coefficient (at 800 mm) of 1.0 mm-1, and a refractive index of 1.56. The scattering fluid is an aqueous solution of Intralipid and near-infrared dye, with corresponding optical properties of 0.007 mm-1 and 0.8 mm-1. A stepper motor is mounted on the lid of the box, which enables a small target to be rotated within the liquid, suspended on a short length of rigid wire. The target consists of an epoxy resin cylinder, 10 mm in length and 10 mm in diameter, with the same scattering properties as the surrounding liquid, but ten times the absorption. The motor enables the target to be rotated in a plane perpendicular to the face of the phantom in a circle of diameter 60 mm, and at a speed of up to 1 revolutions per second.

Figure 3: Dynamic phantom and optical topography probe. .

Figure 4: Images of dynamic phantom at three different depths.

Figure 3 shows the topography array attached to the epoxy resin wall of the phantom. Data were collected at 10 Hz while the target revolved within the liquid, and 3D absorption images were generated from each 100 ms of data. The closest approach the target makes to the plane of the sources and detectors is 15 mm. Each column of figure 4 shows three slices through the 3D image of absorption taken parallel to the plane of the sources and detectors at depths of 5, 10 and 15 mm respectively. From left to right, figure 4 shows the target cylinder entering and crossing the field of view. The maximum change in absorption is 0.04 mm-1 and occurs at a depth of 5 mm. While this is less deep than the true position of the target, the depth of the target can be seen to change as it moves across the field of view, suggesting that there is some qualitative depth discrimination. This image was reconstructed from data with no averaging.

4. Adult motor cortex experiment

An experiment was performed on an adult volunteer in order to determine the ability of the topography system to display functional activation of the motor cortex. The thermoplastic pad was moulded to fit the shape of the volunteer's head as shown in figure 5, directly over the estimated position of the left motor cortex. The subject was asked to perform a finger opposition task with their right hand for 30 seconds, followed by 40 seconds of rest. This cycle was repeated eight times so that the data could be averaged to improve the signal-to-noise ratio. The intensity recorded at the two wavelengths was used to calculate changes in absorption coefficient using the differential pathlength factor (DPF) method [3]. A DPF value of 5 was assumed. The absorption changes were then used to estimate changes in concentrations of oxy- and deoxy-haemoglobin. A typical functional response for one source-detector pair is shown in figure 6. The horizontal green bar represents the period of time during which the subject was performing the motor task. An increase in the concentration of oxyhaemoglobin (red line) of approximately 2.5 µM is clearly evident. There is a corresponding but smaller decrease in the concentration of deoxyhaemoglobin (blue line) of approximately 0.05 µM. The change in total haemoglobin (black line) is also shown.

Figure 5: Optical topography of the adult motor cortex. .

Figure 6: Changes in oxy- and deoxy- and total haemoglobin concentration for one source-detector pair.

A 3D absorption image was reconstructed from averaged intensity data at 850 nm and is shown in figure 7. This shows three horizontal slices across the 3D image at depths of 5, 10, and 15 mm. The reconstruction algorithm assumes that the sources and detectors all lie within the same plane. The absorption increase is larger on the right half of the image, corresponding to the superior region of the motor cortex. The absorption increase at 850 nm corresponds to an increase in blood volume in response to the motor task.

Figure 7: A topographic image showing functional activation of the motor cortex. The image indicates an increase in absorption of 40% at 850 nm due to an increase in blood volume during activation. .

5. Infant visual cortex experiments

Our optical topography system as been used to record oxygenation changes in the visual cortex of 4-month-old infants [4]. The visual stimuli consisted of female faces, scrambled visual noise, and animated cartoons (to attract the babies' attention). The aim was to demonstrate the capability of the system to spatially localize functional activity and study the possibility of depth discrimination in the haemodynamic response. The group data show both face stimulation and visual noise stimulation induced significant increases in oxy-haemoglobin compared to the cartoons, but the increase in oxy-haemoglobin with face stimulation was not significantly different from that seen with visual noise stimulation. Activation maps were obtained using 3D reconstruction methods on multi source–detector separation optical topography data.

Figure 8: Reconstruction of the change in oxy-haemoglobin concentration in an infant projected in a single plane during a) face
stimulation and b) noise stimulation. The black arrow indicates the approximate position of the midline.

Figure 8 above shows tomographic reconstructed images of peak oxy-haemoglobin changes in a baby, averaged between 6 and 12 mm from the surface of the scalp. During face stimulation there is a noticeable increase in the lower area of the visual cortex spread to both sides of the midline. With noise stimulation the maximum increase is shifted towards the central region of the visual cortex, with a possible region of decrease adjacent to the maximum increase. See Blasi et al [4] for more details.

References

1. Gibson, AP, Hebden, JC, and Arridge, SR (2005): Recent advances in diffuse optical imaging, Physics in Medicine and Biology 50, R1-R43. Download PDF file.
2. Everdell, NL, Gibson, AP, Tullis, IDC, Vaithianathan, T, Hebden, JC, and Delpy, DT (2005): A frequency multiplexed near infrared topography system for imaging functional activation in the brain, Review of Scientific Instruments 76, 093705. Download PDF file.
3. Delpy, DT, Cope, M, van der Zee, P, Arridge, S, Wray, S, and Wyatt, JS (1988): Estimation of optical pathlength through tissue from direct time of flight measurement. Physics in Medicine and Biology 33(12), 1433-1442.
4. Blasi, A, Fox, S, Everdell, N, Volein, A, Tucker, L, Csibra, G, Gibson, AP, Hebden, JC, Johnson, MH, and Elwell, CE (2007): Investigation of depth dependent changes in cerebral haemodynamics during face perception in infants, Physics in Medicine and Biology 52, 6849-6864. Download PDF file.

Last update: December 10, 2007