AUTISM SPECTRUM DISORDER

Field of the Invention The present invention relates to electroencephalograph (EEG) parameters for detecting an Autism Spectrum Disorder (ASD) and, in particular, a method for determining the presence of an ASD in a subject, determining the risk of developing an ASD in a subject, determining progression of an ASD or assessing response to therapy of a subject with an ASD using said EEG parameters. The present invention also relates to a system for determining the presence of an ASD in a subject and a kit of parts for carrying out the method of the present invention.

Background to the Invention Autism spectrum disorder (ASD) is the name for a group of developmental disorders often diagnosed in childhood, which affect social interaction, communication and behaviours and interests. Some of the most common ASDs are autism, Asperger syndrome and pervasive developmental disorder not otherwise specified (PDD-NOS). Approximately 1 % of the UK population are affected by an ASD with the incidence being much greater in males (1.8%) than in females (0.2%).

ASDs are thought to follow two distinct developmental routes. In the first most common type, concerns with a child's development are identified during the first two years of life. Problems identified often include delayed development of speech, not responding to others, displaying little interest in interacting with others and having a limited play repertoire. In the second type, a child's initial normal development is followed by a period of regression during the first few years of life.

People can be affected by ASDs in a variety of ways and can demonstrate a wide range of strengths and difficulties. However, there are a number of ASD impairments which are exhibited to some extent by most people suffering from ASD. These impairments include:

1. Problems with non-verbal and verbal communication including difficulty understanding language and delayed development of speech;

2. Problems with social behaviour for example having difficulty playing and empathising with others; and 3. Problems with behaving flexibly and being imaginative, including struggling with change and showing repetitive behaviours.

There is some evidence that a baby who later manifests ASD may develop more short range connections and fewer long range connections in the brain. Consequently, the ASD brain functions differently, perhaps because local rather than global information processing is favoured. These structural differences may affect the sensory information processing systems preventing the baby from having the integrated communication with his or her environment that is essential for proper social development.

Magnetic resonance imaging (MRI) studies have found increased cerebral white matter in ASD brains, differences in the hippocampus and amygdala, and increased total brain volume. These structural variations, linked to difficulty with imitation and social understanding suggest that brain abnormality in ASD is driven by different connections of the underlying neural networks and their mutual interactions. However, MRI techniques do not provide the fast time-scales needed to identify these interactions. In contrast, electroencephalography (EEG) can be used to reveal temporal dynamics of the brain on a relatively short time-scale.

Diagnosis of ASDs currently depends on clinical recognition of significant problems with an individual's communication, social interaction and flexible thought processes. Often this analysis and testing is undertaken by a number of clinicians including paediatricians, psychiatrists, speech therapists, occupational therapists and psychologists and it can therefore be some time before a diagnosis is made. In addition, diagnostic features may appear over time and a prolonged assessment of the individual may be required. There is, at present, no definitive test to confirm or rule out a diagnosis.

Sheikhani, A., et. al. (2012) J. Med. Syst. 36. 957-963 discloses the detection of abnormalities for diagnosing of children with autism disorders using of quantitative electroencephalography analysis.

WO2010/123577 discloses compositions and methods for evaluating cognitive deficits.

US2013/0295016 discloses methods for identifying and evaluating signatures in electroencephalographic oscillations that occur during onset of an exploratory activity in a subject. Existing methods require stimulus of the patient, or the administration of a therapeutic agent, prior to undertaking electroencephalographic analysis.

It would be advantageous to make a diagnosis of ASD early in life and quickly, to prevent the sometimes prolonged uncertainty that some parents experience before a diagnosis can be made or dismissed.

There is no current cure for ASD. However, there are a number of management options and interventions which are intended to improve the quality of life for the patient and their carer. Some of the main interventions include:

• Behavioural therapy;

• Speech and language therapy (for example developing alternative communication methods e.g. using pictures);

• Occupational therapy;

• Making changes to an individual's environment;

• Increasing the knowledge and understanding of the communication and learning style of the individual and the reasons behind particular behaviours;

• Adjusting teaching techniques; and

• Supporting social interaction.

In some cases, medication may also be prescribed to treat some of the symptoms or conditions associated with ASD, for example sleeping problems, depression and aggressive behaviour.

It would be advantageous to make a diagnosis of ASD before symptoms are well established, since this may increase the impact of the interventions discussed above.

It is an object of the present invention to obviate or mitigate one or more of the abovementioned problems.

Summary of the Invention

The present invention relates to EEG parameters indicative of Autism Spectrum Disorder (ASD) and is based in part on studies by the inventors in which they have shown that it is possible to differentiate between ASD individuals and non-ASD individuals using EEG parameters.

In a first aspect of the present invention there is provided a method for determining the presence of an Autism Spectrum Disorder (ASD) in a subject, determining the risk of a subject developing an ASD, determining progression of an ASD or assessing response to therapy of a subject with an ASD. The method of the present invention comprises the steps of: providing electroencephalography (EEG) signal data from the subject;

deriving one or more EEG parameters from the EEG signal data, wherein the

EEG parameter has a parameter value; and

analysing the one or more EEG parameters by comparing the parameter value with a control parameter value to determine whether the subject has an ASD, is at risk of developing an ASD, has an ASD which is progressing, or is responding to an ASD therapy; wherein the one or more EEG parameters are selected from the group comprising wavelet total power, wavelet coherence and wavelet phase coherence.

As will be understood by the skilled person EEG is an electrophysiological monitoring method which is used to record electrical activity of the brain by use of electrodes (or probes) placed on the scalp. In the method of the present invention, EEG signal data can be used to determine whether a subject from which the EEG signal data is derived has an ASD.

The EEG parameters of the invention may be extracted using non-linear time-series analysis methods.

Wavelet power provides an indication of the prevalence of each frequency band (or spectral band) in a signal. In preferred embodiments of the invention, the present invention uses the frequency spectrum of an EEG in the range between 0.8 Hz and 40 Hz divided into frequency bands similar to those conventionally used in clinical EEG. When referring to spectral bands, it is taken to mean the following: delta - 0.8 Hz to 4 Hz; theta - 4 Hz to 7.5 Hz; alpha - 7.5 Hz to 14 Hz; beta - 14 Hz to 22 Hz; and gamma - 22 Hz to 40 Hz. Wavelet coherence indicates the emergence and propagation of similar waves across the scalp of the subject, for a particular frequency band (or spectral band) between pairs of probes across the scalp. Wavelet phase coherence indicates the phase difference between same-frequency waves across the scalp of a subject, for a particular frequency band (or spectral band) between pairs of probes across the scalp.

The present inventors have shown that by analysing one or more of these EEG parameters, EEG signal data can surprisingly be used to determine the presence of an ASD in a subject.

In the present invention, the EEG parameters utilised (wavelet total power, wavelet coherence and/or wavelet phase coherence) may be derived for one or more spectral bands.

For example, in one embodiment, wavelet total power is derived for the alpha spectral band. Therefore, in such an embodiment, at least one of the EEG parameters is alpha wavelet total power. Alternatively, or in addition, wavelet total power may be derived for the delta, theta, beta and/or gamma spectral bands. In such embodiments, at least one of the EEG parameters is delta, theta, beta and/or gamma wavelet total power.

Alternatively, or in addition, wavelet coherence is derived for the alpha spectral band. Therefore, in such an embodiment, at least one of the EEG parameters is alpha wavelet coherence.

Alternatively, or in addition, wavelet coherence may be derived for the delta, theta, beta and/or gamma spectral bands. In such embodiments, at least one of the EEG parameters is delta, theta, beta and/or gamma wavelet coherence.

Alternatively, or in addition, wavelet phase coherence is derived for the alpha and/or theta spectral bands. Therefore, in such an embodiment, at least one of the EEG parameters is alpha and/or theta wavelet phase coherence. Alternatively, or in addition, wavelet phase coherence may be derived for the delta, beta and/or gamma spectral bands. In such embodiments, at least one of the EEG parameters is delta, beta and/or gamma wavelet phase coherence. In a preferred embodiment, the one or more EEG parameters comprises wavelet total power for the alpha spectral band, wavelet coherence for the alpha spectral band and/or wavelet phase coherence for the alpha and/or theta spectral bands. The inventors have demonstrated that this combination of EEG parameters is particularly informative when looking to determine the presence of an ASD in a subject.

Even more preferred is when the one or more EEG parameters comprises wavelet total power for the alpha spectral band, wavelet coherence for the alpha spectral band and wavelet phase coherence for the alpha and theta spectral bands. The skilled person will appreciate that alternative spectral bands may also be utilised in the method of the present invention. For example, the theta-alpha and beta-gamma spectral bands could be merged and then the spectrum divided in low (delta), medium (theta-alpha) and high (beta-gamma) frequency. As will be understood by the skilled person, EEG signal data is obtained by the use of electrodes which are placed on the scalp of a subject. The electrodes (or probes) utilised in the EEG studies described herein are designated: Fp1 , Fp2, F3, F4, F7, F8, Fz, T3, T4, T5, T6, C3, C4, Cz, P3, P4, Pz, 01 and 02. Such probes are positioned at various locations on the scalp, often according to the International 10/20 system (see Figure 2A), an internationally recognised protocol for the location of scalp electrodes. The skilled person will appreciate, however, that alternative probes may be utilised in the method of the present invention, for example those as in high-density arrays.

The 10/20 system for EEG probe placement is defined by the American Electroencephalographic Society (Guideline thirteen: Guidelines for standard electrode position nomenclature, published in the Journal of Clinical Neurophysiology, 1 1 : 11 1-113, 1994).

In the present invention, the EEG parameter wavelet total power may be derived for one or more EEG electrodes (or probes). In one embodiment, the EEG parameter wavelet total power is derived for one or more of the following probes: Fp1 , Fp2, F3, F4, F7, F8, Fz, T3, T4, T5, T6, C3, C4, Cz, P3, P4, Pz, 01 and 02.

In a preferred embodiment, the EEG parameter wavelet total power may be derived for the T4 and/or 02 probe.

In a particularly preferred embodiment, the EEG parameter wavelet total power may be derived for the T4 and/or 02 probe in the alpha spectral band. However, the skilled person will appreciate that any combination of probes and spectral bands may be suitable. So, for example, the EEG parameter wavelet total power may be obtained for the T4 and/or 02 probe in the delta, theta, beta and/or gamma spectral bands.

The EEG parameters wavelet coherence and wavelet phase coherence may be derived for one or more pairs of EEG electrodes (or probes). In one embodiment, the EEG parameters wavelet coherence and wavelet phase coherence may be derived for one or more pairs of EEG electrodes wherein the electrodes are selected from the group comprising: Fp1 , Fp2, F3, F4, F7, F8, Fz, T3, T4, T5, T6, C3, C4, Cz, P3, P4, Pz, 01 and 02. In a preferred embodiment, the EEG parameter wavelet coherence may be derived for the probe pair Cz and C4.

In a particularly preferred embodiment, the EEG parameter wavelet coherence maybe derived for the probe pair Cz and C4 in the alpha spectral band. However, the skilled person will appreciate that any combination of probe pairs and spectral bands may be suitable. So, for example, the EEG parameter wavelet coherence may be derived for the probe pair Cz and C4 in the delta, theta, beta and/or gamma spectral bands.

In a preferred embodiment, the EEG parameter wavelet phase coherence may be derived for the probe pair Cz and F3 and/or Cz and F4.

In a particularly preferred embodiment, the EEG parameter wavelet phase coherence may be derived for the probe pair Cz and F3 in the alpha spectral band. Alternatively, or in addition, the EEG parameter wavelet phase coherence may be derived for the probe pair Cz and F4 in the theta spectral band. ln a preferred embodiment of the invention, the method comprises deriving between one and nine EEG parameters, preferably between three and seven EEG parameters, more preferably five EEG parameters.

In an alternative preferred embodiment of the invention, the method comprises deriving between only one and nine EEG parameters, preferably between only three and seven EEG parameters, more preferably only five EEG parameters.

In an embodiment of the invention in which the method comprises deriving five EEG parameters, the EEG parameters may preferably comprise: i) wavelet total power for probe T4 in the alpha spectral band;

ii) wavelet total power for probe 02 in the alpha spectral band;

iii) wavelet coherence for probe pair Cz and C4 in the alpha spectral band;

iv) wavelet phase coherence for probe pair Cz and F3 in the alpha spectral band; and

v) wavelet phase coherence for probe pair Cz and F4 in the theta spectral band.

In an embodiment of the invention in which the method comprises deriving five EEG parameters, the EEG parameters may consist of: i) wavelet total power for probe T4 in the alpha spectral band;

ii) wavelet total power for probe 02 in the alpha spectral band;

iii) wavelet coherence for probe pair Cz and C4 in the alpha spectral band;

iv) wavelet phase coherence for probe pair Cz and F3 in the alpha spectral band; and

v) wavelet phase coherence for probe pair Cz and F4 in the theta spectral band.

As will be appreciated, the probes referred to are EEG probes and placements are in accordance with the 10/20 system referred to above.

In testing performed by the present inventors, it has been shown that such a selection of EEG parameters provides a sensitivity of 100% for distinguishing between subjects with and without an ASD. However, it will be appreciated by the skilled person that it is not necessary for the method of the present invention to provide a sensitivity of 100%. Preferably the method of the present invention provides a sensitivity of at least 60%, at least 70%, at least 80% or at least 90%.

As will be appreciated by the skilled person, the EEG signal data provided in step i) of the method of the first aspect of the invention may be provided from an EEG signal database, for example. In an alternative, the method of the present invention may further comprise a step of obtaining EEG signal data from a subject. EEG is a technique commonly used in clinical settings and therefore the skilled person will appreciate how such obtaining is performed. In general such obtaining makes use of an EEG device including electrodes (or probes) which are applied to a subject's scalp.

In embodiments, the EEG signal data from the subject is obtained from the subject in a resting state. For example, the step of obtaining EEG signal data from a subject may take place whilst the subject is in a resting state (or at rest). Similarly, the EEG data provided in step i) of the method may originally have been obtained from a subject in a resting state (or at rest). For example, the subject may be sitting or lying down and may not be exposed to external stimuli such as sounds, images, flashing lights etc. By providing or obtaining the EEG signal data from a patient at rest, differences in the 'default network' can be observed. Although the provision of external stimuli may enhance characteristics that are important for diagnosis, a more complex response may be produced which may mask important features of the data. Providing or obtaining EEG signal data from a patient at rest avoids stimulus which could insert bias due to inter-subjects approach/interest in the stimulus (especially in an age of brain maturation that is pre-school). In the method of the present invention, the step of deriving one or more EEG parameters from the EEG signal data comprises an integral transformation, in particular a wavelet transformation.

The use of a wavelet transform provides particular advantages over other transforms, such as Fourier transforms. Fourier transform, for example, does not take into account time- variability and assumes strict periodicity which is not present in the EEG signals. Although a short time Fourier transform could be utilised, the constant window would not be fully appropriate. It has been found that the use of a wavelet transform provides optimal time localisation and optimal time-frequency resolution when calculating power. Utilising a wavelet transform also allows investigation of time-dependent phase dynamics. The wavelet transformation is applied to the EEG signal (x). In embodiments, the mother wavelet ψ selected is the complex Morlet (shown in equation 1 below) with a central frequency of 1 Hz.

The use of equation 1 results in a matrix of complex coefficients having n columns and / rows, with n number of time samples of the signals and / number of frequency steps investigated. For every time-frequency pixel of the matrix, the coefficients are given by equation 2 shown below.

WT(f, t) - j (/, (t T)):d 7 k r (2) Although a wavelet transform can be obtained by many different wavelets, the Morlet wavelet has been shown to be particularly advantageous because of the relatively straightforward relationship between scale and frequency. Due to its shape, the complex Morlet wavelet is a suitable choice when attempting to identify an oscillation in the signal. Use of the Morlet wavelet also allows the pairing of amplitude and phase investigation. A central frequency (or resolution parameter) of 1 Hz was utilised since this was found to be the optimum based on the nature of the frequencies investigated.

The EEG parameters of the present invention comprise a parameter value. In determining whether the subject has an ASD, is at risk of developing an ASD, has an ASD which is progressing, or is responding to an ASD therapy, this parameter value is compared to a control parameter value. As the skilled person will appreciate, the origin of the control parameter will depend upon the particular subject being tested. However, the control parameter may be derived, for example, from EEG signal data obtained from a subject unaffected by an ASD. The control parameter may be obtained from EEG signal data obtained from an age-matched subject and/or a subject of the same sex.

In general, a higher power in the gamma band for the central-parietal probes (for example C3, Cz, C4, P3, Pz and P4), a lower power in the alpha band for the temporal- occipital probes (for example T3, T4, T5, T6, 01 and 02) and a lower power in the theta band for the temporal-central-parietal areas (T3, T45, T5, T6, C4, Cz, P4, Pz) are indicative of the presence of an ASD. ln embodiments, the control parameter value may be a threshold value or range and determining whether the subject has an ASD, is at risk of developing an ASD, has an ASD which is progressing, or is responding to an ASD therapy will be determined based on whether the parameter value is higher or lower than the threshold value or range.

In embodiments where the one or more EEG parameters is wavelet total power for probe T4 in the alpha spectral band, an appropriate threshold range is around 10-15%, preferably around 12-13%, more preferably 12.1-12.7%, or 12.16-12.64%, or in the range of 12.16-13.67%. A most preferred threshold value (or control parameter value) is around 12.5%, preferably 12.55%. In such embodiments of the invention, a parameter value of less than or equal to the threshold range or threshold value is indicative of the presence of an ASD, whereas a parameter value of more than the threshold range or threshold value is indicative of the absence of an ASD. Such threshold ranges may be used to indicate the severity of the ASD. For example, in such instances, a value nearer to the top value of the range (for example around 15%) may be indicative of an "almost normal" case of ASD, whereas a lower value (for example around 10% or lower) may be indicative of a "severe" case of ASD. In embodiments where the one or more EEG parameters is wavelet total power for probe 02 in the alpha spectral band, an appropriate threshold range is around 18-24%, preferably around 20-22%, more preferably around 20.2-21.7% or in the range of 20.21- 21.70%. A most preferred threshold value (on control parameter value) is around 21.6%. In such embodiments, a parameter value of less than or equal to the threshold range or threshold value is indicative of the presence of an ASD, whereas a parameter value of more than the threshold range or threshold value is indicative of the absence of an ASD. Such threshold ranges may be used to indicate the severity of the ASD. In such instances, a value nearer to the top value of the range (for example around 24%) may be indicative of an "almost normal" case of ASD, whereas a lower value (for example around 18% or lower) may be indicative of a "severe" case of ASD.

In embodiments where the one or more EEG parameters is wavelet coherence for probe pair Cz and C4 in the alpha spectral band, an appropriate threshold range is around 45-60%, preferably around 47-59% or 47-50%, more preferably around 47.4-49.2% or 47.41- 49.18%. A most preferred threshold value (or control parameter value) is around 48.9%, preferably 48.88%. In such embodiments, a parameter value of more than or equal to the threshold range or threshold value is indicative of the presence of an ASD, whereas a parameter value of less than the threshold range or threshold value is indicative of the absence of an ASD. Such threshold ranges may be used to indicate the severity of the ASD. In such instances, a value nearer to the top value of the range (for example around 60% or greater) may be indicative of a "severe" case of ASD, whereas a lower value (for example around 45%) may be indicative of an "almost normal" case of ASD.

In embodiments where the one or more EEG parameters is wavelet phase coherence for probe pair Cz and F3 in the alpha spectral band, an appropriate threshold range is around 24-36%, more preferably 27-31 %, more preferably 27.7-30.9% or 27.79-30.91 %. A most preferred threshold value (or control parameter value) is around 28.2%, preferably 28.22%. In such embodiments, a parameter value of more than or equal to the threshold range or threshold value is indicative of the presence of an ASD, whereas a parameter value of less than the threshold value is indicative of the absence of an ASD. Such threshold ranges may be used to indicate the severity of the ASD. In such instances, a value nearer to the top value of the range (for example around 36% or greater) may be indicative of a "severe" case of ASD, whereas a lower value (for example around 24%) may be indicative of an "almost normal" case of ASD. In embodiments where the one or more EEG parameters is wavelet phase coherence for probe pair Cz and F4 in the theta spectral band, an appropriate threshold range is around 0-35%, or around 0-32%, more preferably around 20-30%, more preferably around 25.4- 29.4% or 25.45-29.34%. A most preferred threshold value (or control parameter value) is around 29.0%. In such embodiments, a parameter value of more than or equal to the threshold range or threshold value is indicative of the presence of an ASD, whereas a parameter value of less than the threshold value is indicative of the absence of an ASD. Such threshold ranges may be used to indicate the severity of the ASD. In such instances, a value nearer to the top value of the range (for example around 35% or greater) may be indicative of a "severe" case of ASD, whereas a lower value (for example around 20%) may be indicative of an "almost normal" case of ASD.

In embodiments of the present invention in which multiple EEG parameters are analysed, it is not necessary for every EEG parameter to be indicative of an ASD for an ASD diagnosis to be obtained. For example, in embodiments where five EEG parameters are analysed it may be possible to make an ASD diagnosis when only three EEG parameters are indicative of the presence of an ASD. Preferably, at least 50% of the EEG parameters analysed are indicative of ASD to diagnose the presence of an ASD in a subject, more preferably at least 70%, more preferably at least 90%, most preferably 100%.

In embodiments where the method is being used to determine whether an ASD is progressing, the method may comprise monitoring over time to determine whether an ASD has progressed. In such an embodiment, an initial EEG parameter analysis (as described above) may be compared with one or more EEG parameter analyses undertaken on EEG signal data obtained later in time. The EEG parameter value(s) obtained may change over time, to be further removed (either higher or lower) from a control parameter value or threshold range. Alternatively, in examples where multiple EEG parameters are utilised, for example five EEG parameters, in an initial EEG parameter analysis only three of the five EEG parameter values may be indicative of ASD. In the later EEG parameter analysis an increased number of EEG parameter values may be indicative of ASD, which may indicate that ASD is progressing.

In embodiments where the method is being used to determine whether an ASD is responding to therapy, the method may comprise monitoring over time to determine whether an ASD is responding to therapy. In such an embodiment, an initial EEG parameter analysis (which may be before therapy has commenced) may be compared with one or more EEG parameter analyses undertaken on EEG signal data obtained later in time (for example after therapy has commenced). The EEG parameter value(s) obtained may change over time, to be further removed (either higher or lower) from a control parameter value or threshold range. Alternatively, in examples where multiple EEG parameters are utilised, for example five EEG parameters, in an initial EEG parameter analysis, five out of the five EEG parameter values may be indicative of ASD. In the later EEG parameter analysis, fewer EEG parameter values may be indicative of ASD, which may indicate that an ASD is responding to therapy. Alternatively, in the later EEG parameter analysis, the number of EEG parameter values indicative of ASD may remain the same or increase, which may indicate that an ASD is not responding to therapy.

The method of the present invention may also be used to determine whether a drug is effective at treating ASD, in a similar manner to that described above in relation to determining whether an ASD is responding to therapy. The use of "effective" is used to indicate that a treatment reduces or alleviates signs or symptoms of ASD, improves the clinical course of the disease, decreases the number or severity of exacerbations or reduces any other objective or subjective indicia of the disease. As mentioned above, there is no current cure for ASD; instead, treatment focusses on management options and interventions which are intended to improve the quality of life for the patient and their carer. However, some drugs are used in the management of ASD (for example in treating comorbidities such as anxiety and depression) and the method of the present invention can be used to determine whether such drugs, in addition to other drugs developed to treat ASDs are effective.

The method of the present invention may also be used to determine if a subject is at risk of developing an ASD. In such embodiments, a parameter value as discussed above (for example a phase coherence value) may be compared to a control parameter value from (i) an ASD subject and/or (ii) a subject unaffected by an ASD. In such embodiments, a parameter value more similar to a control parameter value from an ASD subject than to a control parameter value from a subject unaffected by an ASD may be indicative of a high risk of the subject developing an ASD. Alternatively, a parameter value more similar to a control parameter value from a subject unaffected by an ASD than to a control parameter value from an ASD subject may be indicative of a low risk of the subject developing an ASD.

Since a diagnosis of a disease is often not based on the results of a single test alone, the method of the present invention may be used to determine whether a subject is more likely than not to have an ASD based on comparison of one or more EEG parameters having a parameter value with a control parameter value. Thus, for example, a patient with a putative diagnosis of an ASD may be diagnosed as being "more likely" or "less likely" to have an ASD in light of the information provided by the method of the present invention. The present invention may therefore be used to assist a clinician with the diagnosis of an ASD.

The method of the present invention may, in certain embodiments, comprise detecting other signs or symptoms of ASD, conducting clinical tests of ASD and/or measuring other ASD markers. For example, the method of the present invention may comprise performing traditional testing such as clinical recognition of significant problems with an individual's communication, social interaction and flexible thought processes by clinicians and/or cognitive and motoric tests.

As will be appreciated by the skilled person, the above description is not limited to making an initial identification (or diagnosis) of an ASD in a subject, but is also applicable to confirming a provisional diagnosis of an ASD or "ruling out" such a diagnosis. ln embodiments of the invention, the ASD is autism, Asperger syndrome or pervasive developmental disorder not otherwise specified (PDD-NOS).

In a second aspect of the invention there is provided a system for determining the presence of an Autism Spectrum Disorder (ASD) in a subject, determining the risk of a subject developing an ASD, determining progression of an ASD or assessing response to therapy of a subject with an ASD. The system of the second aspect of the invention comprises a processor which is arranged to:

i) process EEG signal data from a subject to obtain one or more EEG parameters, the one or more EEG parameters each having a parameter value; and

ii) analyse said one or more EEG parameters by comparing the parameter value with a control parameter value to determine whether the subject has an ASD, is at risk of developing and ASD, has an ASD which is progressing, or is responding to an ASD therapy; wherein the one or more EEG parameters are selected from the group comprising wavelet total power, wavelet coherence and wavelet phase coherence. In embodiments, the system may further comprise an electroencephalography (EEG) component arranged to obtain EEG signal data from a subject. The EEG system may be an EEG device, for example. Such devices will be well known to the skilled person.

The processor could comprise one or more functional or physical parts, for example one for processing the data, one for analysing the data, etc. In embodiments, the processor is further arranged to provide an output indicating whether the subject has an ASD, is at risk of developing an ASD, has an ASD which is progressing, or is responding to therapy. In embodiments, the output may be provided on a display. In embodiments, the processor and display may be integrated in a single device. In further embodiments, the EEG system may also be integrated into said single device. However, in alternative embodiments, the EEG system, processor and display may each be single devices which are interconnected to form the system of the invention. The system of the second aspect of the invention may be a system which is configured to implement the method of the first aspect of the present invention. ln embodiments of the invention, the ASD is autism, Asperger syndrome or pervasive developmental disorder not otherwise specified (PDD-NOS).

In yet a third aspect of the invention there is a provided a kit of parts for performing the method of the first aspect of the invention, the kit of parts comprising:

(i) an electroencephalography (EEG) device; and

(ii) a processor for processing EEG signal data to obtain one or more EEG parameters, and analysing said one or more EEG parameters.

The present invention also provides a method comprising the steps of:

(i) providing electroencephalography (EEG) signal data from a subject;

(ii) deriving one or more EEG parameters from the EEG signal data, wherein the EEG parameter has a parameter value; and

(iii) analysing the one or more EEG parameters by comparing the parameter value with a control parameter value; wherein the one or more EEG parameters are selected from the group comprising wavelet total power, wavelet coherence and wavelet phase coherence.

In a yet further aspect of the invention there is provided a computer readable medium comprising code configured to implement the method of the abovementioned aspects of the invention. The system or kit of parts described above may be, comprise or use the computer readable medium. The computer readable medium may be a non transitory computer readable medium.

There is also provided a computer program configured (e.g. coded) to implement the method of the abovementioned aspects of the invention. The system or kit of parts described above may comprise or use the computer program.

The described and illustrated embodiments are to be considered as illustrative and not restrictive in character, it being understood that only the preferred embodiments have been shown and described and that all changes and modifications that come within the scope of the inventions as defined in the claims are desired to be protected. Moreover, any one or more of the above described preferred embodiments could be combined with one or more of the other preferred embodiments to suit a particular application.

It should be understood that while the use of words such as "preferable", "preferably", "preferred" or "more preferred" in the description suggest that a feature so described may be desirable, it may nevertheless not be necessary and embodiments lacking such a feature may be contemplated as within the scope of the invention as defined in the appended claims. In relation to the claims, it is intended that when words such as "a," "an," or "at least one," are used to preface a feature there is no intention to limit the claim to only one such feature unless specifically stated to the contrary in the claim.

Detailed Description of the Invention

The present invention will now be further described with reference to the following figures which show:

Figure 1 : Flowchart showing an example embodiment of the present invention in which summation of parameter scores are used to make a diagnosis of Autism Spectrum Disorder.

Figure 2: (A) Location of EEG probes on scalp according to the International standard 10-20; (B) Example of an EEG signal; (C) Power spectrum of EEG signal of B; (D) Wavelet transform of EEG signal of B.

Figure 3: Box plot showing differences between control group (CG) and Autism Spectrum Disorder (ASD) group for exemplary EEG parameters of the present invention. (A) Parameter: Wavelet power, probe T4, alpha. (B) Parameter: Wavelet power, probe 02, alpha. (C) Parameter: Wavelet phase coherence, probe pair Cz-F3, alpha. (D) Parameter: Wavelet phase coherence, probe pair Cz-F4, theta. (E) Parameter: Wavelet coherence, probe pair Cz-C4, alpha.

Figure 4: Scores given to each subject by an example EEG parameter algorithm of the invention. Figure 5: Brain activity maps showing significant difference between ASD group and control group (CG) in the power of theta, alpha and gamma. Circled regions = CG>ASD, Squared regions = ASD>CG. Figure 6: Brain activity maps showing significant differences between ASD group and control group (CG) in the phase coherence of the different bands. Left column = ASD>CG; right column = CG>ASD; thicker lines indicate greater statistical significance).

Figure 7: Brain activity maps showing significant differences between ASD and control group (CG) in the wavelet coherence of the different bands. Left column = ASD>CG; right column = CG>ASD; thicker lines indicate greater statistical significance).

Figure 8: A block diagram showing an example embodiment of the system of the present invention.

The present inventors undertook testing of the use of EEG parameters in determining the presence of ASD in a test cohort. The EEG parameters were analysed in a group of male children having a previous clinical diagnosis of ASD and a control group (CG). The two groups were selected according to the following criteria:

Inclusion criteria:

• A clear diagnosis of ASD (for ASD group) or normal development (for control group);

• Aged between 3 years, 0 months and 4 years, 11 months; and

• Parents willing and able to provide informed consent.

Exclusion criteria:

• Epilepsy or undiagnosed seizure episodes;

• The use of medications known to affect brain function;

• Structural brain and chromosome abnormalities identified in the subject;

· Uncertainty regarding diagnosis or developmental progress; or

• Having a first degree relative with ASD diagnosis (for control group only).

In total, a cohort of 13 ASD male subjects (age: 4 years, 0.0 months ±7.6 months) were involved as the study group and 9 male subjects as the control group (age: 4 years, 1.9 months ± 5.4 months). Recordings

Electroencephalography (EEG) is a non-invasive technique whose use is very common in clinical settings. An advantage of EEG is that it allows high frequency resolution in recording the electric signals of the brain which result from activity of neuronal networks in the brain. The use of EEG also allows higher frequencies to be investigated (compared to those investigated with fMRI, for example).

The EEG technique utilised in the examples below makes use of a number of electrodes (or probes) placed on the scalp. These probes measure voltage fluctuations resulting from ionic current within the neurons of the brain. The probes used in EEG are designated: Fp1 , Fp2, F3, F4, F7, F8, Fz, T3, T4, T5, T6, C3, C4, Cz, P3, P4, Pz, 01 and 02, and are placed as shown in Figure 2. In order to ensure the presence of a window of continuous resting state during the testing procedure, recordings of around 30 minutes were acquired from each subject. The subject was either sitting on a parent's lap or independently sitting on a chair.

Equipment

During testing a continuous EEG system was used: Nicolet cEEG (Viasys Healthcare). 19 channels recording the signals at 256 Hz frequency resolution were utilised.

Montage

The 19 probes were placed on the scalp according to the International 10/20 system (shown in Figure 2A), an internationally recognised protocol for the location of scalp electrodes. Pre-processing

Signal outputs of the EEG system were visually inspected and, at the same time, a video of the acquisition session was viewed to select a one minute window in which the subject was in a resting condition with minimal movements and interactions with the environment. A sample of a selected time frame is shown in Figure 2B. Wavelet Transform

The use of a wavelet transform provides particular advantages over other transforms, such as Fourier transforms. Fourier transform, for example, does not take into account time- variability and assumes strict periodicity which is not present in the EEG signals. Although a short time Fourier transform could be utilised, the constant window would not be fully appropriate. It has been found that the use of a wavelet transform provides optimal time localisation and optimal time-frequency resolution when calculating power. Utilising a wavelet transform also allows investigation of time-dependent phase dynamics.

In this testing protocol, a Wavelet transform (WT) was applied to the output signal x. The mother wavelet ψ selected was the complex Morlet (shown in equation 1 below) with a central frequency of 1 Hz. The range of frequencies investigated was from 0.8 Hz to 40 Hz. This method provides a matrix of complex coefficients having n columns and / rows, with n number of time samples of the signals and / number of frequency steps investigated. For every time-frequency pixel of the matrix, the coefficients are given by equation 2 shown below.

WT

The method allows adjustable time-frequency resolution, chosen to be logarithmic in this particular example. This is the optimal feature for investigating such a broad spectrum of frequencies. Absolute values of the complex coefficients describe the power of the wave having a specific frequency at every sampling time (see Figure 2C), while their angles give information about the phase of the wave. Time average of these matrices give one-dimensional vectors (from n to 1 columns) or power spectral density (see Figure 2D) and phase coherence.

Wavelet Power Power investigations allow determination of the prevalence of each frequency band in a signal. Due to the amplitude of the oscillations being measured in arbitrary units, and because it is strongly dependent on features such as skin/hair characteristics, each power spectrum was normalised to a unitary total area before being compared.

Power values for each spectral band were computed as follows:

• The power spectral density of each signal was obtained by averaging the square of the absolute value of the WT along the columns (from n to 1 columns);

· Every point of the spectrum was divided by the original total spectral area, so that the total area of the normalised spectrum had unitary value; and

• For each frequency band, the relative power content was evaluated by calculating which percentage of the total power was contained in each band. Wavelet Coherence

The inventors evaluated the emergence and propagation of similar waves across the scalp by means of wavelet coherence, for various frequency bands, between pairs of probes across the scalp. The wavelet coherence was computed for various combinations of different probes.

Wavelet coherence is computed as shown in equation 3 below:

WCif) = < U / W/. / _{; .} ,jH ^' 7 ^ /. / _{; .} >< I I /r(/. / _; . ,,j H 7 :> /. / : _{. .} ,J >

(3) · Multiplying correspondent values in WT1 and complex conjugate of WT2;

• Averaging over time;

• Multiplying correspondent values in WT2 and complex conjugate of WT1 ;

• Averaging over time;

• For each frequency value, calculating the square root of the product between these two quantities; and

• Averaging the coherence values within the investigated bands.

Wavelet Phase Coherence The inventors also evaluated the consistency of the phase difference between same- frequency waves across the scalp by the mean of phase coherence, for various frequency bands, between pairs of probes across the scalp. Phase coherence was computed for each combination of different probes.

Phase coherence is computed as shown in equation 4 below:

• Obtaining the matrix of angles of the complex coefficients;

• Subtracting the matrix for every pair of probes, obtaining φ;

· For every frequency value, computing the square of time-average values of sine of φ;

• For every frequency value, computing the square of time-average values of cosine of φ;

• Calculating the square root of the sum of these two quantities.

WPC(f) - < *¾n(<2 ⁾ ( ....„)} > ² + < a½(/, 0( ... _n )) > ² (4) · Averaging the coherence values within the investigated spectral bands.

Definition of the marker

In this example, to enable the inventors to analyse the ability of selected parameters to distinguish between the ASD group and the control group, statistical analysis was performed on the results using the Wilcoxon unpaired text.

In this particular study, five parameters describing strongly significant differences between the two groups (p<0.005) (see Figure 3) were found. The parameters used in this study were:

• Wavelet phase coherence for the probes Cz-F4: theta, probes Cz-F3: alpha;

• Wavelet coherence for the probes Cz-C4: alpha; and

• Wavelet power for the probes T4 and 02: alpha.

Values for the groups, p-values, medians and thresholds are reported in Table 1 below. Box- plots showing the values reported are shown in Figure 3. Table 1 | Table of significance values, group values and medians and corresponding thresholds for the selected parameters. Underlined values indicate a wrong sub-classification (based on the given threshold).

In this example, for each subject, the thresholds have been chosen around the midpoints between the median of the groups, and a partial score of minus one is given for each parameter when lying on the side of the threshold closer to the median of the control group (CG) and plus one is given if it is on the side of the threshold closer to the median of the ASD group distribution. Therefore only odd-numbers scores are possible. A negative score indicates the prevalence of CG features (full score is minus five), and a positive score the prevalence of ASD features (full score is plus five).

Performance

Figure 1 depicts an example embodiment of the method of the present invention which was used with the data derived in the testing described above. In a first step 1 , an EEG signal dataset was provided and was computed using a wavelet transformation. This transformation was performed using the data derived from EEG probes Cz, C4, F4, T3, T4 and 02 in the frequency interval of 0.8 - 40 Hz. As mentioned above, in this study five EEG parameters were utilised:

• Wavelet phase coherence for the probes Cz-F4: theta, probes Cz-F3: alpha;

• Wavelet coherence for the probes Cz-C4: alpha; and

· Wavelet power for the probes T4 and 02: alpha.

To derive these EEG parameters and corresponding EEG parameter values, the wavelet transform data was computed using the equations set out above to derive total power, wavelet coherence and wavelet phase coherence 3, 5, 7, 9, 1 1. Next, each EEG parameter was compared against a corresponding threshold value 13, 15, 17, 19, 21. Based on this comparison, each marker was allocated a partial score of 1 if the parameter was indicative of ASD 23, or a partial score of -1 if the parameter was indicative of non-ASD 25. The five partial scores were then summed 27. If the sum of the partial scores was greater than 0, the subject was marked as an ASD subject 29, whereas if the sum of the partial scores was less than 0, the subject was marked as a non-ASD subject 31.

As shown in the box plots of Figure 3, each EEG parameter selected (Figure 3A-E) was able to distinguish between the ASD group and control group (CG). Using the example embodiment of the invention shown in Figure 1 , the EEG parameters discriminated between the ASD subjects and control group subjects with a sensitivity of 100% and a sensibility of 100%. All the ASD subjects were identified correctly as ASD, and all of the control group subjects were identified correctly as non-ASD (see Figure 4).

Out of the thirteen correctly identified ASD subjects, seven of the thirteen subjects gained a full score of 5, three of the thirteen subjects gained a score of 3 and a further three of the thirteen subjects gained a score of 1 (Figure 4). Out of the nine correctly identified control group subjects, six of the nine subjects gained a full score of -5, one of the nine subjects gained a score of -3 and two of the nine subjects gained a score of -1 (Figure 4).

The results are summarised in Table 2 (below) and Figure 4.

Table 2 | Table of marker performances, with success rates and individual scores given to every subject.

As shown, the protocol described above can be used to distinguish between ASD patients and a control group using specific EEG parameters. As described above, the technique is based on different electrical activity in the brain. Brain maps showing the different electrical activity between the ASD patients and the control group are shown in Figures 5-7. These brain maps show significantly different values of wavelet power, wavelet phase coherence and wavelet coherence for the ASD group and the control group, for the different spectral bands used. Figure 8 shows a block diagram depicting a system in accordance with an embodiment of the invention. The system 35 comprises an EEG system 35 which could be an EEG device, for example a continuous EEG system such as a Nicolet cEEG (Viasys Healthcare). The EEG system 35 is used to obtain EEG signal data from a subject of interest, for example, a patient to be tested for ASD. There is further included a processor 37, which is used to process EEG signal data received from the EEG system 35. Processing of the EEG signal data by the processor 37 results in at least one EEG parameter being derived. The processor 37, in this example, takes the EEG signal data and performs a wavelet transform on the data to obtain processed EEG signal data. In this example, this processed signal data is then further processed by way of wavelet power, wavelet coherence and wavelet phase coherence equations (as outlined above) to produce EEG parameters (wavelet power, wavelet coherence and wavelet phase coherence). The processor 37 further analyses the EEG parameters by comparing a parameter value to a control parameter value. This processing allows determination as to whether the subject from which the EEG data is derived has an ASD. The result of the processing is displayed to a user of the system via display 39.

Although the system 33 is shown as being formed from separate components 35, 37 and 39, it will be appreciated by the skilled person that the processor 37 and the display 39 could be integrated in a single device, for example a computer. Similarly, the EEG system 35 could be an integral part of a single device which also includes the processor 37 and display 39.

It will be appreciated that numerous modifications to the above described method may be made without departing from the scope of the invention as defined in the appended claims. For example, although in the example shown above the invention is described in relation to combinations of particular EEG parameters for particular spectral bands and particular EEG probes, it will be appreciated that other combinations of EEG parameters could be utilised. For example, various combinations of wavelet power, wavelet phase coherence and wavelet coherence could be utilised for spectral bands and EEG probes other than those described in the example given above.