Brain consciousness state determination

Abstract

An apparatus for determining a brain consciousness state of a patient, the apparatus includes one or more processors, the apparatus configured to: determine sensor data associated with an electrode arrangement, the sensor data indicative of a change in a corneo-retinal standing potential of an eye of the patient caused by ocular microtremor; and determine, based on the sensor data, the brain consciousness state of the patient.

Description

[0001] BRAIN CONSCIOUSNESS STATE DETERMINATION

[0002] FIELD AND BACKGROUND

[0003] The present invention relates to brain consciousness state determination, and more particularly apparatuses, methods, computer-readable storage media, computer programs, and systems for determining a brain consciousness state of a patient.

[0004] Determining information relating to a patient, such as a brain consciousness state, can be useful for various reasons. For example, a brain consciousness state, such as a sleep state, sedation state and / or a brain death state, can be used to inform a medical procedure or research.

[0005] The techniques discussed herein provide improved techniques for brain consciousness state determination and increased accuracy across a wide variety of patient scenarios and clinical settings.

[0006] SUMMARY

[0007] Aspects of the invention are set out in the accompanying claims.

[0008] The present inventor has identified a novel technique to determine a brain consciousness state of a patient, which is based on using electrooculography (EOG) to measure ocular microtremor (OMT) of an eye of the patient.

[0009] OMT is a continuous, low amplitude, high frequency involuntary tremor of the eye that correlates with pontine activation (and hence brain consciousness), and underlies micro-drift and microsaccades. OMT typically occurs at a frequency from about 30 Hz to 100 Hz, where the eye is estimated to move between approximately 150 and 2000 nm. OMT has the smallest amplitude of all eye movements, typically about four thousandths of a degree, and an amplitude significantly smaller than microsaccades.

[0010] EOG is a non-invasive technique for measuring the standing electrical potential that exists between the cornea and retina of the eye (a corneo-retinal standing potential). Electrodes can be placed on the skin surrounding the eye, and the electrodes detect a change in electrical potential as a result of movement of the eyeball. In particular, the eye can act as a dipole, whereby the retina may be negatively charged and the cornea may be positively charged. When the eye moves, the orientation and movement of the dipole relative to the skull (and thus the electrodes) can be measured using the electrodes.

[0011] EOG has not previously been used to measure OMT. There has been a long-felt technical prejudice that EOG is not suitable for measuring low amplitude movements of the eye, for example because the amplitude may be near the level of noise in the EOG system. Furthermore, electroencephalography (EEG) has long been established as the technique for determining a brain consciousness state of a patient due to its ease of use. However, the present inventor has identified that EOG can be used to measure OMT, and has further realised that the acquired data can be used to determine a brain consciousness state of the patient in a more accurate, efficient, and non-invasive manner than existing techniques to determine a brain consciousness state.

[0012] While the present techniques may be used to determine a variety of brain consciousness states, such as a sleep state, a sedation state and a brain death state, the present techniques have particular application to the field of anaesthesia and patient sedation, where existing approaches to determine a level of sedation of a patient are manual, subjective, inefficient and in some cases invasive.

[0013] Viewed from a first aspect, there is provided an apparatus for determining a brain consciousness state of a patient, the apparatus comprising one or more processors, the apparatus configured to: determine sensor data associated with an electrode arrangement, the sensor data indicative of a change in a corneo-retinal standing potential of an eye of the patient caused by ocular microtremor; and determine, based on the sensor data, the brain consciousness state of the patient.

[0014] Hence, EOG is used to detect OMT of the patient’s eye to determine a brain consciousness state of the patient. The present inventor has counter-intuitively realised that EOG can be used to non- invasively measure the OMT of a patient’s eye, and that the measured OMT can be indicative of a brain consciousness state of the patient.

[0015] This provides numerous advantages over existing approaches. For example, the brain consciousness state can be measured more accurately and in a non-invasive way, and accuracy is increased across a wide variety of patient scenarios (such as different injury and illness types) and clinical settings (such as elective and intensive care). This also allows for the accurate determination of different levels of patient sedation using a non-invasive technique.

[0016] Indeed, the present inventor has realised that techniques that use EEG to determine a brain consciousness state can be inaccurate and unsuitable for a variety of patient scenarios and clinical settings, and in particular for patients with brain injuries and / or patients in intensive care. This is because EEG signals can be confounded by the brain injury or the underlying disease / pathology that caused the patient to be admitted to intensive care in the first place. For example, infection, sepsis, trauma, burns, and metabolic derangements can all affect an EEG signal from a patient. Hence, EEG may not be suitable for determining a brain consciousness state, such as a level of sedation, of a patient with a brain injury or in intensive care.

[0017] In contrast, the present techniques use EOG to measure OMT. OMT is downstream of many of the processes that confound EEG signals. The OMT signal originates from the base of the brain, in the pons. It is understood that the inputs to the pons, from higher cortical centres, are noisy. As the eye motor control is directly linked to the pons, the orbit tremors in a low-amplitude, high frequency, manner that correlates with this noise. Thus, the microtremor of the eye (i.e. the OMT) correlates with the activity of cortical areas. The electrical signal associated with the OMT can thus be used to characterise a patient’s brain state, and determine different brain consciousness states of the patient and also the depth of sedation in anaesthesia (see S. Bojanic, T. Simpson, and C. Bolger, ‘Ocular microtremor: A tool for measuring depth of anaesthesia?’, Br. J. Anaesth., vol. 86, no. 4, pp. 519-522, Apr. 2001 , doi: 10.1093 / bja / 86.4.519, which is hereby incorporated by reference).

[0018] Hence, using EOG to detect OMT can provide increased accuracy of the brain consciousness state determination. Further, the present technique can be used in intensive care settings or in patients having brain injuries with a reduced likelihood that the accuracy of the technique will be affected (for example as a result of spurious signals).

[0019] Further, measuring OMT using EOG is non-invasive, and can rely on an arrangement of skinsurface electrodes placed around the eye(s) of a patient. As such, invasive techniques, such as placing a piezoelectric crystal directly on the eyeball or using a probe placed on the sclera can be avoided. This increases the efficiency of the determination process and increases patient comfort.

[0020] In some examples, the electrode arrangement is an electrooculography electrode arrangement. Hence, as described, EOG electrodes can be used to efficiently determine the sensor data (which is indicative of the OMT present in the eye).

[0021] In some examples, the corneo-retinal standing potential is an electric potential difference between a cornea and retina of the eye. The electric potential difference may thus be measured by the electrode arrangement, and the present apparatus may be configured to determine the sensor data.

[0022] The way in which the apparatus determines the sensor data may vary depending on implementation. In some examples, the apparatus may determine the sensor data by receiving the sensor data, for example from the electrode arrangement or interface associated with the electrode arrangement.

[0023] In some examples, the sensor data comprises or is indicative of a time-varying potential difference. This time varying potential difference is thus indicative of the change in the standing potential of the eye caused by OMT, and hence is indicative of the OMT in the eye. As a result, the sensor data can be efficiently determined using the electrode arrangement and the sensor data can be manipulated using one or more signal processing techniques (discussed further below). In some examples, the brain consciousness state comprises a: sleep state; brain death state; and / or sedation state of the brain. Hence, the present technique can be used to determine a variety of different brain consciousness states.

[0024] The present inventor has realised that the sensor data output from the electrode arrangement and indicative of the OMT in the eye varies depending on the brain consciousness state of the patient. Hence, by determining the sensor data (which is indicative of the underlying OMT), different brain consciousness states can be identified and differentiated, as each brain consciousness state may be associated with specific sensor data characteristics that can be used to differentiate between the different states.

[0025] In some examples, the sedation state of the brain comprises a plurality of sedation states, each indicative of a different level of sedation of the patient. Again, the present inventor has identified that different levels of sedation result in different OMT characteristics, which can in turn be differentiated and identified based on the sensor data associated with the electrode arrangement.

[0026] Hence, the present techniques can be used to determine a sedation state (and in some examples a sedation level) of a patient. This can be done in an automated, accurate, and objective manner. This can be useful because existing approaches to determining a sedation state / level of a patient can be manual, inefficient, and subjective. For example, in some cases, a healthcare practitioner may evaluate a patient’s sedation state using a scoring system based on interactions with the patient. This subjective, human-in-the-loop process can lead to unintended under or over sedation. Under sedation can result in patient discomfort, and over sedation can lead to short and long-term side effects including low blood pressure, loss of muscle mass, reduced immunity, increased risk of delirium, and increased length of stay (see Y. Shehabi et al., ‘Sedation depth and long-term mortality in mechanically ventilated critically ill adults: a prospective longitudinal multicentre cohort study’, Intensive Care Med., vol. 39, no. 5, pp. 910-918, May 2013, doi: 10.1007 / s00134-013-2830-2; M. M. Treggiari, ‘Randomized trial of light versus deep sedation on mental health after critical illness’, Ph.D., University of Washington, United States - Washington, 2007. Accessed: Apr. 18, 2023; and R. J. Stephens et al., ‘Practice patterns and outcomes associated with early sedation depth in mechanically ventilated patients: a systematic review and meta-analysis’, Crit. Care Med., vol. 46, no. 3, pp. 471-479, Mar. 2018, doi: 10.1097 / CCM.0000000000002885, which are all hereby incorporated by reference).

[0027] Hence, by providing a technique for automatically determining a sedation state based on objective sensor data, adverse effects associated with under and over sedation can be avoided, and the efficiency of the sedation state determination can be increased.

[0028] In some examples, the apparatus is configured to output an indication of the brain consciousness state. Thus, a user of the apparatus may be informed of the brain consciousness state. The user may then take action in response to the indication. This can be useful when the user is a healthcare practitioner, and action may then be taken in response to the indication, such as a change in sedative supplied to the user.

[0029] In other examples, the indication informs a user of the information relating to the patient. In some cases, the apparatus comprises an output interface to output the indication, such as a screen to output a graphical indication, and / or a speaker to output an audio indication. In other examples, the apparatus outputs an indication to a further device, for example for display by the further device. The present technique is not particularly limited in this respect.

[0030] In some examples, the electrode arrangement comprises a plurality of electrodes arranged in use on a skin-surface of the patient proximal to the eye to acquire the sensor data. Hence, the electrode arrangement can non-invasively determine the sensor data indicative of the OMT of the eye. This increases patient comfort.

[0031] In some examples, the electrode arrangement comprises one or more of: one or more pairs of bipolar electrodes; and a ground reference electrode. The electrode arrangement may comprise one or more unipolar electrodes, or advantageously one or more bipolar electrodes. This can reduce the noise in the sensor data.

[0032] In some examples, the sensor data comprises sensor data associated with multiple measurement channels of the electrode arrangement. This increases the signal to noise ratio of the sensor data and can increase the accuracy of the determination. Each channel may be associated with an electrode pair positioned around the eye, and so an improved measurement of the standing potential can be made using measurements from different electrode pairs.

[0033] In some examples, determining the brain consciousness state is based on a frequency of the sensor data.

[0034] In some examples, to determine the brain consciousness state, the apparatus is configured to analyse the sensor data using one or more signal processing techniques. As a result, the brain consciousness state can be efficiently and accurately determined from the sensor data. The signal processing technique is not particularly limited and may vary depending on implementation. By using one or more signal processing techniques, the brain consciousness state determination can be further improved to increase the likelihood of an accurate brain consciousness state determination.

[0035] Indeed, the signal processing technique may comprise one or more of wavelet decomposition, Fourier analysis, a machine learning technique (for example a machine learning classifier trained using training data that includes sensor data labelled with a brain consciousness state), Hilbert- Huang transforms, empirical mode decomposition, Kalman filtering, adaptive autoregression, and iterative filtering. Hence, various different signal processing techniques may be used to determine the brain consciousness state from the sensor data, and the present technique is not particularly limited in this respect.

[0036] In some examples, to analyse the sensor data using one or more signal processing techniques, the apparatus is configured to determine from the sensor data frequency information using the one or more signal processing techniques (such as frequency spectrum information). In some examples, the apparatus is configured to transform the sensor data by applying a mathematical function to the sensor data. Again, the mathematical function is not particularly limited and may vary depending on implementation. This supports an efficient analysis of the sensor data, and an accurate determination of the brain consciousness state.

[0037] In some examples, to analyse the sensor data, the apparatus is configured to de-noise the sensor data using one or more machine learning de-noising techniques. Machine learning de-noising techniques known in the art may be used.

[0038] In some examples, to analyse the sensor data, the apparatus is configured to apply one or more filters to the sensor data. This can increase the signal to noise ratio of the sensor data and thus increases the accuracy of the brain consciousness state determination.

[0039] In some examples, the one or more filters remove frequencies in the sensor data below 30 Hz and above 150 Hz. By removing frequencies less than 30 Hz, signals associated with EEG can be removed from the sensor data. This component of the sensor data may be initially present as the EOG electrodes surrounding the eye are in close proximity to the brain and so some EEG signal will be detected with the EOG electrodes. Also, signal associated with EMG (i.e. muscles) can also be removed as these are typically present around 15 Hz. Removing this component improves the signal to noise ratio. By removing frequencies greater than 150 Hz, non-biological signals can be removed from the sensor data. A notch filter may also be used to remove frequencies around 50 Hz, and its harmonics at 100 Hz, 150 Hz, etc., as these signals are typically associated with mains power hum.

[0040] In some examples, the one or more filters comprise a Butterworth filter, a band-pass filter and / or a notch filter. Hence, a variety of different filters may be used (and may be used in combination) to clean the sensor data.

[0041] The above filtering may be applied before other signal processing techniques. As a result, the sensor data contains less noise associated with unwanted background contributions (such as EEG, EMG, mains power, non-biological sources).

[0042] In some examples, the one or more signal processing techniques comprise one or more of a: Fourier transform, wavelet decomposition, and machine learning technique. Hence, a variety of different signal processing techniques may be used (and may be used in combination) depending on implementation. This provides a flexible approach to brain consciousness state determination.

[0043] The signal processing technique(s) may be selected based on use-case. For example, where a large amount of compute power is available, a machine learning technique may be used to analyse the sensor data. In other examples, a fast Fourier transform (based on a relatively small time window) may be used.

[0044] In some examples, the one or more signal processing techniques comprise wavelet decomposition based on a wavelet. Wavelet decomposition may provide a suitable balance of efficiency and accuracy, and thus may be well-suited to a real-time implementation (where the sensor data is analysed at least as fast as it is determined). Further, the present inventor has identified that an OMT signal may be non-stationary (where the signal itself changes in time and amplitude) and so wavelet decomposition may be well-suited to analysing the sensor data. The wavelet may be one or more of a: morse; morlet; Gaussian; and Mexican hat wavelet. The exact form of the wavelet may vary depending on implementation.

[0045] In some examples, to analyse the sensor data, the apparatus is configured to determine power spectral density information for a given time window based on the sensor data. In some examples, this is performed after wavelet decomposition (or fast Fourier transform), and hence may be performed based on the generated frequency spectrum information.

[0046] In other examples, to analyse the sensor data, the apparatus is configured to determine a frequency of the sensor data itself. This can then be used to determine the brain consciousness state. For example, based on previous investigation, the frequency of the sensor data can be characterised so that the brain consciousness state can be determined of unseen sensor data. In other examples, to analyse the sensor data, the apparatus is configured to determine a power- weighted frequency of the sensor data. In other examples, to analyse the sensor data, the apparatus is configured to determine a unitless measure of brain consciousness state, for example on a scale of 0 (no activity) to 100 (normal brain activity).

[0047] This power spectral density information can be used to determine the brain consciousness state. In some examples, to determine the brain consciousness state, the apparatus is configured to compare the power spectral density information to predetermined brain consciousness state definition information. Hence, various brain consciousness states (and in examples different levels of sedation) can be efficiently determined. The predetermined brain consciousness state definition information may be information determined based on previous experimental characterisation. Viewed from a second aspect, there is provided a system for determining a brain consciousness state of a patient, the system comprising: the apparatus described herein; and the electrode arrangement, wherein the apparatus is arranged in use to determine the sensor data from the electrode arrangement.

[0048] Hence, in some examples a system is provided that comprises the apparatus (that determines the brain consciousness state) and the electrode arrangement. In this case, the electrode arrangement acquires the sensor data, and the apparatus if configured to determine the sensor data from the electrode arrangement. It will be appreciated, that this arrangement may vary depending on implementation. For example, in some examples, the apparatus itself comprises the electrode arrangement (the electrode arrangement being integrated). In other cases, the apparatus is configured to be connected to the electrode arrangement, and the electrode arrangement is provided separately from the apparatus as a separate arrangement, which may allow for separate disposal of the electrode arrangement from the apparatus.

[0049] Viewed from a third aspect, there is provided a computer-implemented method for determining a brain consciousness state of a patient, the method comprising: determining sensor data associated with an electrode arrangement, the sensor data indicative of a change in a corneo- retinal standing potential of an eye of the patient caused by ocular microtremor; and determining, based on the sensor data, the brain consciousness state of the patient.

[0050] Viewed from a fourth aspect, there is provided a computer-readable storage medium comprising instructions which, when executed by one or more processors, cause the one or more processors to carry out the method described herein.

[0051] Viewed from a fifth aspect, there is provided a computer program comprising instructions which, when the computer program is executed by one or more processors, cause the one or more processors to carry out the method described herein.

[0052] Viewed from a sixth aspect, there is provided a method for determining a brain consciousness state of a patient, the method comprising: arranging an electrode arrangement on a skin-surface proximal to an eye of the patient to acquire sensor data, the sensor data indicative of a change in a corneo-retinal standing potential of the eye caused by ocular microtremor; and determining, based on the sensor data, the brain consciousness state of the patient.

[0053] Viewed from a seventh aspect, there is provided a method for controlling sedation of a patient, the method comprising: arranging an electrode arrangement on a skin-surface proximal to an eye of the patient to acquire sensor data, the sensor data indicative of a change in a corneo-retinal standing potential of the eye caused by ocular microtremor; determining, based on the sensor data, a sedation state of the patient; and controlling, based on the sedation state of the patient, an amount of sedation supplied to the patient. In this way, the administering of anaesthesia can be improved, increasing the likelihood of an improved patient outcome.

[0054] Viewed from an eighth aspect, there is provided a method of operating the apparatus described herein, the method comprising: determining, with the apparatus, sensor data associated with an electrode arrangement, the sensor data indicative of a change in a corneo-retinal standing potential of an eye of the patient caused by ocular microtremor; and determining, with the apparatus, based on the sensor data, a brain consciousness state of the patient.

[0055] Viewed from a ninth aspect, there is provided a computer-implemented method for obtaining information from a patient, the method comprising: determining sensor data associated with an electrode arrangement, the sensor data indicative of a change in a corneo-retinal standing potential of an eye of the patient caused by ocular microtremor; and determining, based on the sensor data, a sedation state of the patient.

[0056] Viewed from a tenth aspect, there is provided a use of the apparatus described herein to determine a sedation state of a patient.

[0057] Viewed from an eleventh aspect, there is provided a use of the apparatus described herein to obtain information from the patient. In some examples, the information obtained from the patient comprises the sensor data.

[0058] Viewed from a twelfth aspect, there is provided a use of the apparatus describe herein to diagnose a patient.

[0059] Other aspects will also become apparent upon review of the present disclosure, in particular upon review of the Brief Description of the Drawings, Detailed Description and Claims sections.

[0060] BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Examples of the disclosure will now be described, by way of example only, with reference to the accompanying drawings in which:

[0062] Figure 1 shows an example standing potential across an eye;

[0063] Figure 2 shows an example electrode arrangement according to the present techniques;

[0064] Figure 3 shows example steps for determining a brain consciousness state according to the present techniques;

[0065] Figure 4 shows example steps for determining a brain consciousness state according to the present techniques; Figure 5 shows example steps for determining a sedation state of a patient using wavelet decomposition;

[0066] Figure 6 shows a relationship between a power-weighted frequency and time;

[0067] Figure 7 shows example steps for determining a sedation state of a patient;

[0068] Figure 8 shows a relationship between mean power spectral density and time; and

[0069] Figure 9 shows an example electronic device that may implement the present techniques.

[0070] While the disclosure is susceptible to various modifications and alternative forms, specific example approaches are shown by way of example in the drawings and are herein described in detail. It should be understood however that the drawings and detailed description attached hereto are not intended to limit the disclosure to the particular form disclosed but rather the disclosure is to cover all modifications, equivalents and alternatives falling within the spirit and scope of the claimed invention.

[0071] It will be recognised that the features of the above-described examples of the disclosure can conveniently and interchangeably be used in any suitable combination.

[0072] DETAILED DESCRIPTION

[0073] Although illustrative teachings of the disclosure have been described in detail herein with reference to the accompanying drawings, it is to be understood that the invention is not limited to those precise teachings, and that various changes and modifications can be effected therein by one skilled in the art without departing from the scope and spirit of the invention as defined by the appended claims.

[0074] Figure 1 illustrates a corneo-retinal standing potential of an eye 2. In particular, as explained herein, the eye 2 can approximate a dipole, due to a negative charge located at a cornea 4 of the eye 2 and a relatively positively charge located at a retina 6 of the eye 2 (as labelled in figure 1 with -ve and +ve). This difference of electrical potential between the anterior (i.e. the retina 6) and posterior (i.e. the cornea 4) is referred to herein as the corneo-retinal standing potential.

[0075] When the eye moves, the orientation and movement of the dipole approximated by the eye 2 (i.e. the corneo-retinal standing potential) relative to the skull changes. This change in the dipole and the corneo-retinal standing potential can be measured to determine information indicative of the movement of the eye 2.

[0076] Figure 2 shows an example electrooculography electrode arrangement for measuring changes in the corneo-retinal standing potential of one or more eyes 2 of a patient 10. As shown in figure 2, an electrode arrangement 8 is positioned on a skin surface of the patient 10. The electrode arrangement 8 includes a plurality of electrodes, and in this example, includes two pairs of electrodes and a ground reference electrode. The electrodes of the electrode arrangement 8 are positioned around one or more eyes 2 of the patient 10 and are configured to acquire sensor data indicative of a change in the corneo-retinal standing potential of the eye 2 of the patient 10 caused by movement of the eye 2. The sensor data can comprise or be indicative of a time-varying potential difference.

[0077] The electrode arrangement 8 is referred to as an electrooculography electrode arrangement because it is configured to measure the corneo-retinal standing potential of the eye.

[0078] In the example of figure 2, a first electrode pair of the electrode arrangement 8 is configured to acquire sensor data indicative of horizontal movement of the eye 2, and a second electrode pair of the electrode arrangement 8 is configured to acquire sensor data indicative of vertical movement of the eye 2. Thus, this example electrode arrangement comprises multiple measurement channels, a first for measuring horizontal eye movement and a second for measuring vertical eye movement. Each measurement channel is thus associated with a different electrode pair.

[0079] It will be appreciated that the number, arrangement and positions of electrodes may be varied and the present disclosure is not particularly limited in this respect. For example, in some implementations a single electrode pair may be used and in others a plurality of electrode pairs may be used. The electrodes may be arranged in a different configuration from that shown in figure 2, where figure 2 merely shows one possible configuration. The electrodes may be unipolar or may be bipolar depending on implementation, and in some cases a single measurement channel may be used.

[0080] As discussed above, movement of the eye causes a change in the corneo-retinal standing potential, which in turn can be measured using an electrode arrangement.

[0081] However, an electrooculography electrode arrangement has not previously been used to measure ocular microtremor. The present inventor has realised that electrooculography can be used to measure ocular microtremor, and has further realised that the acquired data can be used to determine a brain consciousness state of the patient in a more accurate, efficient, and non- invasive manner than existing techniques to determine the brain consciousness state.

[0082] Figure 2 also shows an apparatus 12 for determining a brain consciousness state of the patient 10. OMT experienced by a patient can be indicative of the brain consciousness state of the patient. For example, a patient that is asleep, sedated, or brain dead will experience different OMT characteristics. Further, different levels of sedation of the patient are associated with different OMT characteristics. By determining sensor data associated with the electrode arrangement 8, the apparatus 12 can then determine the brain consciousness state of the patient 10 based on the sensor data.

[0083] The apparatus 12 can determine the sensor data associated with the electrode arrangement 8 in various ways. For example, the apparatus 12 may receive as an input the sensor data from the electrode arrangement 8 as shown in figure 2. In other examples, the sensor data may be received from one or more intermediate devices (such as a computer, server, etc.). The input of sensor data to the apparatus 12 may be directly wired from the electrode arrangement 8 as shown in figure 2 by the lines, or may be indirect. For example, over one or more communication networks or using one or more communication protocols (i.e. over Bluetooth, Wi-Fi, a cellular connection, etc.). The sensor data may be acquired by the electrode arrangement 8 and determined by the apparatus 12 in real-time, or the sensor data may be acquired and stored for offline determination by the apparatus 12. The present techniques are not particularly limited in these aspects.

[0084] As shown in figure 2, apparatus 12 optionally includes an output 12a for outputting the determined brain consciousness state. This may for example be a graphical output, and the apparatus 12 may comprise a screen for displaying the graphical output indicative of the determined brain consciousness state. In other examples, the output 12a is an audio output or the like. Alternatively, or additionally, the apparatus 12 may provide the determined brain consciousness state to a further device, for example for output.

[0085] It will be appreciated that figure 2 shows an example apparatus 12, and in other examples the apparatus 12 may be distributed across a plurality of devices (such as electronic devices, servers, and / or computers). For example, the sensor data may be received by a remote server and the remote server may determine the brain consciousness state from the sensor data. In this example, additional compute power of the server can be used. This may be useful in implementations that use machine learning to determine the brain consciousness state.

[0086] Figure 3 shows a method 300 for determining a brain consciousness state of a patient. Method 300 includes steps 301 , 302, 303, and 304. Steps 303 and 304 may be performed by apparatus 12 of figure 2.

[0087] At step 301 , a patient (such as patient 10 of figure 2) experiences a given brain consciousness state. The brain consciousness state may be one of a sleep state, a sedation state, or a brain death state, for example. In some cases, the brain consciousness state may be one a plurality of different sedation states, each associated with a different level of sedation of the patient. Hence, the present method may be used to determine a level of sedation of a patient, for example while anaesthesia is being supplied to the patient. At step 302, the patient experiences OMT which is indicative of the given brain consciousness state. As discussed herein, different brain consciousness states (and levels of sedation) result in different OMT characteristics. By measuring the OMT with an electrode arrangement, the brain consciousness state can be determined. It will be appreciated that steps 301 and 302 may occur at the same time.

[0088] At step 303, sensor data associated with the electrode arrangement is determined, the sensor data indicative of a change in a corneo-retinal standing potential of an eye of the patient caused by ocular microtremor. The electrode arrangement of step 303 may correspond to electrode arrangement 8 of figure 2.

[0089] Step 304 includes determining, based on the sensor data, the brain consciousness state of the patient. The brain consciousness state can be determined based on analysing the sensor data. For example, the apparatus 12 may analyse the sensor data using one or more signal processing techniques to determine the brain consciousness state. The signal processing techniques used to analyse the sensor data is not particularly limited. Analysing the sensor data using the one or more signal processing techniques to determine the brain consciousness state may be based on generating frequency information from the sensor data. The one or more signal processing techniques may comprise one or more signal decomposition techniques. The one or more signal decomposition techniques may comprise Fourier decomposition, wavelet decomposition, Hilbert- Huang transforms, empirical mode decomposition, Kalman filtering, adaptive autoregression, and iterative filtering.

[0090] In another example, a machine learning technique may be used to analyse the sensor data. For example, the sensor data may be input to a neural network trained to output a brain consciousness state based on input sensor data. The neural network may be trained based on training data comprising sensor data labelled with the associated brain consciousness state. The neural network may be a classifier neural network.

[0091] Figure 4 shows another method 400 for determining a brain consciousness state of a patient. Method 400 follows from step 303 of figure 3 and thus may correspond to step 304 of figure 3. Method 400 may be performed by apparatus 12 of figure 2.

[0092] At step 401 , the sensor data is filtered. While it will be appreciated that this step may be optionally performed, it can be useful to reduce the noise in the sensor data and increase the signal-to-noise ratio of the sensor data to increase the accuracy and efficiency of the brain consciousness state determination. This may include applying one or more filters to the sensor data. This can be particularly useful for the present technique that uses EOG to measure OMT, which can be sensitive to noise. The filtering may remove frequencies in the sensor data below 30 Hz. Such frequencies may be associated with EEG signals and so this can be removed to provide a clearer EOG signal. As described herein, as a result of the placement of the electrodes, some EEG signal may necessarily be present in the sensor data (despite the electrodes being arranged around the eye). Hence, filtering to remove signals of about 30 Hz or lower can remove unwanted EEG signals that can increase the noise in the EOG signal and can reduce the accuracy of the brain consciousness state determination. This frequency filtering can also remove signals associated with EMG, which typically occur around 15 Hz.

[0093] The filtering can additionally or alternatively remove frequencies in the sensor data greater than 150 Hz. Such frequencies may be associated with non-biological activity. Further, a notch filter may be used to remove frequencies typically associated with mains power hum (such as 50 Hz and harmonics thereof).

[0094] The types of filter that may be applied are not particularly limited and may be selected and varied based on implementation. In some cases, the one or more filters that are applied to the sensor data include one or more of a Butterworth filter, a band-pass filter and a notch filter.

[0095] In some examples, a neural network / machine learning de-noiser may be used to de-noise the sensor data. Hence, in some examples, the apparatus may be configured to de-noise the sensor data using a machine learning denoising technique. This may be performed in addition, or instead of, the filtering at step 401. When performed in addition to the filtering at step 401 , the machine learning de-noiser may be used after the sensor data is filtered at step 401. Machine learning denoising techniques known in the art may be used.

[0096] At step 402, the sensor data is analysed using one or more signal processing techniques. The signal processing technique (or combination thereof) used to analyse the sensor data is not particularly limited and may vary depending on implementation. Step 402 may comprise generating frequency spectrum information using the one or more signal processing techniques.

[0097] In some examples, the signal processing technique is a signal processing technique configured to determine frequency spectrum information from the sensor data. For example, the signal processing technique may include a Fourier transform or wavelet transform.

[0098] As an example, Fourier analysis may be used to analyse the sensor data. For example, a fast Fourier transform may be applied to a portion of sensor data corresponding to a given time period to generate frequency spectrum information from the sensor data for that given time period. As the OMT signal (and thus the sensor data) is non-stationary in time, a relatively short time period may be chosen (such as 1 second) so as to approximate stationary behaviour. By doing this, the time series sensor data can be mapped to a frequency spectrum. As explained further below, power spectral density information may then be determined based on the frequency spectrum information, and compared to known values to determine the brain consciousness state.

[0099] As a further example, wavelet decomposition may be used to analyse the sensor data. For example, wavelet decomposition using an analytical wavelet (such as morse, morlet, Gaussian, Mexican hat, etc.) can be applied to a portion of sensor data corresponding to a given time period. The time period could be longer than for a Fourier example, such as 10s, because wavelet decomposition can be used with non-stationary signals Hence, in some examples, wavelet decomposition may provide a more accurate representation of the signal in the frequency spectrum. The wavelet decomposition analysis generates frequency spectrum information from the sensor data. As explained further below, power spectral density information may then be determined based on the frequency spectrum information, and compared to known values to determine the brain consciousness state.

[0100] At step 403, power spectral density information for a given time window is determined based on the analysed sensor data (i.e. the sensor data from step 402). This step may therefore use the frequency spectrum information generated at step 402. While in this example a power spectral density is used, it will be appreciated that other information may be used instead, such as a power- weighted frequency.

[0101] At step 404, this power spectral density information is compared to predetermined brain consciousness state definition information to determine the brain consciousness state.

[0102] Various brain consciousness states (levels of sedation, sleep, brain death, etc.) can be characterised to generate the predetermined brain consciousness state definition information, so that when a patient experiences an unknown brain consciousness state, by performing the present techniques, the unknown brain consciousness state can be identified.

[0103] For example, in ICU implementations, the Richmond Agitation Sedation Score (RASS) may be used, where higher readouts from the present process correspond to higher wakefullness measured by RASS. Similarly, in Anaesthesia implementations, a higher readout corresponds with lower sedation levels as measured by the infusion pumps (infusing propofol for example which is an anaesthetic).

[0104] An example of analysing sensor data using wavelet decomposition to determine a brain consciousness state will now be described with reference to figure 5. In this example, a patient is being sedated and an EOG electrode arrangement is used to determine the sensor data during according to the present techniques during the sedation.

[0105] Figure 5 shows a representation of sensor data 14 as a time-varying potential difference. In particular, the x-axis shows time (in seconds) and the y-axis shows potential difference (in mV). The sensor data 14 has been determined from an EOG electrode arrangement arranged on a face of a patient and is indicative of the OMT of an eye of the patient. As the eye experiences OMT, the corneo-retinal standing potential changes and this change can be detected using the EOG electrode arrangement. One or more filters, such as those described above, may have been applied to the sensor data to generate the sensor data 14.

[0106] Once the sensor data 14 has been determined (and optionally filtered), wavelet decomposition is performed on the sensor data 14 using a wavelet (such as a morse wavelet) to generate frequency spectrum information 16.

[0107] Once the frequency spectrum information 16 has been generated, the power spectral density information can be determined for a given time window. The value of the power spectral density for a series of frames (each corresponding to a given time window) is shown by the relationship of power spectral density information against time 18. This value, and how it changes over time, can be used to determine the brain consciousness state. For example, the present inventor has characterised various brain consciousness states in terms of how the power spectral density changes over time such that unseen sensor data can be used to identify an unknown brain consciousness state.

[0108] For example, the present inventor has realised that this technique can be used to specifically determine when a patient has entered a given state of sedation. In particular, point 20 in figure 5 shows a local peak in the value of the power spectral density information. Point 20 corresponds to a point in time when the patient loses their eye-lash reflex, a known clinically relevant end point for loss of consciousness in anaesthesia. This occurs when the patient loses consciousness, and so can also be used to determine the brain consciousness state of the patient.

[0109] Figure 6 shows a relationship between a power-weighted frequency of OMT and time as anaesthesia is induced in a patient (i.e. this is an alternative to the final stage of figure 5). In this example, the signal processing technique used to analyse the sensor data from an EOG electrode arrangement comprises a Hilbert-Huang transform, rather than wavelet decomposition.

[0110] As shown in figure 6, the frequency of the OMT increases significantly just before approximately the 200 mark on the x-axis. This corresponds to a point in time when the patient loses their "eyelash" reflex indicating appropriate anaesthesia, whereupon their airway is instrumented (hence the slight increase in the frequency of the OMT). This is then followed by a period of inactivity to ensure stability and a slow decrease in the frequency of the OMT as further sedation is being infused.

[0111] Figure 7 shows an example method 700 for determining a level of sedation of a patient. As explained above, existing approaches to determining a patient sedation level are subjective, manual, and prone to human error. By using EOG to measure OMT as a way to determine the patient sedation level, the patient sedation level can be determined more accurately and efficiently.

[0112] At step 701 , the patient experiences a sedation state indicative of a given level of sedation. For example, a healthcare professional (i.e. anaesthetist) may supply anaesthesia to a patient. The patient then starts to experience a sedation state. At step 702, the patient experiences OMT which is indicative of the sedation state. As discussed herein, the characteristics of OMT that may be experienced by a patient are effected and related to the sedation state of the patient (i.e. how much anaesthesia the patient has received).

[0113] At step 703, sensor data associated with the electrode arrangement is determined, the sensor data being indicative of a change in a corneo-retinal standing potential of an eye of the patient caused by the OMT. Hence, EOG can be used to determine the characteristics (e.g. amplitude, frequency) of the OMT, the OMT itself being representative of the level of sedation of the patient.

[0114] At step 704, the sedation state of the patient is determined based on the sensor data. This step may include using one or more signal processing techniques to determine the brain consciousness state as described herein (such as with respect to figures 2 to 5).

[0115] In this way, the sedation state (i.e. level of sedation) of the patient can be determined accurately and efficiently in a non-invasive manner.

[0116] Figure 8 shows a relationship between mean power spectral density (PSD) and time, for a patient before and after anaesthesia. In this example, wavelet decomposition is used to determine the power spectral density, but it will be appreciated that other techniques may be used instead. In this example, a two channel (Ch) electrode arrangement is used, one horizontal and one vertical as discussed previously.

[0117] As shown in figure 8, at the point of induction of anaesthesia, the mean PSD values start to reduce, with a reduction in the gradient of the trend line for both channel 1 and channel 2. Hence, the PSD values, which can be determined from the sensor data from an EOG arrangement, can be used to derive a patient’s brain consciousness state, in this case their sedation level. In this example, a PSD value of between -50 and -80 may indicate that the patient is consciousness, and a PSD value of below -80 may indicate that the patient is unconsciousness (noting that the y-axis is a logarithmic scale).

[0118] Figure 9 schematically illustrates an example of an electronic device 900 (such as a computing device) which can be used to implement teachings described above, including method 300, 400, 500, and 700 described in relation to figures 3 to 7. Electronic device 900 may correspond to apparatus 12 of figure 1. The electronic device 900 has processing circuitry 910 for performing data processing in response to program instructions and data storage 920 for storing data and instructions to be processed by the processing circuitry 910. In some examples, the processing circuitry 910 includes one or more caches for caching recent data or instructions. The data storage 920 may have a database 930 which can, for example, store predetermined brain consciousness state definition information used for the comparison. The device further includes a communication interface 940 which can be used, for example, to obtain / receive the sensor data and information relating to the predetermined brain consciousness state definition information and the signal processing techniques(s) (such as the wavelet parameterisation). It will be appreciated that Figure 9 is merely an example of possible hardware that may be provided in the device and other components may also be provided. The device 900 may additionally or alternatively be provided with one or more user input / output device(s) 950 to receive the sensor data (directly or indirectly) from the electrode arrangement or to output information (e.g. the determined brain consciousness state information). This output device 950 may output the information on a screen of the device 700 or provide the information to a user device or other computing device for example.

[0119] The methods discussed above may be performed under control of a computer program executing on a device. Hence a computer program may comprise instructions for controlling a device to perform any of the methods discussed above. The program can be encoded in a computer- readable medium. A computer-readable medium may include non-transitory type media such as physical storage media including storage discs and solid state devices. A computer-readable medium may also or alternatively include transient media such as carrier signals and transmission media. A computer-readable storage medium is defined herein as a non-transitory memory device. A memory device includes memory space within a single physical storage device or memory space spread across multiple physical storage devices.

[0120] In the present application, the words “configured to...” are used to mean that an element of an apparatus has a configuration able to carry out the defined operation. In this context, a “configuration” means an arrangement or manner of interconnection of hardware or software. For example, the apparatus may have dedicated hardware which provides the defined operation, or a processor or other processing device may be programmed to perform the function. “Configured to” does not imply that the apparatus element needs to be changed in any way in order to provide the defined operation.

[0121] Although illustrative teachings of the disclosure have been described in detail herein with reference to the accompanying drawings, it is to be understood that the invention is not limited to those precise teachings, and that various changes and modifications can be effected therein by one skilled in the art without departing from the scope and spirit of the invention as defined by the appended claims. Examples are set out in the following numbered clauses.

[0122] 1. An apparatus for determining a brain consciousness state of a patient, the apparatus comprising one or more processors, the apparatus configured to: determine sensor data associated with an electrode arrangement, the sensor data indicative of a change in a corneo-retinal standing potential of an eye of the patient caused by ocular microtremor; and determine, based on the sensor data, the brain consciousness state of the patient.

[0123] 2. The apparatus of clause 1 , wherein the electrode arrangement is an electrooculography electrode arrangement.

[0124] 3. The apparatus of any preceding clause, wherein the corneo-retinal standing potential is an electric potential difference between a cornea and retina of the eye.

[0125] 4. The apparatus of any preceding clause, wherein the sensor data comprises a timevarying potential difference.

[0126] 5. The apparatus of any preceding clause, wherein the brain consciousness state comprises a: sleep state; brain death state; and / or sedation state of the brain.

[0127] 6. The apparatus of clause 5, wherein the sedation state of the brain comprises a plurality of sedation states, each indicative of a different level of sedation of the patient.

[0128] 7. The apparatus of any preceding clause, wherein the apparatus is configured to output an indication of the brain consciousness state.

[0129] 8. The apparatus of any preceding clause, wherein the electrode arrangement comprises a plurality of electrodes arranged in use on a skin-surface of the patient proximal to the eye to acquire the sensor data.

[0130] 9. The apparatus of any preceding clause, wherein the electrode arrangement comprises one or more of: one or more pairs of bipolar electrodes; and a ground reference electrode.

[0131] 10. The apparatus of any preceding clause, wherein the sensor data comprises sensor data associated with multiple measurement channels of the electrode arrangement.

[0132] 11. The apparatus of any preceding clause, wherein to determine the brain consciousness state, the apparatus is configured to analyse the sensor data using one or more signal processing techniques.

[0133] 12. The apparatus of clause 11, wherein to analyse the sensor data using one or more signal processing techniques, the apparatus is configured to transform the sensor data by applying a mathematical function to the sensor data.

[0134] 13. The apparatus of any of clauses 11 or 12, wherein to analyse the sensor data, the apparatus is configured to apply one or more filters to the sensor data.

[0135] 14. The apparatus of clause 13, wherein the one or more filters remove frequencies in the sensor data below 30 Hz and above 150 Hz. 15. The apparatus of any of clauses 13 or 14, wherein the one or more filters comprise a Butterworth filter, a band-pass filter and / or a notch filter.

[0136] 16. The apparatus of any of clauses 11 to 15, wherein the one or more signal processing techniques comprise one or more of a: Fourier transform, wavelet decomposition, and machine learning technique.

[0137] 17. The apparatus of any of clauses 11 to 16, wherein the one or more signal processing techniques comprise wavelet decomposition based on a wavelet.

[0138] 18. The apparatus of clause 17, wherein the wavelet is one or more of a: morse; morlet; Gaussian; and Mexican hat wavelet.

[0139] 19. The apparatus of any of clauses 11 to 18, wherein to analyse the sensor data, the apparatus is configured to determine power spectral density information for a given time window based on the sensor data.

[0140] 20. The apparatus of clause 19, wherein to determine the brain consciousness state, the apparatus is configured to compare the power spectral density information to predetermined brain consciousness state definition information.

[0141] 21. A system for determining a brain consciousness state of a patient, the system comprising: the apparatus of any of clauses 1 to 20; and the electrode arrangement, wherein the apparatus is arranged in use to determine the sensor data from the electrode arrangement.

[0142] 22. A computer-implemented method for determining a brain consciousness state of a patient, the method comprising: determining sensor data associated with an electrode arrangement, the sensor data indicative of a change in a corneo-retinal standing potential of an eye of the patient caused by ocular microtremor; and determining, based on the sensor data, the brain consciousness state of the patient.

[0143] 23. A computer-readable storage medium comprising instructions which, when executed by one or more processors, cause the one or more processors to carry out the method of clauses 22.

[0144] 24. A computer program comprising instructions which, when the computer program is executed by one or more processors, cause the one or more processors to carry out the method of clause 22.

[0145] 25. A method for determining a brain consciousness state of a patient, the method comprising: arranging an electrode arrangement on a skin-surface proximal to an eye of the patient to acquire sensor data, the sensor data indicative of a change in a corneo-retinal standing potential of the eye caused by ocular microtremor; and determining, based on the sensor data, the brain consciousness state of the patient.

[0146] 26. A method for controlling sedation of a patient, the method comprising: arranging an electrode arrangement on a skin-surface proximal to an eye of the patient to acquire sensor data, the sensor data indicative of a change in a corneo-retinal standing potential of the eye caused by ocular microtremor; determining, based on the sensor data, a sedation state of the patient; and controlling, based on the sedation state of the patient, an amount of sedation supplied to the patient.

[0147] 27. A method of operating the apparatus of any of clauses 1 to 20, the method comprising: determining, with the apparatus, sensor data associated with an electrode arrangement, the sensor data indicative of a change in a corneo-retinal standing potential of an eye of the patient caused by ocular microtremor; and determining, with the apparatus, based on the sensor data, a brain consciousness state of the patient.

[0148] 28. A computer-implemented method for obtaining information from a patient, the method comprising: determining sensor data associated with an electrode arrangement, the sensor data indicative of a change in a corneo-retinal standing potential of an eye of the patient caused by ocular microtremor; and determining, based on the sensor data, a sedation state of the patient.

[0149] 29. Use of the apparatus of any of clauses 1 to 20 to determine a sedation state of a patient.

[0150] 30. Use of the apparatus of any of clauses 1 to 20 to obtain information from the patient.

[0151] 31. Use of the apparatus of any of clauses 1 to 20 to diagnose a patient.

Claims

CLAIMS1. An apparatus for determining a brain consciousness state of a patient, the apparatus comprising one or more processors, the apparatus configured to: determine sensor data associated with an electrode arrangement, the sensor data indicative of a change in a corneo-retinal standing potential of an eye of the patient caused by ocular microtremor; and determine, based on the sensor data, the brain consciousness state of the patient.

2. The apparatus of claim 1 , wherein the electrode arrangement is an electrooculography electrode arrangement.

3. The apparatus of any preceding claim, wherein the sensor data comprises a time-varying potential difference.

4. The apparatus of any preceding claim, wherein the brain consciousness state comprises a: sleep state; brain death state; and / or sedation state of the brain.

5. The apparatus of claim 4, wherein the sedation state of the brain comprises a plurality of sedation states, each indicative of a different level of sedation of the patient.

6. The apparatus of any preceding claim, wherein the apparatus is configured to output an indication of the brain consciousness state.

7. The apparatus of any preceding claim, wherein the sensor data comprises sensor data associated with multiple measurement channels of the electrode arrangement.

8. The apparatus of any preceding claim, wherein to determine the brain consciousness state, the apparatus is configured to analyse the sensor data using one or more signal processing techniques.

9. The apparatus of claim 8, wherein to analyse the sensor data using one or more signal processing techniques, the apparatus is configured to determine from the sensor data frequency information using the one or more signal processing techniques and optionally wherein the apparatus is configured to transform the sensor data by applying a mathematical function to the sensor data.

10. The apparatus of any of claims 8 or 9, wherein to analyse the sensor data, the apparatus is configured to apply one or more filters to the sensor data.

11. The apparatus of claim 10, wherein the one or more filters remove frequencies in the sensor data below 30 Hz and above 150 Hz.

12. The apparatus of any of claims 10 or 11 , wherein the one or more filters comprise a Butterworth filter, a band-pass filter and / or a notch filter.

13. The apparatus of any of claims 8 to 12, wherein the one or more signal processing techniques comprise one or more of a: Fourier transform, wavelet decomposition, and machine learning technique.

14. The apparatus of any of claims 8 to 13, wherein the one or more signal processing techniques comprise wavelet decomposition based on a wavelet.

15. The apparatus of any of claims 8 to 14, wherein to analyse the sensor data, the apparatus is configured to determine power spectral density information for a given time window based on the sensor data.

16. The apparatus of claim 15, wherein to determine the brain consciousness state, the apparatus is configured to compare the power spectral density information to predetermined brain consciousness state definition information.

17. A system for determining a brain consciousness state of a patient, the system comprising: the apparatus of any of claims 1 to 16; and the electrode arrangement, wherein the apparatus is arranged in use to determine the sensor data from the electrode arrangement.

18. A computer-implemented method for determining a brain consciousness state of a patient, the method comprising: determining sensor data associated with an electrode arrangement, the sensor data indicative of a change in a corneo-retinal standing potential of an eye of the patient caused by ocular microtremor; and determining, based on the sensor data, the brain consciousness state of the patient.

19. A computer-readable storage medium comprising instructions which, when executed by one or more processors, cause the one or more processors to carry out the method of claim 18.

20. A computer program comprising instructions which, when the computer program is executed by one or more processors, cause the one or more processors to carry out the method of claim 18.

21. A method for controlling sedation of a patient, the method comprising: arranging an electrode arrangement on a skin-surface proximal to an eye of the patient to acquire sensor data, the sensor data indicative of a change in a corneo-retinal standing potential of the eye caused by ocular microtremor; determining, based on the sensor data, a sedation state of the patient; and controlling, based on the sedation state of the patient, an amount of sedation supplied to the patient.

22. A method of operating the apparatus of any of claims 1 to 16, the method comprising: determining, with the apparatus, sensor data associated with an electrode arrangement, the sensor data indicative of a change in a corneo-retinal standing potential of an eye of the patient caused by ocular microtremor; and determining, with the apparatus, based on the sensor data, a brain consciousness state of the patient.

23. A computer-implemented method for obtaining information from a patient, the method comprising: determining sensor data associated with an electrode arrangement, the sensor data indicative of a change in a corneo-retinal standing potential of an eye of the patient caused by ocular microtremor; and determining, based on the sensor data, a sedation state of the patient.

24. Use of the apparatus of any of claims 1 to 16 to determine a sedation state of a patient.

25. Use of the apparatus of any of claims 1 to 16 to obtain information from the patient.

Images