WO2007144711A2

WO2007144711A2 - Method and apparatus for evoking a response from a subject

Info

Publication number: WO2007144711A2
Application number: PCT/IB2007/001431
Authority: WO
Inventors: Richard Reilly; Edmund C. Lalor; Barak A. Pearlmutter
Original assignee: National University Of Ireland Maynooth; University College Dublin, National University Of Ireland, Dublin
Priority date: 2006-06-12
Filing date: 2007-05-31
Publication date: 2007-12-21
Also published as: WO2007144711A3

Abstract

A method of evoking a response from a subject is disclosed. The method comprises modulating a characteristic of a stimulus with a signal comprising a non-binary sequence, and measuring the subject's response to the stimulus. The subjects measured response is correlated with the modulated signal to produce an impulse response for the subject.

Description

Method and apparatus for evoking a response from a subject

The present invention relates to a method and apparatus for evoking a response from a subject.

A visual evoked potential (VEP) is a routinely used tool in both research and clinical settings for the evaluation of visual sensory and perceptual processing of a subject. In clinical testing, the so-called transient VEP is typically evoked by the repeated presentation of a visual stimulus at a rate of less than or equal to two presentations per second. The VEP is extracted from an electroencephalogram (EEG) by averaging the subject's measured response to each stimulus. When derived from such repetitive stimulation, the transient VEP shows several distinct components (e.g. Cl, Pl, Nl) for a given scalp topography on the occipital scalp.

A primary advantage of the VEP technique is its temporal resolution, which is limited only by the sampling rate of the measurement device.

As well as being used to evaluate optic neuritis and tumors, retinal disorders and demyelinating diseases such as multiple sclerosis, recent work has also shown that certain components of the transient VEP are affected in disorders such as schizophrenia, autism and depression.

If the rate of repetitive presentation of the visual stimulus exceeds 4-8 Hz, the separate components mentioned above of the transient VEP are no longer seen, and instead a periodic frequency-following response known as the steady-state visual evoked potential (SSVEP) is elicited. The periodicity of this response matches that of the stimulus, and provided stimulus presentation is precise, SSVEP power extends over an extremely narrow bandwidth. Spectral analysis with high frequency resolution allows rapid and continuous quantification of the SSVEP magnitude with high signal-to-noise (SNR) ratios. The advantage of the ability to rapidly obtain the SSVEP comes at the cost of the intrinsic timing information that comes with the distinct components of the transient VEP.

Both methods have also been used to study attention mechanisms in the brain. It has been reported that SSVEP amplitude modulations correlate strongly with certain peaks of the transient VEP but not with others, suggesting that SSVEP studies which measure only response amplitude changes as a function of attention will exhibit degraded performance as compared to techniques which also monitor response latencies or the complete response profile.

While studies of attentional modulation of transient VEPs abound, this technique is hampered by the need to average over many trials to derive a stable response profile with a typical minimum number in the region of 60 trials and as many as 200-300 being preferable. As mentioned above, each of these trials needs to be separated by at least 500 ms in order to obtain a sufficiently high signal-to-noise ratio. The length of time required to acquire this number of trials and the discrete nature of the trials present a serious challenge to the continuous measurement of short term focusing and shifting of attention. That is, many of the experimental paradigms necessitated by this arrangement become decidedly monotonous and can be extremely taxing for subjects.

Furthermore, in standard VEP attention studies, suddenly-onsetting stimuli 'appearing in unattended space might actually grab attention exogenously, causing involuntary, transient reorienting from cued to uncued or -distracter locations. Thus, it is not certain that the observed modulations of endogenously cued stimuli solely reflect endogenous attentional influences. Another issue regarding the standard VEP and attention is that, because the responses to each of multiple stimuli presented together cannot be separated, stimuli must be presented in isolation to the attended or unattended space at different times. This represents a serious limitation of the technique, in that it is not possible to emulate the more common real-life situation where both relevant and irrelevant information are present in the visual field at the same time.

Significant work has been done developing techniques to obtain visual evoked responses to more than one simultaneously presented stimulus. Much of this work has been based on the SSVEP, for example U.S. Patents 5,474,082 and 6,829,502. Here, control of a system is afforded based on the user's interaction with multiple stimuli each at a different frequency.

Another method that has been employed with some success is that of modulating each of a number of stimuli on a screen using a binary sequence such that every sequence is uncorrelated with every other, for example U.S. Patents 4,846,567 and 6,688,746. This allows for the determination of the response to each simultaneously presented stimulus. A similar idea using a null stimulus condition and sparsely presented stimuli is described in U.S. Patent application 2003/0163060.

Still, it would be of great use to have a method for rapidly and continuously measuring the visual evoked response where a complete temporal profile could be obtained.

According to the present invention there is provided a method of evoking a response from a subject comprising: modulating a characteristic of a stimulus with a signal comprising a non- binary sequence, measuring the subject's response to the stimulus, and correlating the subjects measured response with the modulated signal to produce an impulse response for said subject.

Preferably, the stimulus is visual.

•Preferably, the non-binary sequence comprises a sequence having a Gaussian (distribution. Alternatively, a coloured continuous distribution or a continuous random walk could be used.

Preferably, the power spectrum of the modulated signal is shaped to target specific frequency spectra to specific neural cellular subsystems, e.g. parvocellular, magnocellular, koniocellular .

Because the waveforms that are used to modulate, for example, the luminance of the stimulus are not limited to binary values, characteristics of the waveforms can be chosen more freely. For example, by changing the frequency spectra of these modulating waveforms it may be possible to target specific neural pathways in a controlled manner which is not possible using the prior art. This can be further achieved by restricting the dynamic range of the input waveform. For example, by restricting the range over which the contrast of a visual stimulus can be modulated to 0-10% the stimulus can be biased toward the magnocellular system.

By putting power in all regions of the time-frequency domain, limited only by considerations of minimizing annoyance to the subject or avoiding saturation of the visual system, the present invention is able to accelerate the process of acquisition of subject information as compared to conventional pulsed stimuli. This facilitates easy design of input stimuli with any desired spectra as well as the design of visually unobtrusive stimuli for which the impulse response of the system can be readily estimated.

In the preferred embodiment, the luminance or contrast of a visual stimulus is smoothly modulated using a spread spectrum waveform.

Preferably, signal processing is employed to recover the linear and non- linear impulse response functions of the visual system.

In alternative implementations, the stimulus is selected from tactile stimuli, auditory stimuli, visual stimuli or any combination thereof. Information rich multi-sensory stimuli with very specific statistical inter-relationships are easily implemented using the present invention and, thus, may be a valuable tool in the investigation of multi-sensory -integration effects. For example, it would be possible to examine how the magnitude of multi-sensory effects changes with respect to the correlation between waveforms controlling each modality.

The assumption that the modulation of the luminance or contrast of a visual stimulus is reflected in a linear way in the EEG is likely to be most true for the relatively simple cells in early visual areas. As such, the impulse response obtained using luminance or contrast modulation may reflect mostly activity in these early areas. However, other attributes of a visual stimulus may be modulated in order to target other specific areas of the visual system. For example, by modulating the positions of a series of dots such that the total image appears to vary smoothly between an object and a random array of dots it may be possible to obtain the impulse response of higher order areas, such as infero-temporal cortex.

Preferably, the stimulus is presented while concurrently recording EEG, MEG, fMRI, near-Infrared or any other functional brain imaging modality or any combination thereof.

In preferred embodiments, more than one stimulus is presented and the impulse responses to one or more of these stimuli are recovered. This can be of particular value in clinical diagnosis of conditions like hemineglect. This can also be very useful in specific research paradigms that depend on retinotopy such as early visual attention studies. The invention facilitates the rapid acquisition of a visual evoked potential with a complete temporal profile and high SNR. This speed of acquisition is of particular value in clinical settings.

The non-binary nature of the stimulus results in a more sensitive measure of the subject's response. This means that smaller changes in brain state (e.g. pathology) should be detected using the present invention than in the prior art.

Embodiments of the invention will now be described by way of example with reference to the accompanying drawings, in which:

Figure 1 (a) shows a flow diagram of a preferred embodiment of the present invention; Figure 1 (b) shows some exemplary visual stimuli used in the preferred embodiment;

■Figure 2 shows an example response signal acquired according to Figure 1;

Figure 3 shows scalp maps showing the topographical evolution (at 75, 100,

110, 130 and 155 ms) of a response using the present invention (top) vis-a- vis a standard VEP (bottom) ; and

Figure 4 shows another example response signal acquired according to Figure

1 where the dynamic range of modulation was restricted to 0-10% in order to bias it toward the magnocellular subsystem.

Referring now to Figure 1 (a) , firstly, a stochastic waveform is generated using any desired method, step 10. In the preferred embodiment, this is

Gaussian with uniform power in the range 2-30 Hz to produce a signal x(t) .

It should be noted that the frequency spectrum of x(t) can be tuned in any desired way in order to potentially target specific visual neural pathways.

In the preferred embodiment, this signal x(t) is then mapped to the luminance or contrast of a stimulus on a computer screen, step 12. In the preferred embodiment, it is mapped onto the luminance of the stimulus according to a linear relation with the zero-point of the waveform corresponding to a luminance of 50%, and scaled to allow three standard deviations within the displayable dynamic range. The stimulus is then presented to the subject while their EEG y(t) is recorded, step 14. Sample stimuli are shown in Figure 1 (b) . The first is a snowflake image containing a large uniform area while also having numerous sharp edges that increase the activation of Vl . Striate cortex holds columns of neurons that become active when lines or edges are presented, with each column responding to a specific orientation (Hubel and Wiesel, 1959) . The snowflake image subtended visual angles of 5.25° vertically and horizontally. The second image is a standard checkerboard pattern, which at maximum contrast comprises equal numbers of black and white checks. Each check subtends a visual angle of 0.65° both horizontally and vertically, while the checkerboard as a whole subtends visual angles of 5.25° vertically and horizontally. In the case of both the snowflake images and the checkerboard patterns, the luminances of the black and white areas were measured as 0.1 cd/m2 and 164 cd/m2 respectively, giving a Michelson contrast of 99.9%.

Using the white snowflake image of Fig. 1 (b) as a template, 256 snowflake images were generated where the white area of each image was assigned a gray-scale value between 0 and 255. Fig. 1 (b) shows three such images. The underlying spread spectrum waveform was mapped to the luminance level according to a linear relation, with the zero-point of the waveform corresponding to a luminance of 50%, and scaled to allow ± three standard deviations within the displayable dynamic range. On every refresh of the computer monitor, the snowflake image corresponding to the current sample of the input waveform was displayed and the EEG data was tagged with the corresponding value of the luminance.

For checkerboards, 68 checkerboards were generated where the mean of the luminance of the lighter checks and the darker checks approximately equal for each checkerboard. For example, the checkerboard comprising dark checks of gray-scale level 0 and light checks of gray-scale level 255 has a mean luminance of approximately 82 cd/m2. Similarly the checkerboard comprising dark checks of gray-scale level 129 and light checks of gray-scale level 230 has a mean luminance of approximately 82 cd/m2. Finally the uniform image consisting of pixels at gray-scale level 188 also has a mean luminance of 82 cd/m2. The underlying spread spectrum waveform was mapped to these images according to a linear relation, with the zero-point of the waveform corresponding to checkerboard 34, and scaled to allow + three standard deviations within the range of the images. Again, on every refresh of the computer monitor, the checkerboard image corresponding to the current sample of the input waveform was displayed. In this case, because the mean luminance of all the checkerboards was the same, the EEG data was tagged with the value of the luminance of the light checks minus the luminance of the dark checks. Fig. l(b) shows three of these constant mean luminance checkerboards.

Other suitable stimuli might comprise multiple spatially overlapped features (e.g. horizontal and vertical bars).

In an exemplary test, subjects were seated 60 cm from a 19 inch computer monitor driven by an NVIDIA GeForce FX5200 video card, at a refresh rate of 60 Hz. EEG data were recorded from 64 electrode positions, low-pass filtered below 134 Hz and digitized at a rate of 512 Hz using a BioSemi Active Two system (http: //www.biosemi . com/faq/cms&drl .htm) . Synchronization between the video display and the EEG signals was ensured by including the signal on the parallel port of the presentation computer, controlled by the presentation software, among the signals acquired by the analog-to-digital converter bank. The response properties of the video monitor used for stimulus presentation were measured using a Nuclear Associates photometer, model 07-621, with an ambient light shield. The monitor was found to have a gamma of 2.3.

Based on this recorded EEG y(t) and the corresponding stimulus waveform x(t), the impulse response of the visual system can be estimated by a method such as least-squares estimation to produce a signal w(t), step 16.

Fig. 2 shows an exemplary impulse response signal w(t) obtained using this method and this is termed a VESPA (Visually Evoked Spread Spectrum Analysis) response.

In this case, the VESPA shows as a large response to the stimulus with clear negative peaks at around 75 (Cl), 125 (Nl) and 175ms (N2) and with clear positive peaks at around 100 (Pl) and 150ms (P2).

As such, it can be seen that the VESPA allows for the same analysis methods as for a conventional VEP but can be measured continuously using less obtrusive stimuli. Another possibility for application of this method is in the isolation of different visual neural pathways by altering the characteristics of the input waveform or the stimulus characteristics or by choosing different stimulus attributes to be modulated. Scalp topographic mapping of the VESPA under the present stimulus parameters revealed highly specific scalp topographies, quite distinct from that obtained using the standard VEP. The abiding characteristic of the early VESPA maps in Fig. 3 was the persistently delimited focus over midline occipital scalp without any evidence for the characteristic early bilateral spread over lateral occipital scalp regions consistently seen for the standard VEP. This pattern suggests that the VESPA may well have a distinct cellular activation pattern from that of the VEP, favoring midline structures such as striate cortex and neighboring retinotopically mapped extrastriate regions as well as regions in the dorsal visual stream, activation of which are known to produce mxdline scalp topographies.

In another exemplary test the ability of the method to target specific neural subpopulations the dynamic range over which the contrast of the aforementioned checkerboards was varied was limited to 0-10%. This was chosen in order to bias the stimulus towards magnocellular cells. The VESPA obtained from this test is shown in Figure 4. It has a morphology that is very distinct from that of the VESPA in Figure 2 and, thus, provides strong evidence that it was generated by a distinct neural subpopulation.

In variations of the preferred embodiment, it is possible to obtain VESPAs for more than one stimulus simultaneously, making the method very useful for conducting various types of experiments targeting the visual and attention systems.

Two examples of such multiple stimuli can be seen at the bottom of Fig.

1 (b) . The first comprises two snowflakes situated 1° to the right and left of a central fixation point marked by a cross hair. The second comprises a small snowflake occluding a larger snowflake. For both of these arrangements, the modulating waveforms for each stimulus were different instantiations of the same random process, and therefore had identical statistics .

For the purposes of illustrating that the input waveform can be shaped as desired and still elicit the desired response, one of the waveforms could for example be filtered by scaling coefficients corresponding to frequencies below 1 Hz by a factor of 0.1 and those corresponding to frequencies between 1 Hz and 10 Hz by a factor of 0.3.

Many applications of the present application are possible, for example, the response can be used in:

• Diagnosis by comparison with models based on data from healthy controls and/or clinical populations.

• Investigation, by altering the characteristics of the input waveform, of the cellular underpinnings of sensory deficits in schizophrenia. • Classification of neuronal degeneration in ageing adults, including, but not limited to, alzheimer's disease and dementia.

• Diagnosis of autism.

• Biometric identification.

• Monitoring of alertness in people with alertness-critical jobs, e.g. truck drivers, pilots, sonar operators, air traffic controllers and soldiers .

• - Alertness testing as a proxy for response time in alertness critical jobs .

• Checking of the integrity of the visual system and of attentional modulatory systems.

• Assessment of retinopathies

• Monitoring of attention while a subject is reading.

• Producing retinograms in place of binary white noise or binary M- sequences . • Scanning for visual attentional deficits.

• Controlling a Brain-Computer Interface by modulation of attention to different stimuli.

• Research on binocular rivalry by presenting a different stimulus to each eye .

In improvements or variations of the preferred embodiments, many additional techniques may be employed. For example, Independent Component Analysis or another source separation method can be used to improve the signal-to-noise ratio of the response. Also, solving for the second-order elements of the impulse response parallel to the diagonal could be considered analogous to conventional paired-pulse gating stimulation. In some applications, measurement of interhemispheric transfer can be facilitated by measuring the ipsilateral response to laterally independent stimuli and by measuring the shared response to correlated lateral stimuli.

Claims

Claims :

1. A method of evoking a response from a subject comprising: modulating a characteristic of a stimulus with a signal comprising a non- binary sequence, measuring the subject's response to the stimulus, and correlating the subjects measured response with the modulated signal to produce an impulse response for said subject.

2. The method according to claim 1 wherein the stimulus is visual.

3. The method according to claim 1 wherein the non-binary sequence comprises a sequence having a Gaussian distribution.

4. The method according to claim 1 wherein the non-binary sequence comprises a coloured continuous distribution or a continuous random walk.

5. The method according to claim 1 comprising shaping a power spectrum of the modulated signal to target specific neural cellular subsystems.

6. The method according to claim 1 comprising restricting the range of the modulated signal to target specific neural cellular subsystems.

7. The method according to claim 5 wherein the subsystems comprise one or more of: parvocellular, magnocellular, or koniocellular .

8. The method according to claim 1 comprising modulating a luminance of a visual stimulus smoothly using a spread spectrum waveform.

9. The method according to claim 1 comprising modulating at least a second characteristic of a visual stimulus smoothly using a spread spectrum waveform in order to obtain responses from specific visual areas.

10. The method according to claim 2 comprising recovering the linear and non-linear impulse response functions of the visual system.

11. The method according to claim 1 wherein the stimulus is selected from tactile stimuli, auditory stimuli, visual stimuli or any combination thereof.

12. The method according to claim 1 comprising concurrently recording EEG, MEG, fMRI, near-Infrared or any other functional brain imaging modality or any combination thereof while presenting the stimulus.

13. The method according to claim 1 comprising presenting more than one stimulus; and recovering the impulse responses to one or more of said stimuli .