The present disclosure relates to an EEG signal processor, and associated methods. More particularly to processing of EEG signals to provide an indicator of a participant’s olfactory bulb function.

According to a first aspect of the present disclosure there is provided a system comprising a processor, wherein the processor is configured to:

receive first-EEG-signalling, which is representative of electrical activity in a participant’s brain for a period of time following an instant at which the participant has been exposed to an odour;

determine, as EBG-power-data, the power in the first-EEG-signalling at one or more frequencies that are greater than 25Hz; and

provide an indicator of the participant’s olfactory bulb function based on the EBG- power-data.

Unexpectedly, it has been found that EEG signalling with a frequency greater than about 25Hz provides an accurate representation of olfactory bulb function.

The processor may be configured to receive the first-EEG-signalling from one or more electrodes attached to the participant’s scalp or face (such as the participant’s forehead). In this way, a non-invasive system for providing an indicator of a human’s olfactory bulb function can be provided, which surprisingly is not unduly impacted by the proximity of the olfactory bulbs to the sinuses.

The processor may be configured to: compare the EBG-power-data with one or more threshold values; and provide the indicator of olfactory bulb function based on the comparison.

The processor may be configured to: determine, as the EBG-power-data, the power in the first-EEG-signalling at one or more frequencies between 25Hz and 100Hz.

The processor may be configured to determine, as the EBG-power-data, the power in the first-EEG-signalling:

i) at one or more frequencies that are greater than or equal to 40Hz, 50Hz or 60Hz, and / or ii) at one or more frequencies that are less than or equal to 100 Hz, 80Hz or 70Hz; and / or

iii) for a period of time that begins 75ms, 125ms or 200ms after the instant at which the participant is exposed to the odour; and / or

iv) for a period of time that ends 350ms, 300ms or 250ms after the instant at which the participant is exposed to the odour.

The processor may be configured to:

compare the EBG-power-data with a plurality of threshold values that are representative of predetermined power levels, and

provide a graphical display of the EBG-power-data based on the comparison.

The processor may be further configured to:

compare the EBG-power-data with a plurality of threshold values; and

attribute a score to the olfactory function.

The processor may be further configured to:

provide an indicator of a functioning olfactory bulb if the EBG-power-data or the score is greater than a functioning-threshold; and / or

provide an indicator of a non-functioning olfactory bulb if the EBG-power-data or the score is less than the functioning-threshold.

The first-EEG signalling may be representative of electrical activity in one or both hemispheres of the participant’s brain.

The first-EEG signalling may be representative of electrical activity in the left hemisphere of the participant’s brain. The processor may be further configured to perform one or more of the following steps:

determine, as left-bulb-EBG-power-data, the power in the first-EEG-signalling at one or more frequencies that are greater than 25Hz;

provide an indicator of the participant’s left olfactory bulb function based on the left- bulb-EBG-power-data;

receive second-EEG-signalling, which is representative of electrical activity in the right hemisphere of the participant’s brain for the period of time following the instant at which the participant has been exposed to an odour;

determine, as right-bulb-EBG-power-data, the power in the second-EEG-signalling at one or more frequencies that are greater than 25Hz; provide an indicator of the participant's right olfactory bulb function based on the right-bulb-EBG-power-data; and

provide a Parkinson’s disease indicator based on the indicator of the left olfactory bulb function and the indicator of the right olfactory bulb function.

The processor may be configured to receive the second-EEG-signalling from one or more electrodes attached to the participant’s scalp or forehead.

The processor may be further configured to: provide the Parkinson's disease indicator based on a difference between the indicator of the left olfactory bulb function and the indicator of the right olfactory bulb function.

The processor may also be configured to:

determine, as other-olfactory-system-power-data, the power in the first-EEG- signalling at one or more frequencies between 12.5Hz and 30Hz; and

provide an indicator of the function of other processing nodes in the participant’s olfactory system based on the other-olfactory-system-power-data.

The system may further comprise: one or more EEG electrodes configured to be attached to the participant’s scalp or forehead in order to provide the first-EEG-signalling. For instance, the one or more EEG electrodes may be configured to be attached to at least one side of the nasal extension above the medial portion of the eyebrow of the participant in order to provide the first-EEG-signalling.

There may be provided a method of providing an indicator of a participant’s olfactory bulb function, the method comprising:

receiving first-EEG-signalling, which is representative of electrical activity in the participant’s brain for a period of time following an instant at which the participant has been exposed to an odour;

determining, as EBG-power-data, the power in the first-EEG-signalling at one or more frequencies that are greater than 25Hz;

providing the indicator of the participant’s olfactory bulb function based on the EBG- power-data.

The method may further comprise:

attaching one or more EEG electrodes to the participant's scalp or forehead in order to provide the first-EEG-signalling. For instance, attaching the one or more EEG electrodes to at least one side of the nasal extension above the medial portion of the eyebrow of the participant, in order to provide the first-EEG-signalling.

There may be provided a computer program, which when run on a computer, causes the computer to configure any apparatus, including a processor or device, disclosed herein or perform any method disclosed herein. The computer program may be a software implementation, and the computer may be considered as any appropriate hardware, including a digital signal processor, a microcontroller, and an implementation in read only memory (ROM), erasable programmable read only memory (EPROM) or electronically erasable programmable read only memory (EEPROM), as non-limiting examples. The software may be an assembly program.

The computer program may be provided on a computer readable medium, which may be a physical computer readable medium such as a disc or a memory device, or may be embodied as a transient signal. Such a transient signal may be a network download, including an internet download.

One or more embodiments will now be described by way of example only with reference to the accompanying drawings in which:

Figure 1 shows a cross-sectional view of a human head;

Figure 2 shows an example embodiment of a system that processes one or more EEG signals;

Figures 3a to 3c show schematically the acquisition of EEG-signalling, and a visual indicator of the power in the EEG-signalling that is representative of olfactory bulb function;

Figures 4a and 4b show one example of processing performed by one or more processors in order to determine the power in pre-processed EBG-signalling;

Figure 5 shows an example embodiment of a system that processes one or more EEG signals and provides: an indicator of left olfactory bulb function, and also an indicator of right olfactory bulb function; and

Figure 6 shows schematically an example embodiment of a method of providing an indicator of a participant’s olfactory bulb function.

Figure 1 shows a cross-sectional view of a human head, including the brain. The brain includes two olfactory bulbs 102 (a left olfactory bulb in the left hemisphere of a person’s brain, and a right olfactory bulb in the right hemisphere of the person's brain), one of which is shown in Figure 1. Each olfactory bulb is a neural structure of the vertebrate forebrain which is one part of the human olfactory system that contributes to a person's sense of smell (olfaction). Other processing nodes of the olfactory system include, among others, olfactory receptor neurons (ORNs) and the piriform cortex.

There are no non-invasive methods to acquire signals from the olfactory bulb (OB) 102 of a human. It has been found that attempts to acquire signals from the human OB 102 using functional magnetic resonance imaging (fMRI) have failed due to the OB’s proximity to the sinuses, one of which is shown in Figure 1 with number reference 104. The sinuses include large cavities, which create large distortions of the magnetic signal and can preclude good signal strength in the OB area.

It has been found that the sinuses do not affect electroencephalography (EEG) recordings. However, a significant problem exists in localizing the EEG signal to the olfactory bulb, and: how to separate the EEG response of olfactory receptor neurons from that of processing in the olfactory bulb on the one hand; as well as separation of cortical and olfactory bulb processing on the other.

The OB 102 was long assumed to function only as a relay station that processed an odour signal on its route to cortical areas where value-added processing started. However, it has now become clear that the OB 102 is intimately involved in the processing of many olfactory tasks: odour discrimination and intensity, concentration-invariant recognition, odour segmentation, and to some extent, odour pattern recognition. No techniques exist that allow acquisition of functional signals from the OB 102.

The OB 102 is also linked to several neurodegenerative diseases, with the foremost among them being Parkinson's disease (PD). PD patients have early olfactory dysfunction due to the OB 102 being the location of the first PD-dependent cerebral insults. However, significant degeneration of the OB 102 is needed before a drop in odour behavioural performance can be detected. A reliable measure of OB signals would therefore not only serve as an invaluable research tool to define the basic mechanism of olfactory processing in healthy individuals, but also to potentially act as a very early (years before behavioural odour problems) biomarker for PD.

Figure 2 shows an example embodiment of a system 200 that processes one or more EEG signals. The EEG signals are representative of electrical activity in a participant’s brain, and provide an indicator of olfactory bulb function by using the processing described herein. As will be discussed below, the indicator can be a visual display, a binary indicator of whether or not the olfactory bulb is considered to be functioning or non-functioning, or it could be a non-binary score value.

The system 200 includes a first EEG sensor system 202 that provides first-EEG-signalling 204 to a processor 206. The first EEG sensor system 202 includes one or more EEG electrodes that can be attached to the scalp of a person in order to provide the first-EEG- signalling 204. In some examples, the first EEG sensor system 202 includes a plurality of electrodes that are placed in the vicinity of both of the olfactory bulbs such that the first- EEG-signalling 204 is representative of the functioning of both olfactory bulbs. In other examples, the first EEG sensor system 202 can be located on a person’s scalp such that it obtains signals from only one of the olfactory bulbs. Optionally, a second EEG sensor system 212 can then be used to provide second-EEG-signalling 214 representative of the other olfactory bulb. For instance, the first EEG sensor system 202 can obtain signals from the left olfactory bulb, and the second EEG sensor system 212 can obtain signals from the right olfactory bulb, or vice versa. Further details of where the electrodes can be placed on a person’s scalp will be provided below.

The processor 206 receives the first-EEG-signalling 204, which is representative of operation of the olfactory bulb for a period of time following an instant at which a patient has been exposed to an odour. As will be discussed below, in a clinical setting a person can be exposed to a controlled odour at a predetermined instant in time. The processor 206 can then determine the power in the first-EEG-signalling 204 at one or more frequencies that are greater than about 25Hz. The parts of the first-EEG-signalling at the one or more frequencies that are greater than 25Hz, for which the power is determined, will be referred to herein as electro-bulbograms signalling (EBG-signalling). That is, EBG- signalling can include one or more frequency components of EEG-signalling that are greater than 25Hz. The determined power of the EBG-signalling can be referred to as EBG-power-data.

The processor 206 can optionally include the functionality of a filter 208 to enable the EBG- signalling to be extracted from the first-EEG-signalling 204, such that processing can be performed on a restricted set of frequency ranges / bins. In some examples, the first-EEG- signalling 204 may already have been restricted to the frequencies of interest (the EBG- signalling) before it is provided to the processor 206 such that the first-EEG-signalling 204 is the same as the EBG-signalling. Either way, the processor 206 can provide the indicator of olfactory bulb function based on the EBG-power-data. In some examples, as will be discussed below, the processor 206 can compare the EBG-power-data with one or more threshold values. The processor 206 can then provide the indicator of olfactory bulb function based on the comparison.

Unexpectedly, it has been found that EEG signalling with a frequency greater than about 25Hz provides an accurate representation of olfactory bulb function. In some examples, EEG-signalling in the gamma band has been found to provide good results. The gamma band can be defined as frequencies between 25Hz and 100Hz. Importantly, if a skilled person attempted to make human odour EEG recordings, they may expect to use low- pass filters because human cortical processing of odours may be expected to operate at around 5Hz. This would exclude the detection of the EBG-signalling described herein, which have one or more frequencies that are greater than 25Hz.

Providing an indicator of olfactory bulb function in this way can advantageously enable further explorations of the role fulfilled by the olfactory bulb in the human olfactory system. Furthermore, the process can be easily implemented as an everyday screening tool for Parkinson’s disease given the role of the olfactory bulb as an initial point of cerebral insult in Parkinson’s disease progression.

Figures 3a to 3c show schematically the acquisition of EBG-signalling, and a visual indicator of the power in the EBG-signalling that is representative of olfactory bulb function.

Figure 3a shows an illustration of a person’s head with four EBG electrodes 316a-d attached to the scalp. The locations of the left and right olfactory bulbs are also marked with stars 318a-b.

To record EBG-signalling, in the following example a system is used that includes an 8 electrode active-electrode electroencephalogram system, an olfactometer, and a system for recording nasal breathing. Nasal breathing signals can be monitored to control when the olfactometer delivers non-trigeminal odors to the nasal cavity, for example at the onset of inhalation. The EBG-signalling is recorded from the EBG electrodes 316a-d placed on the forehead, centrally above the eyebrows as will be described below. In some examples, the electrodes can be placed on the participant’s head or face anywhere above their eye line, such as on the forehead and / or on the scalp (for instance above the hairline).

In this example, participants were seated in-front of a computer screen and 8 active- electrodes (BioSemi, Active two, Netherland) were attached: one to each side of the head over the mastoids as reference electrodes (not shown), one above and one below the lateral end of the left eye to assess eye blinks (not shown), a first pair of electrodes 316a, 316b on one side of the nasal extension above the medial portion of the eyebrow, and a second pair of electrodes 316c, 316d on the other side of the nasal extension above the medial portion of the eyebrow. These two pairs of electrodes 316a-d will be referred to as EBG electrodes. The precise locations of the EBG electrodes in this example were assigned based on simulated lead-field originating from dipoles placed in the left and right olfactory bulb. In other examples, the electrodes can be placed in different positions on the participant’s scalp. Such differently placed electrodes can provide EEG-signalling that is processed differently to that described below.

Participants rested their head in a headrest and a nose piece that delivers odours from the olfactometer was inserted about 0.5cm into each nostril. The participant’s task was, after each trial, to rate how intense each odour was on a visual analogue scale appearing about 1 .5 seconds after each odour presentation on the screen in-front of them, replacing a visual fixation cross, using a computer mouse held in their dominant hand.

Iso-intense odours were delivered birhinally for 1 s for each trial using the computer- controlled olfactometer. To avoid trigeminal responses originating from drying out the intranasal respiratory mucosa, a total birhinal airflow of 3.5 I/m was used (1 .75 I/m per nostril). Triggering of odours took place at the onset of nasal inhalation meaning that all the odours were delivered at the beginning of inhalation phase to achieve a consistent timing in relation to sniff cycle across all trials. Sniff-triggering also reduces the potentially unwanted bias of attention and anticipation that cue-triggering might induce on the signal of interest. Nasal respiration was measured using a thermopod (Adlnstruments, Bolder, CO), attached in close proximity to the right nostril. As the subject inhales, the temperature falls in comparison to the warmer exhalation air and the onset of inhalation can be inferred by setting a threshold to zero at the midpoint where the onset of inhalation can then be detected based on the falling edge below the threshold. Sniff cycles were recorded at 400Hz and a first derivative (i.e. change) of measured temperature with window length of 81 samples was calculated to level out extreme values and to increase sensitivity to signal change. Moreover, to assure odour onset at early inhalation phase and to enable adjustment of odour onset-timing, odour timing was measured in real-time from the odour tube using a photo-ionization detector (mini-PID, Aurora Scientific).

The participant was exposed to different odours while EBG-signalling was continuously recorded from the four EBG electrodes 316a-d. The EBG-signalling was sampled at 512Hz and online high-pass filtered (0.01 Hz). During testing, participants were only using their nose to breath and were presented with either odorized or odorless air as described above.

EBG data were epoched from 500 ms pre-stimulus to 1500 ms post stimulus. Epochs containing artifacts were identified in two separate steps in this example. Firstly, three different types of artifact were identified (i.e.‘jumps’, muscle, eye movement). To identify data containing jumps, the EBG data were z-transformed, median filtered with the order of 9, and then absolute differences were calculated to detect jumps. Subsequently, muscle artifacts were identified by band-pass filtering of data (Butterworth order 8, cutoff frequency — 110 - 140 Hz) followed by a standard z-transform and Hilbert transformation to extract amplitudes followed by smoothing the transformed signal (which can be done by convolving data with a boxcar function using a 200 ms width). Lastly, eye blinks and other eye-related artifacts (such as the participant moving their eyes to the left or right) were identified by using data obtained by the additional electrodes on the forehead within a Butterworth filter (order 4, cutoff frequency - 1 - 15 Hz) and, in a similar way to muscle artifact detection, implemented as a z-transform, Hilbert transformation and smoothing. All trials with identified artifacts were rejected from further processing.

In an optional second step, visual inspection of the EBG data can be performed by first determining the variance across trials and then removing trials with high variance manually.

The remaining EBG data were band-pass filtered (Butterworth order 4, cutoff frequency -1 Hz - 100 Hz) and line-filtered at 50 Hz for further analysis. The line filtering was performed with the help of discrete Fourier transform (DFT) filter with a 3Hz bandwidth. The DFT filtering was implemented with help of Fourier coefficients estimation using a fast Fourier algorithm and zeroed of the line coefficient.

Figure 3b shows the pre-processed EBG-signalling 319a-c obtained for 3 separate odour- exposure-events, and also shows pre-processed EBG-signalling 319d acquired for no odour exposure. In Figure 3b, each pre-processed EBG-signalling line 319 is an average of the signals returned from the four EBG electrodes 316 for a single odour-exposure- event. In some examples, each pre-processed EBG-signalling line 319 can correspond to only one of the olfactory bulbs. In further examples still, the pre-processed EBG-signalling lines 319 can correspond to an average of: i) the signals returned from EBG electrodes 316 for a plurality of the same type of odour-exposure-events (for example, all events where the odour is chocolate) for a single participant; ii) the signals returned from EBG electrodes 316 for a plurality of different types of odour-exposure-events for a single participant; and iii) the signals returned from EBG electrodes 316 for a plurality of the same or different types of odour-exposure-events for a plurality of participants

Figure 3c shows a plot of EBG-power-data 321 across time (on the horizontal axis) and frequency (on the vertical axis). The EBG-power-data 321 is derivable from the EBG- signalling 319 as will be discussed below. An odour is presented to the participant at time - 0s, identified with reference 320 in Figure 3c. In this example a subset of the EBG- power-data 321 is shown within a dotted box 322 in Figure 3c - this subset of EBG-power- data 321 will be referred to as selected-EBG-power-data 322.

In this example, the EBG-signalling can be defined as the EEG signalling in a plurality of frequency bins between 30Hz and 100Hz, during an EBG-period-of-time from about 0.3s before the instant 320 at which the participant is exposed to an odour, until about 0.7s after. Furthermore, the selected-EBG-power-data 322 can be defined as the EBG-power- data 321 in a plurality of frequency bins between 40Hz and 90Hz, during a selected-EBG- period-of-time from 0.05s to 0.15s following the instant 320 at which the participant is exposed to the odour.

The EBG-power-data 321 for the entire EBG-signalling is shown in Figure 3c, in addition to the selected-EBG-power-data 322. It will be appreciated that in some examples the EBG-signalling can be filtered so that only the frequency and time components that are required for the selected-EBG-power-data 322 remain before the power is calculated. In this way, only the selected-EBG-power-data 322 is calculated.

In generating the plot of Figure 3c, a processor can be considered as comparing the EBG- power-data 321 with one or more threshold values in order to determine how to visually display it. In this example, the individual data points of the EBG-power-data 321 are compared with a plurality of power threshold values to determine how they are displayed. In this way, the plot of Figure 3c can be considered as providing a visual indicator of olfactory bulb function based on the comparison.

The EBG-power-data 321 shows estimated power difference (between odour and air) across time and frequency, which represents how an olfactory bulb responds to an odour.

It will be appreciated that the EBG-power-data 321 and also the selected-EBG-power-data 322 do not need to be defined by the specific time and frequency ranges that are illustrated in Figure 3c. For example, different ranges can be used, or only a single frequency bin and / or only a single timeslot may be used. In some instances, the EBG-power-data 321 and / or the selected-EBG-power-data 322 can represent the power in the first-EEG- signalling at: i) one or more frequencies that are greater than or equal to 40Hz, 50Hz or 60Hz, and / or ii) one or more frequencies that are less than or equal to 100 Hz, 80Hz or 70Hz.

Also, the EBG-power-data 321 / selected-EBG-power-data 322 (and therefore also the corresponding portion of the first-EEG-signalling) can be for a period of time that starts a predeterm ined-duration after the instant at which the participant is exposed to the odour. In some instances, the EBG-power-data 321 and / or the selected-EBG-power-data 322 can represent the power in the first-EEG-signalling for a period of time that: i) begins 50ms, 75ms, 125ms or 200ms after the instant at which the participant is exposed to the odour, and / or ii) ends 350ms, 300ms, 250ms or 150ms after the instant at which the participant is exposed to the odour.

In this example, a processor can provide an alternative representation of the selected- EBG-power-data 322. As shown with reference 324, the processor can calculate a standard error of the mean (SEM) of the selected-EBG-power-data 322. This can be considered as a non-binary score of the olfactory bulb function. In some examples, this score can be compared with a functioning-threshold value and the processor can: provide an indicator of a functioning olfactory bulb if the score is greater than the functioning- threshold; and provide an indicator of a non-functioning olfactory bulb if the score is less than the functioning-threshold.

It will be appreciated that various other statistical measures of the EBG-power-data 321 and / or selected-EBG-power-data 322 can be used to represent that data, including to attribute a score to the EBG-power-data 321 / selected-EBG-power-data 322.

Figures 4a and 4b show one example of processing that can be performed by one or more processors in order to determine the power in pre-processed EBG-signalling 419. The processing of Figures 4a and 4b is relevant for the 8 electrode arrangement that is described above with respect to Figure 3. The preprocessing described herein is optional in some examples.

In order to provide the pre-processed EBG-signalling 419, the processor first removes a linear trend and mean centers the data. This can be simply done with ordinary least square estimation to regress out the linear trend and signal offset. Then, the signal is padded with zeros to expand the length of samples to the next closest power of two to improve the speed of estimation of frequency coefficients. As part of the preprocessing, filtering can also be performed. In this example, bandpass filtering is performed to retain frequency components between 20Hz and 100Hz and thereby to convert received EEG-signalling to EBG-signalling.

The processor applies a fast Fourier transform (FFT) to the pre-processed EBG-signalling 419 in order to provide frequency-domain-EBG-signalling 426, which in this example is complex signalling with a real and imaginary component. In a similar way to that described above, the processor can provide the frequency-domain-EBG-signalling 426 by applying a FFT to pre-processed EBG-signalling 419 that represents one or more odour-exposure- events, of one or more different types, for one or more participants.

Separately, Figure 4a shows a time domain representation of a first taper 428, which is a discrete prolate spheroidal sequence. A first set of wavelets 430 is shown, also in the time domain, that is derived from the first taper 428. A frequency domain representation 432 of the first set of wavelets 430 is also shown in Figure 4a, for example following an FFT of the time domain representation of the first taper 428.

The processor then multiplies, in the Fourier space, (i) the coefficients of the frequency- domain-EBG-signalling 426, and (ii) the frequency domain representation 432 of the first set of wavelets 432, in order to generate a first-frequency-domain-convolved-signal 433. The processor then performs an inverse FFT on the first-frequency-domain-convolved- signal 433 to provide a first-time-domain-convolved-signal 434. In this way, the processor convolves the EBG-signalling 419 with a first set of wavelets 430.

In this example, the processor performs similar processing to that described above in relation to a second taper 436, which is a different discrete prolate spheroidal sequence to the first taper 428. A second set of wavelets 438 is shown, that is derived from the second taper 436. The processor then multiplies, in the Fourier space, (i) the coefficients of the frequency-domain-EBG-signalling 426, and (i) a frequency domain representation 440 of the second set of wavelets 438, in order to generate a second-frequency-domain- convolved-signal 441 . The processor then performs an inverse FFT on the second- frequency-domain-convolved-signal 441 to provide a second-time-domain-convolved- signal 442. In this way, the processor convolves the EBG-signalling 419 with a second set of wavelets 438. Turning now to Figure 4b, the processor performs an absolute squared operation 444 on the first-time-domain-convolved-signal 434. This generates a first-power-signal 448 that is a representation of the power in the EBG-signalling 419. Similarly, the processor performs an absolute squared operation 446 on the second-time-domain-convolved-signal 442 to generate a second-power-signal 450 that is another representation of the power in the EBG-signalling 419.

The processor then performs an averaging operation 452 on the first-power-signal 448 and the second-power-signal 450, in order to generate a combined-power-signal 454 that represents a power estimation of one or both of the olfactory bulbs (depending upon which electrodes are used to provide the EBG-signalling 419). Figure 4b shows a graphical representation 456 of the combined-power-signal 454, which is an example of EBG-power- data, that is the same as the plot shown in Figure 3c. A selected part of the combined- power-signal 454 is shown with reference 458.

In some applications, the processor can perform additional averaging operations 452. For instance:

• if the frequency-domain-EBG-signalling 426 is representative of a single olfactory bulb, then the processor can perform averaging operations 452 with power signals determined for the other olfactory bulb;

• if the frequency-domain-EBG-signalling 426 is representative of a single odour- exposure-event, then the processor can perform averaging operations 452 with power signals determined for other odour-exposure-events;

• if the frequency-domain-EBG-signalling 426 is representative of a single type of odour-exposure-event, then the processor can perform averaging operations 452 with power signals determined for other types of odour-exposure-events;

• if the frequency-domain-EBG-signalling 426 is representative of one or more odour-exposure-events for a single participant, then the processor can perform averaging operations 452 with power signals determined for other participants.

It will be appreciated that a different number of tapers 428, 436 can be used. The number of tapers 428, 436 can be selected so as to achieve an acceptable trade-off between leakage and sensitivity. Selecting tapers depends on the leakage from the lobes and the sensitivity to the frequency studied. It is advantageous to maximize the power in the main lobe and minimize the power in the side lobes. As more tapers are added, sensitivity is reduced and the signal is increasingly smoothed but, in return, this will minimize leakage and side band interference.

The processing described above can assess how the power develops across time and frequency. In a more specific example, frequency coefficients at each time bin are approximated by employing multi-taper sliding window and wavelets in the frequency range of 20 - 100 Hz (with step 0.1 Hz across time window 100 - 300 ms) using steps of 5 ms power at each time/frequency point. This frequency range spans the gamma band spectrum, which has been found to provide a good indication of olfactory bulb function. This results in 801 frequency bins and 81 time bins. Power is then estimated at each bin (total of 64,881 bins) using complex wavelet 430, 438 transforms with two tapers 428, 436 from discrete prolate spheroidal sequences. The window length is adjusted to capture 3 cycles of signal at each frequency bin. Two sets of complex wavelets 430, 438, formed from two tapers 428, 436, are convolved with at least a selected part (corresponding to the selected part 458 of the combined-power-signal 454) of the EBG-signalling 419 to estimate the power at each time / frequency bin. In some applications, the selected part of the EBG- signalling 419 may the entire EBG-signalling 419.

The convolution is implemented in the frequency domain in this example as a multiplication of fast Fourier coefficients of the EBG-signalling 426 and the coefficients 432, 440 of the two sets of wavelets 430, 438. In some applications, this processing is more computationally efficient in the frequency domain. However, in some examples at least some of the above processing can be performed in the time domain.

An inverse Fourier transform is subsequently applied to return the convolved signals into the time domain, and the resultant amplitudes are averaged 452 for the different tapers. The power at each time/frequency point is estimated as the absolute squared 444, 446 of these numbers. This estimated power is shown as EBG-power-data 456 in Figure 4b.

Power within the designated area of interest (dashed box 458 in Figure 4b) is extracted.

In some applications, it can be beneficial to present the EBG-power-data 456, 458 at a group level (for a plurality of participants) to enable statistical comparisons to be readily performed.

Figure 5 shows example embodiment of a system 500 that processes one or more EEG signals and provides: an indicator of left olfactory bulb function 510a, and also an indicator of right olfactory bulb function 510b. Features of the system 500 of Figure 5 that are also shown in Figure 2 have been given corresponding reference numbers in the 500 series and will not necessarily be described again here.

In this example, the system 500 includes a first EEG sensor system 502 that provides first- EEG-signalling 504 that is representative of operation of the left olfactory bulb of a participant. The first EEG sensor system 202 includes one or more EEG electrodes that can be attached to the scalp of the participant, in the vicinity of the left olfactory bulb, in order to provide the first-EEG-signalling 504. The system 500 includes a second EEG sensor system 512 that provides second-EEG-signalling 514 that is representative of operation of the right olfactory bulb of the participant. The second EEG sensor system 512 includes one or more EEG electrodes that can be attached to the scalp of the participant, in the vicinity of the right olfactory bulb, in order to provide the second-EEG- signalling 514.

In this example, the processor 506 includes: (i) a first processing pipeline 506a for processing the first-EEG-signalling 504 and providing the indicator of left olfactory bulb function 510a; and (ii) a second processing pipeline 506b for processing the second-EEG- signalling 514 and providing the indicator of right olfactory bulb function 510b. Each of the processing pipelines 506a, 506b can provide the functionality of the processor that is described with reference to any one of Figures 2, 3 and 4 such that: (i) the indicator of left olfactory bulb function 510a can include left-bulb-EBG-power-data; and (ii) the indicator of right olfactory bulb function 510b can include right-bulb-EBG-power-data. It will be appreciated that in other examples, the processing functionality could be provided by a plurality of discrete processors, or by a single processing pipeline that is appropriately controlled.

In this example, the processor 506 provides a statistical representation 524a of the left- bulb-EBG-power-data. This statistical representation 524a can be provided as the indicator of left olfactory bulb function 510a, or can be derived from it. Similarly, the processor 506 can provide a statistical representation 524b of the right-bulb-EBG-power- data. These statistical representations 524a, 524b can be provided in the same way that is described above with reference to Figure 3c.

It has been found that early olfactory dysfunction in PD patients includes different responses to odour by the left and right nostril (analogously to the fact that motor symptoms also appear initially unilaterally) - that is PD affects one of the olfactory bulbs before the other. In a healthy person, there is no significant difference between the responses of the two olfactory bulbs. In this example, the processor 506 processes the indicator of left olfactory bulb function 510a and the indicator of right olfactory bulb function 510b in order to determine a Parkinson’s disease (PD) indicator 560.

For example, the processor 506 can provide the PD indicator 560 based on a difference between the indicator of left olfactory bulb function 510a and the indicator of right olfactory bulb function 510b. In some applications, the processor 506 can subtract a score attributed to the left olfactory bulb with a score attributed to the right olfactory bulb, and then compare the result of that subtraction with a PD-threshold. If the result of the subtraction is greater than the PD-threshold, then the processor 506 can set the PD indicator 560 as indicative of a biomarker for PD. In other applications, the processor 506 may determine a ratio of the score attributed to the left olfactory bulb with the score attributed to the right olfactory bulb. Then, if this ratio is greater than a PD-ratio-threshold, the processor 506 may set the PD indicator 560 as indicative of a biomarker for PD.

In some applications, statistical comparisons can be performed between left-bulb-EBG- power-data and right-bulb-EBG-power-data in order to determine whether or not there is a significant statistical difference between them. If there is a statistical difference, then the processor 506 can set the PD indicator 560 as indicative of a biomarker for PD. For example, permutation testing can be used. If a statistical p-value of less than 0.01 or 0.05 is determined, then this can be interpreted as indicative of a biomarker for PD.

In some examples, the above processing can also be extended to include EBG-signalling with frequency components in the beta band, which can be frequency components in range of between 12.5 and 30 Hz. Advantageously, these frequency components can be representative of the function of other processing nodes in the olfactory system, such as piriform cortex, amygdala, and orbitofrontal cortex (among others). For instance, a processor can determine other-olfactory-system-signalling as one or more components of the received EEG-signalling at one or more frequencies that are in the beta band. This other-olfactory-system-signalling can be at any of the timeframes that are discussed above with reference to the EBG-signalling. The processor can then determine, as other- olfactory-system-power-data, the power in the other-olfactory-system-signalling. The processor can also provide an indicator of the function of the other processing nodes in the olfactory system based on the other-olfactory-system-power-data, which can involve comparing the other-olfactory-system-power-data with one or more threshold values in a similar way to that discussed above, including an other-olfactory-system-functioning- threshold.

Figure 6 shows schematically an example embodiment of a method of providing an indicator of a participant’s olfactory bulb function.

At step 670, the method includes attaching one or more EEG electrodes to at least one side of the nasal extension above the medial portion of an eyebrow of the participant. This can be considered as an unconventional location for EEG electrodes.

At step 672, the method includes receiving first-EEG-signalling. As discussed above, the first-EEG-signalling is representative of electrical activity in the participant’s brain for a period of time following an instant at which the participant has been exposed to an odour. In some examples, the method can also include receiving second-EEG-signalling (not shown in Figure 6).

The method continues at step 674 by determining, as EBG-power-data, the power in the first-EEG-signalling at one or more frequencies that are greater than 25Hz. For example, by processing EEG signalling in the gamma band. Then at step 676 the method provides an indicator of the participant’s olfactory bulb function, based on the EBG-power-data, using any of the examples described herein.

The design disclosed herein can enable signals to be simultaneously and directly recorded from within each stage of olfactory processing, including the olfactory bulb. These measures can be used to directly assess correlations between each recorded stage and the EBG surface recordings.

The inventors have performed numerous studies to confirm that the EBG-signalling that is described herein does indeed represent the function of the olfactory bulb, and not another source. For example, the inventors have confirmed that the EBG response does not originate from: the olfactory receptor neurons (ORNs) or the piriform cortex. This included:

• Confirming that the indicators of olfactory bulb function were relatively stable over repeated exposures - the piriform cortex demonstrates a rapid habituation to repeated odour exposure, whereas olfactory bulb responses appear relatively stable over repeated exposures.

• Obtaining responses from a participant that is missing an olfactory bulb due to congenital anosmia, but who still demonstrates intact peripheral responses to odours - a negative indicator of olfactory bulb function was determined for this individual, as expected.

Confirming that rapid odour presentation only affects the simultaneously-recorded event-related potential (ERP) from cortical scalp electrodes, and that rapid odour presentation does not eliminate the EBG signal. If the EBG signal were eliminated, then this could indicate a major cortical source instead of the OB as primary source. Identifying that the recorded EBG response occurs too early after stimulus presentation (about 150ms) to derive from higher cortical odor processing.

Generating a parametric map of signal source reconstruction, which showed the olfactory bulb as the EBG signal source.

The functionality disclosed herein establishes a novel measure of human olfactory bulb responses to odour stimuli, and can allow assessment of neural responses from one of the early stages of human olfactory processing using an inexpensive, non-invasive, and temporally-precise recording method. Furthermore, equipment that already exists in numerous labs and clinics, including an EEG system, can be used. For the first time, it is possible to acquire data from the complete human olfactory system. Thus, a method of assessing olfactory bulb processing in an awake human using a noninvasive and relatively inexpensive technique is of great importance to the understanding of human olfactory processing in both healthy individuals and patients with Parkinson’s disease.

In addition, localizing disease-related changes in human central olfactory processing can require information about each stage of the olfactory pathway - information currently unobtainable. Thus, the technique disclosed herein, which allows measurement of human olfactory bulb signals, will greatly aid future olfactory-related translational work and establish a new paradigm for studies of human olfactory processing. This can enable fundamental questions to be explored, such as what role the human olfactory bulb plays in processing odour pleasantness, quality coding, and odour fear learning.

The teachings in this document can allow further investigation of a wide variety of clinical disorders known to affect olfactory processing, such as neurodegenerative and eating disorders as well as schizophrenia. Importantly, as indicated above, the olfactory bulb is closely linked to Parkinson's disease (PD) where clear behavioral olfactory disturbances (early occurrence in PD -91 %) commonly precede the characteristic motor symptoms defining the disease (early occurrence -75%) by several years. The reliable measure of olfactory bulb signals that can be achieved using the processing described herein therefore can serve as an invaluable research tool to define the basic mechanism of olfactory processing in healthy individuals, and also serve as potentially a very early (years before behavioral odor problems) biomarker for PD. In this way, examples disclosed herein can significantly broaden the accessibility to olfactory neuroscience for many more laboratories and also pave the way to potentially use the measure as an early biomarker of Parkinson’s disease in everyday clinical settings.