WO2008016581A2

WO2008016581A2 - Imaging neuronal activity using nonlinear phase modulation as an intrinsic contrast mechanism

Info

Publication number: WO2008016581A2
Application number: PCT/US2007/017067
Authority: WO
Inventors: Martin C. Fischer; S. Warren Warren
Original assignee: Duke University
Current assignee: Duke University
Priority date: 2006-07-31
Filing date: 2007-07-31
Publication date: 2008-02-07
Anticipated expiration: 2009-01-31
Also published as: WO2008016581A3

Abstract

Measurements of two photon absorption (TPA) and/or intrinsic non-linear phase modulation such as self-phase modulation (SPM) or cross- phase modulation (XPM) can be used for neuronal imaging, with potential advantages in speed, penetration depth and molecular contrast. Nonlinear optical processes such as SPM and XPM are known to be sensitive to material structure and are significantly altered by neuronal firing. A pulse shaping technique that extracts weak non-linear phase modulation signatures can be extrapolated to high spatial and temporal resolution while retaining the noninvasive character, thus eliminating the requirement for expressed or exogenous contrast agents.

Description

TITLE

IMAGING NEURONAL ACTIVITY USING NONLINEAR PHASE MODULATION AS AN INTRINSIC CONTRAST MECHANISM

CROSS REFERENCE TO RELATED APPLICATION

[0001] This application claims the benefit of U.S. Provisional

Application No. 60/820,859 filed July 31 , 2006, the entire content of which is incorporated herein by reference.

FIELD

[0002] The technology herein relates to neuronal imaging, and more particularly to a new method and system for neuronal imaging using nonlinear phase modulation as an intrinsic contrast mechanism for neuronal activity with potential advantages in speed, penetration depth and molecular contrast.

BACKGROUND AND SUMMARY

[0003] Philosophers have for centuries pondered the relationship between mind and brain. However, investigators have only relatively recently been able to explore the connection analytically. The ability stems from developments in imaging technology including for example positron- emission tomography and magnetic resonance imaging. Coupled with powerful computers, such imaging techniques can now capture in real time images of the physiology associated with thought processes. They show how specific regions of the brain "light up" when activities such as reading are performed and how neurons and their elaborate cast of supporting cells organize and coordinate their tasks. The mapping of thought can also act as a tool for neurosurgery and elucidate the neural differences of people crippled by devastating mental illnesses, including depression and schizophrenia. See for example Raichley, "Visualizing the Mind," Scientific American (April 1994).

[0004] An ideal neuronal imaging modality would be able to noninvasively record intrinsic signatures of neuronal activity of the entire brain with high spatial and temporal resolution. Many techniques exist to study neuronal activation, none of which comes even close to fulfilling these requirements.

[0005] There thus remains a need to develop a measurement technique that can non-invasively map the rapid dynamics of entire neural circuits with high resolution. Mapping the activity of these circuits in living animals is a fundamental step toward understanding the neural mechanisms that underlie behaviors. Recent studies have suggested that neuronal networks are made of multiple layers of small functional units precisely located in a geometrical map. See K. Ohki et al., Nature 442, 925 (2006). Studying such functional units requires a technique that allows quantitative measure of the activity of neuronal circuits at the single cell level. However, existing noninvasive neuronal imaging techniques cannot meet the conflicting demands of high spatiotemporal resolution and broad field of view that are required to monitor the complex dynamics of neural circuits.

[0006] Most commonly, neural activation patterns are studied by recording the associated electric potentials or currents. However, electrodes that are inserted locally can only yield information at a few specific points. Magnetoencephalography and electroencephalography can measure neural function noninvasive^, but generally with rather limited spatial resolution. Another measurement method for neural functional imaging is to record the hemodynamic response with methods such as functional magnetic resonance imaging (fMRI), near infrared spectroscopy (NIRS) or diffuse optical imaging (DOI). However, not only is the hemodynamics generally relatively slow (at best significant fractions of a second), but these measurement methods generally offer rather low spatial resolution. For example, in realistic applications of DOI or NIRS, either the penetration depth for high-resolution measurements is exceedingly low (tens of microns) or the spatial resolution for deep-tissue measurements is poor (on other order of millimeters). Photoacoustic tomography has somewhat better spatial resolution (perhaps 100 microns) but relatively poor time resolution on the order of 1 second.

[0007] A much higher spatial and temporal resolution can be achieved when recording nonlinear optical responses of expressed or injected contrast agents. It is known that changes in potential can cause voltage-sensitive dyes (VSD) to rapidly alter their optical properties (fluorescence, absorption and birefringence). Common methods include two-photon fluorescence microscopy of calcium-sensitive dyes as in K. Svoboda et al., Nature 385, 161 (1997) and C. Stosiek et al., PNAS 100, 7319 (2003) and two-photon excited fluorescence or second-harmonic generation microscopy of voltage-sensitive dyes as in W.N. Ross et al., J. Membr. Biol. 33, 141 (1977), A. Grinvald et al., Nature Rev. Neuroscience 5, 874 (2004), and D. A. Dombeck et al., J. Neuroscience 24, 999 (2004). One of these nonlinear signatures of exogenous dyes can be used as the signal to monitor the membrane potential change with the benefits of higher spatial resolution, deeper penetration, better contrast and higher threshold of photobleaching. However, the required injection of probes into the brain often damages the brain tissue, restricts imaging to an area around the injection pipette, and sometimes affects the excitability of neurons.

[0008] Noninvasive (intrinsic) signatures of neuronal activity present additional challenges. Numerous studies have shown that neuronal firing is associated with transient changes in light scattering and birefringence on a sub-ms timescale, but these linear processes will give images that are highly degraded by light scattering.

[0009] The technology presented herein provides a new method for neuronal imaging, with potential advantages in speed, penetration depth and molecular contrast. The fundamental principles have been demonstrated by measuring signal changes during neuronal activation ex vivo. Translation of these techniques to live tissue and validation against existing techniques are believed possible. If the transition is successful, the technique holds the potential to revolutionize optical methods for neuronal imaging.

[0010] The technology^{^} presented herein thus takes advantage of nonlinear optical techniques to measure intrinsic signatures of neuronal activation. Self-phase modulation (SPM), a nonlinear optical property that causes self-induced, intensity-dependent phase changes in a light pulse propagating through a medium, has been identified as a contrast agent. In bulk liquids and structured solids, SPM is known to be highly sensitive to molecular composition and submicron structural details. To investigate SPM and other changes during neuronal activation, a recently developed pulse shaping technology has been employed to allow measurement of nonlinear signatures that had previously only been observable with prohibitively high laser powers. (See Tian et al., Opt. Lett. 27, 1634 (2002) and T. Ye et al., "Multiph. Micr. in Biomed. Sciences Vl", A. Periasamy et al. eds., vol. 6089, p. 60891 X (SPIE, San Jose, CA, USA, 2006)). With the enhanced sensitivity of present non-limiting exemplary embodiments, SPM signal changes in hippocampal brain slices have been measured during glutamate-induced neuronal activation, using very modest optical power.

[0011] Nonlinear phase modulations (e.g., self-phase modulation

(SPM) and cross-phase modulation (XPM)) are optical processes that are known to be sensitive to material structure. The present technology shows that SPM is significantly altered by neuronal firing and XPM is expected to show similar behavior. Other non-limiting exemplary embodiments of the present technology include measuring two photon absorption (TPA) in addition or alternatively to SPM and/or XPM. With a certain pulse shaping technique, weak nonlinear phase modulation signatures can be extracted under conditions suitable to in vivo studies. Measuring intrinsic nonlinear phase modulation and/or TPA with this technique provides an entirely new method for neuronal imaging, with potential advantages in speed, penetration depth and molέcular contrast. This method can be extrapolated to high spatial and temporal resolution while retaining the noninvasive character — thus eliminating the requirement for expressed or exogenous contrast agents.

[0012] In one illustrative non-limiting implementation, the nonlinearity implies that the spatial resolution can be submicron, at distances down to about 1 mm. Time resolution is expected to be extremely good. This combination represents highly significant advantages over all other noninvasive methods of imaging neuronal activity, optical or not.

[0013] A present exemplary non-limiting contrast mechanism will likely meet all but one of the stated properties of an ideal imaging method: it will be able to noninvasively record intrinsic signatures of neuronal activity with high spatial and temporal resolution, but it is safe to say that it will not be able to do so for an entire human brain. However, even if the penetration depth turns out to be only a few mm, the cortex of the brain (where most of the currently imaged neural activity occurs) would likely be accessible. Also, the penetration depth is potentially large enough for an entire mouse or even rat brain. "Minimally invasive" imaging in deeper regions of larger brains is certainly possible through endoscopic procedures.

BRIEF DESCRIPTION OF THE DRAWINGS

[0014] These and other features and advantages will be better and more completely understood by referring to the following detailed description of exemplary non-limiting illustrative implementations in conjunction with the following drawings.

[0015] Figure 1 is a diagram of an exemplary illustrative non-limiting implementation of a neuronal imaging system.

[0016] Figure 2 is a more detailed diagram of an exemplary illustrative non-limiting implementation of the neuronal imaging system illustrated in Figure 1. [0017] Figures 3A, 3B and 3C show exemplary illustrative non- limiting results of nonlinear measurements in solutions.

[0018] Figure 4 shows exemplary illustrative non-limiting results of non-linear measurements at several positions in a brain slice during neuronal activation.

[0019] Figures 5A, 5B and 5C show exemplary illustrative non- limiting results of nonlinear measurements in another experiment. In more detail, these figures show time evolution of a lock-in signal at different locations: near the cell body layer (center) in Fig. 5B, and 150 μm displaced on either side in Figs. 5A and 5C. The three curves in each location are spaced 10 μm apart. At the indicated time points (box labelled "G" in Figs 5A-5C), a buffer containing glutamate was applied to the sample.

[0020] Figures 6A and 6B show exemplary illustrative non-limiting results of measurements from another experiment. In particular, a steady- state signal (Fig. 6A) and relative change (Fig. 6B) of transmission and SPM signals during glutamate activation for a scan across a CA1 cell body layer is shown.

[0021] Figure 7 is an exemplary illustrative non-limiting histogram of the relative SPM signal change during glutamate activation resulting from another experiment.

[0022] Figure 8 shows exemplary illustrative non-limiting results of non-linear measurements from another experiment. In particular, Fig. 8 shows a time evolution of the SPM signal for glutamate activations with and without prior TTX application. [0023] Figures 9A and 9B show exemplary illustrative non-limiting results of measurements of another experiment. In particular, these figures show estimated scattering length (Fig. 9A) and SPM coefficient (Fig. 9B) in a neuronal sample during multiple activations.

DETAILED DESCRIPTION

[0024] One exemplary illustrative non-limiting implementation uses nonlinear phase modulation as an intrinsic contrast mechanism for neuronal activity. This effect is demonstrated by measuring self-phase modulation; however, cross-phase modulation (XPM, the multi-wavelength equivalent of SPM) is expected to show similar properties. For experimental demonstration of the present technology, the hole refilling method described in M. C. Fischer, T. Ye, G. Yurtsever, A. Miller, M. Ciocca, W. Wagner and W. S. Warren, "Two-photon absorption and self- phase modulation measurements with shaped femtosecond laser pulses," Opt. Lett. 30, 1551 (2005) was used and is incorporated herein by reference, but other measurement approaches are possible. This provides an intrinsic contrast mechanism that is a rapid indicator of neuronal activity and that can be extracted noninvasively with high resolution. Optical high- resolution imaging modalities of exogenous contrast agents are already in common use. Even though they share some of the advantages with the present non-limiting exemplary method and system (i.e. penetration depth, resolution) the required administration of agents classifies them as invasive. Noninvasive methods have so far not been able to provide high resolution information.

[0025] A working principle of this technique is based on the fact that nonlinear processes can create new frequency components in a laser pulse, but linear processes (such as scattering or absorption) generally cannot do this. Figure 1 shows an example illustrative non-limiting neuronal imaging system. The neuronal imaging system includes a laser system 100, a pulse shaper 200, a microscope 300, an optical spectrum analyzer 400, a data acquisition system 450 and a display 500. Figure 2 shows a more detailed exemplary illustrative non-limiting implementation. As shown in Figure 2, the pulse shaper 200 includes a reflective grating 201 , a collimating lens 203, an acousto-optic modulator (AOM) 205, a focusing lens 207 and a reflective grating 209. A pulse shaping control system 204 controls the AOM 205 so that a pulse (e.g., pulse 101) having a desired spectrum is output from the pulse shaper 200. The microscope 300 includes eyepieces 301, a dichroic mirror 303 and a beamsplitter 305 and is capable of holding sample 307. The optical spectrum analyzer 400 includes a reflective grating 401 , a collimating lens 403, a mask/filter 405 and a detector 407 (e.g., a photodiode).

[0026] In the exemplary implementations shown, a laser system 100 generates ultrafast laser pulses. A pulse shaper 200 receives the ultrafast laser pulses and shapes the spectrum of the ultrafast laser pulses in a way that makes the spectrum sensitive to specific nonlinear interactions. For example, the AOM 205 of the pulse shaper 200 shapes the spectrum of output pulse so that it contains one or more holes in the pulse spectrum (as can be seen in pulse 101 output from pulse shaper 200). The resulting shaped pulses from the pulse shaper 200 were then used to illuminate the sample 307 in a microscope 300. During the nonlinear interaction between the sample 307 and the pulses from the pulse shaper 200, the laser pulse spectrum undergoes changes that are characteristic for specific nonlinear interactions. A spectrum analyzer 400 spectrally analyzes the transmitted and/or backscattered light from the sample 307 to extract information such as two-photon absorption, self-phase modulation and/or cross-phase modulation pertaining to the non-linear interaction. The data acquisition system 450 acquires data from the spectrum analyzer 400. The display 500 displays a resulting neuronal image of the sample 307.

[0027] Self-phase modulation (SPM) is of particular interest for this application, since this non-resonant interaction depends on molecular concentrations and tissue morphology. However, the present technique is not limited to SPM. For example, cross-phase modulation (XPM) should work just as well (alone or along with SPM), as would two-photon absorption (TPA) alone or along with SPM and/or XPM. SPM causes self- induced, intensity-dependent phase changes of a light pulse propagating in a medium. XPM is the process in which a propagating light pulse of one color causes phase changes in a pulse of another color. Similarly, TPA can occur by absorption of two photons from a single light pulse or by absorption of photons from two distinct pulses.

[0028] Present, non-limiting illustrative embodiments can in principle measure any of these effects (SPM, XPM or TPA) or any combination thereof. The experiments described below only specifically investigate single-pulse effects such as SPM and "single-color" TPA. However, experiments using multiple pulses (i.e. pulses with a different center wavelength) could alternatively have been performed to, for example investigate XPM, and could potentially be more sensitive. TPA and SPM are routinely measured simultaneously, as will be appreciated for example from the experimental results illustrated in Figs. 3A-3C herein. TPA and SPM are in fact just two phase components of one acquired signal. In other neuron experiments described in for example Figs. 5-9 herein, no two-photon absorption signature was observed or expected. However, this can change when different wavelengths are used or when exogenous contrast agents are added.

[0029] Pulse shaper 200 may provide acousto-optical pulse shaping of the ultrafast laser pulses based on control from the pulse shaping control system 204. In one exemplary illustrative non-limiting implementation, the simultaneous measurement of two-photon absorption (TPA) and self-phase modulation (SPM) with modest light power levels is made possible by a measurement technique using shaped laser pulses through acousto-optic pulse shaping. See M. C. Fischer, T. Ye₁ G. Yurtsever, A. Miller, M. Ciocca, W. Wagner and W.S. Warren, "Two-photon absorption and self-phase modulation measurements with shaped femtosecond laser pulses," Opt. Lett. 30, 1551 (2005) and U.S. Patent No. 5,526,171 , each of which is incorporated herein by reference. This technique can dramatically increase the sensitivity of nonlinear imaging to such an extent as to be able to measure the generally very small nonlinear TPA and SPM signatures without causing tissue damage. These spectral changes are typically very small for the low pulse energies allowable in biomedical imaging applications. However, spectral changes can efficiently be detected by appropriately pre-shaping the spectrum such that the changes show against a- small background. To achieve this, a narrow hole (e.g., see pulse 101 in Fig. 2) was created in the pulse spectrum using acousto-optic pulse shaping techniques. The central frequency components were not completely eliminated but were left a small portion (a few percent of the peak intensity) to serve as a local oscillator (LO). Nonlinear processes can generate polarizations that contain frequency components at the location of the spectral hole. These components then interfere constructively or destructively with the LO in the spectral hole, depending on the relative phase. TPA is a dissipative process and the created nonlinear polarization exhibits a 180° phase shift with respect to the incident field. In contrast, the nonlinear polarization created by SPM is 90° out of phase with the incident field. With the present pulse shaper 200, the phase of the local oscillator can be adjusted independently from the remainder of the spectrum. A change of the LO phase causes synchronous changes of the spectral intensity at the LO position (the hole) that can be recorded with an optical bandpass filter and a detector such as a photodiode. By using lock-in detection SPM and TPA can be simultaneously extracted as two quadrature components of the photodiode signal.

[0030] Figures 3A, 3B and 3C show an example illustrative non- limiting demonstration of this measurement technique. In Figure 3A, TPA is shown by one line (having data points represented by solid colored circles) and SPM by another line (having data points represented by solid colored squares) for various samples in a cuvette, measured at λ=805 nm. The laser power is only 400 μW in this particular example. These figures compare self-phase modulation and two-photon absorption in Rhodamine 6G (R6G) (Figure 3A) with melanin (OD=0.5 at 600 nm) (Figure 3B) and oxyhemoglobin (2:3 mM) (Figure 3C), all in a cuvette. As can be seen, the samples show SPM and TPA whereas the glass walls only show SPM. The differences in the SPM signals on the far wall come from absorption in the physiological samples. A difference in the SPM signal between the physiological samples is also apparent.

[0031] Figure 4 shows an illustrative non-limiting example of SPM measurements in a rat brain slice during neuronal activation. At the indicated time points, solutions were added to the sample: Buffer (B) as a control, Glutamate (G) for activation, or TTX (T) to block activation. Figure 4 shows the time evolution of the SPM signal at different positions within the brain slice.

[0032] Hippocampal slices with a thickness of 400 μm were prepared from a P6 rat. The slices were mounted on a membrane (0.4-μm culture insert; Millipore) and cultured at 35 ⁰C under 5% CO₂ in minimal essential medium (3 mM glutamine, 30 mM Hepes, 5 mM NaHCO₃, 30 mM D- glucose, 0.5 mM L-ascorbate, 2 mM CaCI₂, 2.5 mM MgSO₄, 1 μg insulin, and 20% horse serum). To change neuronal activity, 2.5 μl_ of glutamate (100 μM) or tetrodotoxin (TTX, 1 μM) in a buffer containing 20 mM HEPES (pH 7.3), 127 mM NaCI, 2.5 mM KCI, 1.25 mM NaH₂PO₄, 5 mM NaHCO₃, 2 mM CaCI₂, 2 mM MgCI₂ was applied on a slice. The solution was absorbed in the slice and the membrane in 1-2 min. Figure 4 depicts the time evolution of the SPM signal at different positions within the brain slice: the expected position for the cell body layer (labeled as 0 mm), a position 100 μm displaced (-0.1 mm), and two positions 300 μm displaced on either side (±0.3 mm). The typical width of the cell body layer is on the order of 100 μm. 200 fs laser pulses were focused to about 20 μm and repeated at 20 kHz (for an average- power of 500 μW, one-tenth of a laser pointer). At specific time points during the recording, glutamate was added to stimulate neuronal firing, TTX to inhibit firing, and buffer as a control. The time points for administration are marked in Fig. 4 as "G", "T" and "B", respectively.

[0033] Pronounced changes in SPM were seen when glutamate was added, but not when a buffer was added. These changes did not occur when the neurons were missed. After adding the TTX to block activation no changes were seen when glutamate was added. The rise time is limited by the slow absorption, but this intrinsic optical signal should be just as fast as birefringence.

[0034] Figures 5-9 show the results of additional experiments. For the experiments described to obtain at least the results of Figs. 5-9, the laser system 100 was a regenerative amplifier (RegA, Coherent) seeded by a mode-locked ThSapphire oscillator (Vitesse, Coherent). The laser pulses entering the pulse shaper 200 at a repetition rate of 20 kHz were ~200 fs long with a center wavelength of 805 nm. The pulse shaper 200 created a local oscillator (spectral hole) of ~3 nm at the center of the pulse spectrum (see pulse 101 in Fig. 2) and the LO phase was rotated at 1 kHz. The beam radius (1/e² intensity) at the focus was chosen relatively large (-10 μm) to increase the likelihood of sampling neurons within the focal volume. The incident power to the sample 307 was 500 μW, corresponding to energies of 25 nJ/pulse. Part of the transmitted signal was directed onto a detector (e.g., a photodiode) as a monitor of sample transmission. The remaining transmitted beam passed through a 1 nm wide bandpass filter (angle-tuned to pass the center region of the spectral hole) onto another photodiode 407. This photodiode signal was then mixed with a 1 kHz reference in a lock-in amplifier and both quadrature components were recorded^with an integration time of 30 ms. The reference phase for the lock-in amplifier was obtained by recording the signal from water, in which only SPM but not TPA is present. Because no two-photon absorbers are present in the present samples, TPA was not observed, but only SPM contributions to the nonlinear signals. However, as noted above, this could have been changed if different wavelengths were used or if exogenous contrast agents were added. To confirm the nonlinear origin of the acquired signal, the input beam was attenuated (thereby attenuating the main pulse as well as the local oscillator) and the expected intensity-squared dependence was observed. The linear signal background caused by residual amplitude modulations in the spectral hole was negligible for the parameters used here.

[0035] In the experiments depicted in Figs. 5-9, SPM signal changes during neuronal activation in brain slices were investigated. Hippocampal slices were prepared from rats (postnatal day 6) with a thickness of 400 μm. The slices were mounted on membranes (0.4-μm culture insert; Millipore) and cultured for two weeks at 35°C under 5% CO₂ in minimal essential medium containing 20% horse serum. For each of the experiments, a brain slice was mounted on its membrane within a horizontal flow cell and completely immersed the slice in buffer containing 25 mM NaH₂PO₄, 127mM NaCI, 25 mM glucose, 2.5 mM KCI, 2 mM MgCI₂ and 2 mM CaCI₂ bubbled with O₂ 95% / CO₂ 5%. The cell was positioned on a movable sample holder in the microscope 300 for measurement with the hole refilling technique. A line was repeatedly scanned on crossing a layer of pyramidal neurons (CA1) in the brain slice to obtain the time course of signal. At specified time points, the inlet of the flow cell was switched to buffer containing glutamate (G, 200 μM) to stimulate neuronal firing, buffer containing tetrbdotoxin (TTX, 0.5 μM) to inhibit activation, or buffer containing both. The flow rate (~ 1 mUmin) through the cell remained unchanged during the switch of solutions.

[0036] Figures 5A-5C show a typical time course of the acquired transmission and lock-in signals in a brain slice during glutamate activation. Data is shown at several locations: the approximate position of the cell body layer (labeled as "center") in Fig. 5B and two positions 150 μm displaced on either side (the typical width of the cell body layer is on the order of 100 μm) in Figs. 5A and 5C, The time point for administration of glutamate solution during the recording is marked in Figs. 5A-5C as bars "G". At the position of the cell body layer (center) in Fig. 5B, the SPM signal showed pronounced transient changes followed by a long-term overshoot in response to a glutamate application (indicated as a bar labeled "G", 1 min), while only small changes occurred at locations well displaced from the cell body layer. Small changes were seen in the transmitted power for all locations through the brain slice, presumably due to activity dependent changes in light scattering. (See D. M. Rector et al., "In vivo optical imaging of brain function"(CRC Press, 2002)). The obvious time delay and slow decay of the observed signals upon activation is caused by the slow mixing and wash-out of the glutamate within the buffer- filled reservoir in the flow cell.

[0037] Figures 6A-6B show transmission and SPM signals before and during activation as a function of position for a line scan bisecting the layer of pyramidal CA1 neuron cell bodies. Fig. 6A depicts the steady- state values, i.e. the values before chemical activation. The SPM signal shows a drastic change across the cell layer, whereas the transmission does not. An even more interesting feature can be seen in Fig. 6B, which shows the relative signal change (the ratio of peak height to the pre- activation baseline) of SPM and transmission during an activation period. Drastic localization of the activation signal was observed, although with a somewhat different overall position dependence. In both cases, localization was seen only in the SPM signal. The transmission changes, caused largely by scattering changes, are relatively uniform across the sample. [0038] For the activation experiments, substantial inter-slice variability was seen both in the amplitude and the decay time. In a total of 14 brain slices, 50 line scans were performed, most of which consisted of multiple activations. The positioning of the scan line within the slice was based on weak visible contrast through an epi-illuminated microscope and was therefore approximate. In addition, some line scans might not have been wide enough to cover areas of response.

[0039] Figure 7 shows a histogram of all SPM signal responses. In the average of all the activation, an average of 26 % relative signal change with a standard deviation of 17 % was observed.

[0040] In order to establish neuronal activity as the cause of the observed SPM signal, change control experiments were performed. First, the membrane potential induced by glutamate using a patch clamp technique was measured. Cells depolarized from - 50 mV to 10 mV after glutamate application. The depolarization decayed within 5-10 min and was followed by a hyperpolarization. The time course of the polarization change is very similar to the observed SPM changes. The depolarization was largely inhibited by the sodium channel blocker tetrodotoxin TTX, suggesting that the cell depolarization is partially caused by the network activity as well as direct depolarization by glutamate receptors. Consistently, changes in SPM signal were also blocked by TTX (Fig. 8; activation suppression by 85 %, N = 6), suggesting that SPM signal is correlated with the neuronal activity.

[0041] The large relative signal changes and their localization are promising features of the present SPM measurement technique. However, the absolute quantification of these changes is made difficult by the fact that the measurement is influenced by other tissue parameters, for example by scattering. (See D. M. Rector et al., "In vivo optical imaging of brain function"(CRC Press, 2002). The effect of linear light attenuation can be estimated as follows. Scattering (or more generally attenuation) affects the measurements differently depending whether it occurs before or after the focus. Scattering before the focus attenuates the light and reduces the intensity in the focal region. The strength of the generated SPM signal depends quadratically on the focal intensity. Scattering after the focus merely attenuates the already generated nonlinear signal with a linear scaling. Because attenuation influences the measured SPM signal, that is the amplitude in the spectral hole, a change in this signal does not necessarily correspond to an actual change in the SPM coefficient in the sample. This quantification issue is not specific to the SPM measurement, but it is present in most nonlinear measurement methods, including two- photon fluorescence microscopy. In order to get an estimate of the SPM coefficient change, attenuation has to be taken into account, which in the present case is dominated by scattering, before and after the focus. Attenuation before the focus can be avoided by measuring at (or extrapolating to) the surface of the sample. Attenuation after the focus can be taken into account by normalizing the nonlinear signal by the transmitted signal, which is a measure of the linear attenuation factor. By performing measurements at different depths, the attenuation (scattering) coefficient can also be extracted.

[0042] Figures 9A-9B show the result of such a study for multiple activations in a brain slice. For this study, the probe beam was positioned in a region of strong SPM signal during activation, as described previously. SPM and transmission signals were acquired at several depths during activation, calculated the SPM over transmission ratio, and extracted the SPM coefficients at the surface and attenuation constants by exponential fits. The absolute calibration of the SPM coefficients was performed by using the SPM coefficient of water (1.7-10^'16 cm²/W; See W. Liu et al., Appl. Phys. B 76, 215 (2003)). In Figs. 9A and 9B, activity related changes can be seen in both coefficients — linear (scattering) and nonlinear (SPM). With this estimation, the relative changes in SPM coefficients in this sample were about 10 % to 15 %, which is smaller than the change in SPM signal obtained from the directly measured lock-in signal alone (about 45 % at the surface for this case). Even though the present estimation procedure is only approximate it shows that appreciable nonlinear SPM changes are present in the sample during activation.

[0043] In the above experiments intrinsic nonlinear signatures were recorded during neuronal activation. As a general qualitative, a positive peak in SPM signal was observed immediately following the application of glutamate and a successive drop of signal, often below the pre-activation baseline. In most cases, the signal slowly recovers within several tens of minutes to a level close to the pre-activation value. One question concerns the origin of the observed signal change. SPM depends on molecular concentrations as well as tissue morphology. The signal is therefore expected to reflect changes^caused by neuronal firing, as well as long term structural changes. Fast intrinsic changes in scattering and birefringence have been observed previously (e.g. recent observations of linear optical signatures in retina; See X.-C. Yao et al., J. Biomed. Opt. 11, 064030 (2006)), suggesting the possibility of nonlinear firing-related SPM signatures. However, a system might not capable of recording SPM changes on a timescale required for observing neuronal firing events (on the order of milliseconds). The slow mixing- and washout times of activating and inhibiting drugs further complicates interpretation. Drastically enhancing the time resolution and circumventing the mixing issue may be provided by electric neuronal stimulation.

[0044] The samples can be repeatedly stimulated without substantial decrease of signal. Firstly, this fact indicates that the changes induced by the application of glutamate are reversible. Secondly, it indicates that the sample can withstand even prolonged exposure to the laser light (the incident power was merely 500 μW, corresponding to energies of 25 nJ/pulse). It may be determined in the future how much of the residual decrease in signal baseline and peak strength is related to multiple applications of glutamate and how much is due to possible laser damage to the sample.

[0045] Present non-limiting exemplary embodiments are thus capable of detecting the presence of functional self-phase modulation, cross-phase modulation and/or two photon absorption contrast in, for example, hippocampal brain slices. A present hole refilling technique is able to measure any one or combination of these small nonlinear signatures at power levels low enough to be used for in vivo imaging. Intrinsic, high- resolution contrast at depths beyond the reach of existing techniques can thus be obtained.

[0046] While the technology herein has been described in connection with exemplary illustrative non-limiting implementations, the invention is not to be limited by the disclosure. The invention is intended to be defined by the claims and to cover all corresponding and equivalent arrangements whether or not specifically disclosed herein.

Claims

WE CLAIM

1. A neuronal imaging system comprising: a laser illumination system that shapes the spectrum of ultrafast laser pulses and illuminates a sample with said pulses, thereby causing nonlinear interaction between the sample and the pulses; a spectrum analyzer that spectrally analyzes the transmitted or backscattered light from the sample to extract information pertaining to said non-linear interaction; and a display that displays a neuronal image generated at least in part in response to said extracted information.

2. The system of claim 1 wherein said spectrum analyzer analyzes two-photon absorption and nonlinear phase modulation signatures of said transmitted or backscattered light.

3. The system of claim 1 wherein said spectrum analyzer analyzes at least one nonlinear phase modulation signature of said transmitted or backscattered light.

4. The system of claim 3 wherein the nonlinear phase modulation signature is a self-phase rrradulation (SPM) signature.

5. The system of claim 3 wherein the nonlinear phase modulation signature is a cross-phase modulation (XPM) signature.

6. The system of claim 1 wherein said spectrum analyzer analyzes a two-photon absorption signature of said transmitted or backscattered light.

7. The system of claim 1 wherein the laser illumination system that shapes the spectrum of the ultrafast laser pulses creates a hole in the spectrum of the ultrafast laser pulses to serve as a local oscillator.

8. A neuronal imaging method comprising: shaping the spectrum of ultrafast laser pulses; illuminating a sample with said pulses, thereby causing nonlinear interaction between the sample and the pulses; spectrally analyzing the transmitted or backscattered light to extract at least one of two-photon absorption or nonlinear phase modulation signatures; and generating a neuronal image at least in part in response to said extracted signatures.

9. The method of claim 9 wherein spectrally analyzing the transmitted or backscattered light extracts a self-phase modulation (SPM). signature as the nonlinear phase modulation signature.

10. The method of claim 9 wherein spectrally analyzing the transmitted or backscattered light extracts a cross-phase modulation (XPM) signature as the nonlinear phase modulation signature.

11. The method of claim 9 wherein shaping the spectrum of the ultrafast laser pulses includes creating a hole in the spectrum of the ultrafast laser pulses to serve as a local oscillator.