A method involving rearrangement of stimuli in an SSVEP-based brain-macine interface

Abstract

A method for performing an action based on a user's brain activity comprising the steps of: (a) displaying a plurality of labels; (b) displaying a plurality of stimuli wherein each stimulus correspond

Description

This invention relates to a method for performing an action triggered by brain signals and a system capable of interpreting brain signals in order to determine and perform a desired action. In particular the invention relates to a brain-machine interface (BMI) system which allows a person using the system to select an action they wish to perform by looking at the desired action from a number of options, the system carries out a method to determine which action is desired by the user and issues a command for that action to be performed. People with severely restricted mobility, such as sufferers of paralysis, lack the ability to control devices around them that require physical interactions. Further, sufferers of locked-in-syndrome lose control of all of their muscles except the ones that control their eye movements and are unable to communicate verbally. Therefore, they cannot take advantage of the recent influx of voice-controlled devices. This may also be an issue for other sufferers of severe paralysis, such as hemiplegia, who might have lost the ability to speak with sufficient clarity to control voice-controlled devices. Although sufferers of such conditions are typically restricted to a hospital bed, the ability to control the devices around them could provide them with a considerable improvement in both comfort and independence. For example, a motorized bed, a room’s air conditioning system and a television could each provide improved comfort or entertainment. The ability to control such devices without requiring another person’s assistance would significantly improve the independence of a person with a condition such as locked-in-syndrome. Brain-machine interfaces (BMIs) (sometimes referred to as brain-computer interfaces, neural-control interfaces, mind-machine interfaces or direct neural interfaces) give a person the ability to communicate with and control machines using brain signals alone rather than peripheral muscles. There is a wide variety of different types of BMI. Invasive BMIs require surgery to implant electrodes under the skull and into the brain. The main advantage of invasive BMIs is the provision of more accurate brain signal readings. However, disadvantages include sideeffects from the surgery, scar tissue forming which weakens the brain signals and the potential for the body to reject the electrodes. Partially invasive BMIs similarly require surgery to implant electrodes inside the skull, but in this case they rest outside the brain rather than within the brain. The risk of side-effects, such as scar-tissue forming in the brain, is lower than for fully invasive BMIs but the signals from the brain are not as strong. In order to avoid the need for surgery, and the potential side-effects that are associated with fully and partially invasive BMIs, the majority of published research in the field focuses on non-invasive BMIs which involves electrodes attached to the scalp. The main disadvantage of non-invasive BMIs is that the strength of signals from the brain is much lower, especially for high frequencies which are dampened by the skull. The most common types of non-invasive BMI are based on electroencephalography (EEG) - a method of recording the electrical activity of the brain. Known EEG-based BMIs comprise a plurality of electrodes positioned on a person’s scalp. In some cases the electrodes are “wet” and require electrolyte to be applied between the electrode and the skin to improve conduction of the signal between the skin and the electrode. In other cases the electrodes are “dry” and do not require the use of electrolyte. However, dry electrodes typically suffer from high impedance. In some cases the electrodes are “passive” electrodes which simply pass the EEG signals from the scalp to the associated BMI system. In other cases the electrodes are “active” electrodes which locally amplify and buffer the voltage level of EEG signals before transmitting the signal through any cabling to the BMI system. The low output impedance of an active electrode mitigates cable motion artefacts, thus enabling the use of high-impedance dry electrodes for greater user comfort. A subset of EEG-based BMIs use visual stimuli to excite the retina and generate steadystate visually evoked potentials (SSVEPs) which may be measured by electrodes positioned at the back of the head, close to the occipital region of the brain. The main advantage of this type of BMI is that the frequency of the phase reversal of the stimulus used can be clearly distinguished in the spectrum of an EEG signal which makes detection of SSVEP stimuli from the recorded EEG data relatively easy. Another advantage is that the method is based on eye movement which is one of the few abilities that people with locked in syndrome retain. However, an issue with SSVEP-based BMIs is that the visual stimuli required involves images flickering at different frequencies which can be irritating for a person using the BMI over a long period of time. There are also several non-EEG-based BMIs, one of which also relies on eye movement by recording electrooculography (EOG) signals. EOG-based BMIs require at least one electrode placed close to the eye in order to detect a steady electric potential field which originates from the eye. As the eye moves, so does the electric potential field. A corresponding change in voltage may be detected by the electrode. In particular EOGbased BMIs are a very robust technique for measuring eye gaze shifts and blinks. Some known BMIs require a period of training before they may be used to perform their true purpose. This training period may involve the completion of a number of trials in order to determine the best way to interpret a particular user’s brain signals. For example, known SSVEP-based BMIs may ask the user to focus on a number of different stimuli, one-by-one, while recording EEG data to establish which stimuli are most effective at inducing SSVEPs. This process can take several minutes and can cause the user to experience fatigue due to continuous retinal excitation and / or irritation due to the repetitive nature of the trials. BMIs may operate either synchronously or asynchronously. Synchronous operation is cue-based as it requires certain actions to be performed in distinct time periods, whereas asynchronous operation is paced by the user as actions may be performed at any point in time. According to a first aspect of the invention there is provided a method for performing an action based on a user’s brain activity comprising the steps of: (a) displaying a plurality of labels; (b) displaying a plurality of stimuli wherein each stimulus corresponds to a stimulus parameter and is associated with a different label; (c) receiving a brain signal while the user is exposed to the plurality of stimuli; (d) applying a steady state visually evoked potential (SSVEP) detection algorithm to the brain signal in order to detect an SSVEP and determine a stimulus parameter signal from the SSVEP; (e) determining which one of the plurality of stimuli corresponds to a stimulus parameter which is closest to the determined stimulus parameter signal and identifying the label associated with that stimulus as a selected label; (f) indicating the selected label to the user; (g) receiving an eye movement signal while the user is exposed to the indication of the selected label; and (h) determining whether the recorded eye movement signal corresponds to a predetermined cancel eye gesture, wherein: if a cancel eye gesture is determined, performing the further steps of: adding one tally to a counted number of consecutive cancel eye gestures and repeating steps (a) to (h); if a cancel eye gesture is not determined, performing the further step of: performing the action corresponding to the selected label, resetting the counted number of consecutive cancel eye gestures to zero and storing in memory a correct selection data set comprising the brain signal and the stimulus parameter of the stimulus associated with the selected label. By means of the invention, a user of a BMI system may view a display showing a plurality of labels wherein each label defines an action that the user might want to perform. Further, each of the plurality of labels may be associated with one of a plurality of stimuli wherein each stimulus has individual predetermined parameters, such as flicker frequency. If the user wishes to perform any of the actions defined by the plurality of labels, the user may look at the relevant label and the associated stimulus should excite the user’s retinas and induce an SSVEP to be generated in the user’s occipital region of the brain. A brain signal may be received by the BMI system and an SSVEP detection algorithm may detect an SSVEP and determine a stimulus parameter signal from the SSVEP that is representative of a stimulus parameter that is estimated to have caused the SSVEP. In other words, the SSVEP algorithm detects an SSVEP and estimates the features of the stimulus that caused the SSVEP to be generated. The BMI system may then determine which label the user was looking at based on an analysis of which of the plurality of stimuli corresponds to a stimulus parameter that is closest to the determined stimulus parameter signal. The determined label is shown to the user who may then decide whether the determination is correct. If the determined label is not the label that the user intended to choose then the user may perform a predetermined cancel eye gesture, such as a double blink or a wink, which allows the user to repeat the selection process. The number of cancel eye gestures determined with respect to consecutive brains signals received by the BMI system is counted. This means the system keeps track of how many consecutive times a user fails to select the label desired and allows the system to adapt and improve to avoid the user becoming frustrated. If the determined label is the label that the user intended to choose then the user is not required to perform any further action and after waiting a predetermined period of time the BMI system may issue a command for the action defined by the selected label to be performed. Therefore, by means of the invention, the user may use a BMI system to perform an action with only brain activity and eye movements. Meanwhile, the counted number of consecutive cancel eye gestures is reset to zero to acknowledge that the most recent label selection has not been cancelled. Also, a correct selection data set is stored in memory as correct data may be used to train one or more training-based SSVEP algorithms and thereby improve the performance of the BMI system. In some embodiments of the invention, the step of receiving a brain signal while the user is exposed to the plurality of stimuli may be performed continuously until the SSVEP detection algorithm detects an SSVEP and determines a stimulus parameter signal of the SSVEP. This means that the label selection process by the user is asynchronous - it is not confined to a predetermined time frame. As such, the brain signal may be considered as a continuous transfer of information from the user’s brain to the BMI system. Asynchronous operation allows the user to interact with the BMI at any time convenient to the user. This could enable the user to multi-task between using the BMI and performing other actions such as listening to a person speaking to them or watching a television, for example. The freedom offered by an asynchronous BMI system, such as a BMI system capable of performing the method provided by the invention, may be particularly advantageous for a user with severely restricted mobility as they may not have the ability to otherwise interact with devices in the local environment. In other embodiments of the invention, the step of receiving a brain signal while the user is exposed to the plurality of stimuli may be performed during a predetermined period of time. This means that the label selection process is synchronous - it is cue based and confined to predetermined timings. After the predetermined period of time elapses, the SSVEP detection algorithm assigns a label to the received brain signal, irrespective of whether the user was looking at the plurality of stimuli. Therefore, in order to avoid an unwanted action being performed, the user is required to interact with the BMI system within the predetermined time. Synchronous operation limits the freedom that the user has when interacting with the BMI system. However, a BMI system that operates synchronously is not required to differentiate between an idle state and an active state of the user. The task of the SSVEP detection algorithm may therefore be simplified and the accuracy of the SSVEP may be improved as a result. For example, it may be less likely that SSVEPs are missed by a synchronous system as it is not possible for the synchronous system to mistake an SSVEP for an idle state. The SSVEP detection algorithm may be one of many different types of SSVEP detection algorithm. For example, in a first embodiment of the invention, the SSVEP selection algorithm may be a training-free SSVEP detection algorithm that requires no training in order to function sufficiently for a user. However, the performance of such embodiments may have limited accuracy as a result of not being trainable. In a second embodiment of the invention, the SSVEP detection algorithm may be a training-based algorithm which requires predetermined training wherein a user of the BMI system is required to carry out a training session. In such embodiments the BMI system may store correct selection data sets generated during the training session to train the SSVEP detection algorithm and improve the performance of the SSVEP detection algorithm following the training session. In a third embodiment of the invention the SSVEP detection algorithm may initially function without requiring training (similarly to the first embodiment referred to above). This allows a user to instantly use the BMI system without needing to carry out a predetermined (and potentially time consuming) training session, albeit with possibly limited accuracy. However, as the user continues using the BMI system, correct selection data sets may be stored and once a sufficient number of data sets are stored the information may be used to train and improve the accuracy of the SSVEP detection algorithm in a similar way to the second embodiment. In a fourth embodiment of the invention the SSVEP detection algorithm may be preconfigured based on training data obtained from a trial set of users, not including data relating to the current user of the BMI system. This is known as subject independent training as, to use the BMI system, the user does not need to do any training him / herself. During the actual running of the BMI system, correct selection data sets associated with the current user may be obtained and stored to adapt the training data available and fine tune the performance of the SSVEP detection algorithm system to the current user. In other embodiments of the invention, a combination of one or more of the SSVEP detection algorithms described above may be applied in parallel and the predictions generated by each may be fused together in order to determine a stimulus parameter signal from the received brain signal. Accordingly, in embodiments of the invention, the step of applying an SSVEP detection algorithm may comprise: applying a training-free SSVEP detection algorithm configured to operate without training; applying a training-based SSVEP detection algorithm configured to operate according to training based on correct selection data sets stored in memory; or applying a plurality of training-free and / or training-based SSVEP detection algorithms and fusing all predictions to generate a single determined stimulus parameter. In such embodiments of the invention, the SSVEP algorithm or combination of SSVEP algorithms applied may be determined based on the quantity of correct selection data sets stored in memory. In other words, the SSVEP detection algorithm which is used may depend on how long a user has been using the BMI system, i.e., how many correct selection data sets are stored in memory. Initially, a training-based SSVEP detection algorithm may be applied as it will operate with no training, albeit with potentially limited accuracy. However, once a sufficient number of correct selection data sets have been stored in memory, one or more training based SSVEP detection algorithms may be applied. This allows the user to benefit from a system which may be used instantly and also becomes more accurate after continued use. In embodiments of the invention which may apply a training-based SSVEP detection algorithm, the method further comprises the step of using a correct selection data set to train a training-based SSVEP detection algorithm. In particular, the SSVEP detection algorithm may learn how to more accurately determine a stimulus parameter signal from an SSVEP. A more accurate determination of a stimulus parameter signal is a determination in which the determined stimulus parameter signal, for that particular user, is closer to the true stimulus parameter. If the determination of the stimulus parameter signal is more accurate, this means that the chance of the BMI system determining that an incorrect stimulus parameter is closest to the stimulus parameter signal is reduced. Hence the BMI system becomes more accurate at determining which label the user is attempting to select. To improve the degree of training possible, the correct selection data set may further comprise one or more of: the determined stimulus parameter signal, allowing the SSVEP detection algorithm identify differences in the determined stimulus parameter and the actual stimulus parameter; - the selected label and the arrangement of the plurality of labels at the time the brain signal was received, allowing the SSVEP detection algorithm to identify any trends linked to positioning of the label and corresponding stimulus; and - the arrangement of the plurality of stimuli at the time the brain signal was received, allowing the detection algorithm to identify if adjacent stimuli affect the brain signal, for example. In embodiments of the invention, following the step of adding one tally to the counted number of consecutive cancel eye gestures, the method comprises the further steps of: determining whether the counted number is equal to a predetermined rearrangement number wherein: if the counted number is less than the predetermined rearrangement number, proceeding with repeating steps (a) to (h); or if the counted number is equal to the predetermined rearrangement number, rearranging the plurality of stimuli such that each stimulus is associated with a different label, resetting the counted number to zero and proceeding with repeating steps (a) to (h). In such embodiments of the invention, the predetermined rearrangement number may be one such that the plurality of stimuli is rearranged relative to the plurality of labels every time that a cancel eye gesture is detected. This would reduce the likelihood of a user failing to select a particular label twice in a row due to the label being associated with a stimulus that is unsuitable for the user. A stimulus may be considered unsuitable for the user if the stimulus fails to evoke an SSVEP from which the SSVEP detection algorithm(s) can accurately and consistently determine the stimulus parameters. However, an unsuitable stimulus may not always be the reason that a user performs cancel eye gesture. For example, a label may be incorrectly selected because the user was distracted while making their selection, or in cases where the user looked at a label without meaning to select the label and perform the corresponding action. Therefore, it may not be necessary for the predetermined rearrangement number to be one. Instead the predetermined rearrangement number may be two, three or any suitable number that stops the user from being permanently obstructed from selecting a particular label but that also reduces the amount that stimuli are rearranged relative to the labels. Reducing unnecessary rearrangement of the stimuli may improve their effectiveness and / or reduce irritation caused to the user’s eyes as they are able to acclimatise to the positioning of certain stimuli. In embodiments of the invention, if the plurality of stimuli are rearranged and there is no cancel eye gesture determined with respect to a selected label determined from a successively received brain signal, the method comprises the further steps of: identifying that the stimulus parameter associated with the selected label before the rearrangement of the plurality of stimuli is problematic; storing in memory a record of the problematic stimulus parameter; and counting the number of times a stimulus parameter is identified as problematic. In such embodiments of the invention, the BMI system may identify stimulus parameters that are problematic, i.e., potentially unsuitable for the user. The BMI system may also keep track of how many times a particular stimulus parameter is identified as problematic. Each identification of a stimulus parameter as being problematic increases the likelihood that the stimulus parameter is unsuitable for the user. Therefore, in embodiments of the invention, if a stimulus parameter is identified as problematic a predetermined number of times, the method may comprise the further step of updating the problematic stimulus parameter. In such embodiments of the invention, when updating the stimulus parameter, the frequency may be changed to a frequency similar to one that forms part of a stimulus parameter that is functioning well, i.e. one that is present in several recent correct selection data sets. In order to ensure the SSVEP detection algorithm is able to discriminate between the new stimulus parameter and the existing, well-performing stimulus parameter, the new stimulus parameter may be provided with a different phase to the existing stimulus parameter. In embodiments of the invention, the method comprises the further steps, prior to step (b), of: tracking the user’s point of gaze; determining a region of interest based on the point of gaze; and limiting the number of stimuli that are displayed such that the number of stimuli corresponds to the number of labels positioned within the region of interest only, wherein each stimulus displayed in step (b) is associated with a different label within the region of interest. Limiting the number of flickering stimuli that are displayed to the number of labels that are positioned in a region of the display is advantageous because it reduces the amount of irritation that may be caused by having several stimuli flickering with different parameters. Further, it means that fewer stimuli are required. The stimuli within the region of interest may therefore be allocated using stimuli with the most discriminatory stimulus parameters based on information from one or more correct selection data sets stored in memory. The accuracy with which the user is able to operate the BMI system would thereby be improved. In such embodiments of the invention the step of tracking the user’s point of gaze may comprise the steps of: receiving a gaze signal comprising electroencephalography (EEG) data, electrooculography (EOG) data and / or video-oculography (VOG) data; and determining a point of gaze of the user based on the gaze signal. Determining the point of gaze of the user from EEG signals avoids requiring hardware additional to that already required for the receiving of EEG signals from which SSVEPs may be detected, thereby reducing the cost of the BMI system. In embodiments of the invention the method comprises the further steps, prior to step (a), of: continuously recording eye movement signals from the user; determining whether a recorded eye movement signal corresponds to a predetermined activate eye gesture; and if an activate eye gesture is determined, activating the display and performing step (a); or if an activate eye gesture is not determined, repeating the activate eye gesture determining step. In such embodiments of the invention, a BMI system capable of carrying out the method provided by the invention may be fitted to a user, turned on and in a standby mode. While in standby mode the BMI system may continuously record eye movement signals from the user. If the user would like to use the BMI system they may perform a predetermined activate eye gesture, such as a double blink or a wink. While still in standby mode, the BMI system may determine whether an eye movement signal is indicative that the user carried out an activate eye gesture. If no activate eye gesture is identified then the BMI system may repeat this process with the next eye movement signal that is recorded. However, if an activate eye gesture is identified then the BMI system may activate a display and perform step (a). Therefore, by means of the invention, the user may switch the BMI system from a standby mode to an active mode wherein the BMI system is ready to use for performing an action. Further, the user process to activate the BMI system may also be performed asynchronously, without time limitations. This may be particularly advantageous for a user with severely restricted mobility, as the user would be able to use the BMI system intermittently throughout the day without requiring the assistance of another person to manually turn the system on and without having to abide by a predetermined schedule for using the BMI system. Once the BMI system is in the active mode it may operate synchronously or asynchronously. In synchronous operation the BMI system will operate on the basis that the user is attempting to select one of the plurality of labels being displayed until the user chooses to return the BMI system to the standby mode. Accordingly, although the label selection process is synchronous and defined by predetermined time periods, the user has the freedom to enter and exit the label selection process freely by carrying out the activate eye gesture and selecting the label associated with the standby mode respectively. This may therefore provide the user with a good level of independence and freedom in combination with providing a simpler SSVEP detection algorithm which may operate with greater accuracy. Further, the user may be exposed to the flickering stimuli only when he / she has actively chosen to activate the BMI system, thereby reducing irritation that may be caused by the flickering. Alternatively, in asynchronous operation the user may be provided with complete freedom in terms of both when to activate the BMI system and when to select one of the plurality of labels. This may be particularly advantageous in situations where the user wishes to multitask between operating the BMI system and carrying out other activities such as interacting with another person or watching television. In embodiments of the invention, if the selected label determined in step (e) corresponds to an “off” action and a cancel eye gesture is not determined in step (g); then the method comprises the further steps of: deactivating the display; and repeating the steps of continuously recording eye movement signals from the user; determining whether a recorded eye movement signal is indicative that the user carried out an activate eye gesture; and if an activate eye gesture is identified, activating the display and moving to step (a); or if an activate eye gesture is not identified, repeating the activate eye gesture determining step. In such embodiments of the invention, a BMI system may display a plurality of labels including a label defining an action to switch the BMI system to a standby mode which may involve turning off a display forming part of the BMI system. A user of the BMI system who wants to switch the BMI system into the standby mode may look at the relevant label and associated stimulus. The BMI system may determine that the user is looking at the ‘standby label’ by performing steps (c) to (g) as set out above. Once the ‘standby label’ has been determined correctly by the BMI system, the display may be deactivated, and the BMI system switched to a standby mode wherein eye movement signals are recorded and analysed until the user performs an activate eye gesture as set out above. By means of the invention, the user may switch the BMI system from an active mode to a standby mode. The label selection process may be synchronous or asynchronous. Accordingly the process for the user to switch the BMI system from the active mode to the standby mode may be synchronous or asynchronous. Further, in combination with the method steps set out above to switch the BMI system from a standby mode to an active mode, the user is able to switch between the two modes at will. The ability to switch between active and standby modes therefore gives the user freedom in how they interact with the BMI system, which may be particularly advantageous for users with severely restricted mobility who may be unable to otherwise interact with electronic devices. Additionally, the display showing a plurality of flickering stimuli can become an irritation. Therefore the ability for a restricted mobility user to actively turn off the display at his or her own discretion is also advantageous over known BMI systems which may only switch to a standby mode after extended periods of inactivity from the user. In other embodiments of the invention the method may comprise the further steps of determining whether a recorded eye movement signal corresponds to a predetermined deactivate eye gesture; and if a deactivate eye gesture is determined, deactivating the display. In such embodiments of the invention the user may not be required to carry out the label selection process in order to switch the BMI system to the standby mode and may instead perform eye gestures asynchronously to switch the BMI system between modes. In embodiments of the invention, if the selected label determined in step (e) corresponds to an application action and the cancel eye gesture is not determined in step (g); then the step of performing an action corresponding to the selected label comprises issuing a command signal representing a command to perform the application action. The command signal may be transmitted to any suitable device in communication with the BMI system. For example, the command signal may be transmitted to a device that affects the user’s environment (such as a motorised bed, an air conditioning system, lights or motorised blinds) or a device for the user’s entertainment (such as a televisions, radio or smart speaker). By means of such embodiments of the invention, a user with restricted mobility may be able to improve his or her own comfort, by controlling devices such as a motorised bed, without requiring the aid of another person. Additionally, the user could entertain himself or herself while alone by controlling a device such as a television. This could improve the quality of life experienced by the user and reduce the workload for family members or health professionals caring for people with restricted mobility. According to a second aspect of the invention there is provided a system comprising a controller configured to: receive brain signals; receive eye movement signals; determine a label-stimuli configuration wherein each of a plurality of stimuli is associated with a different one of a plurality of labels; cause a display device to display the plurality of labels and the plurality of stimuli according to the label-stimuli configuration; apply an SSVEP detection algorithm to the brain signals in order to detect an SSVEP and determine a stimulus parameter signal from the SSVEP; determine which one of the plurality of stimuli corresponds to a stimulus parameter which is closest to the determined stimulus parameter signal and identify the label associated with that stimulus as a selected label; cause the selected label to be indicated to the user; determine whether an eye movement signal corresponds to a predetermined cancel eye gesture; count consecutive cancel eye gestures; store in memory a correct selection data set comprising a brain signal and a stimulus parameter; and cause an action to be performed that corresponds to the selected label. By means of the invention, a user of the system may view a display device showing a plurality of labels wherein each label defines an action that the user might want to perform. Further, each of the plurality of labels may be associated with one of a plurality of stimuli according to the label-stimuli configuration, wherein each stimulus has individual predetermined parameters, such as flicker frequency. If the user wishes to perform any of the actions defined by the plurality of labels, the user may look at the relevant label and the associated stimulus should excite the user’s retinas and induce an SSVEP to be generated in the user’s occipital region of the brain. Brain signals may be received by the system and the SSVEP detection algorithm may detect an SSVEP from the brain signals and determine a stimulus parameter signal from the SSVEP. The system may determine which label the user was looking at based on which one of the plurality of stimuli corresponds to a stimulus parameter which is closest to the determined stimulus parameter signal. The system may indicate the determined label to the user who may then decide whether the determination is correct. If the determined label is not the label that the user intended to choose then the user may perform a predetermined cancel eye gesture, such as a double blink or a wink, which allows the user to repeat the selection process. If the determined label is the label that the user intended to choose then the user is not required to perform any further action and system may cause the action to be performed that corresponds to the selected label. Therefore, by means of the invention, the user may use the system to perform an action with only brain activity and eye movements. In some embodiments of the invention the system may be configured to apply the SSVEP detection algorithm to brain signals received during a predetermined period of time in order to detect an SSVEP and determine a stimulus parameter signal from the SSVEP. Such embodiments of the invention may therefore operate synchronously according to predetermined timings. In other embodiments of the invention the system may be configured to determine that the user is idle if no SSVEP is detected. Accordingly the system may continuously apply the SSVEP detection algorithm to brain signals received during a predetermined period of time in order to detect an SSVEP until an SSVEP is detected. Only once an SSVEP is detected would the system determine a stimulus parameter signal from the SSVEP. Such embodiments of the invention may therefore operate asynchronously according to a pace set by the user. In embodiments of the invention the controller is further configured to determine a rearranged label-stimuli configuration wherein each of the plurality of stimuli is associated with a different label compared to the label-stimuli configuration; cause a display device to display the plurality of labels and the plurality of stimuli according to the rearranged labelstimuli configuration. In embodiments of the invention the controller is further configured to determine whether a recorded eye movement signal corresponds to a predetermined activate eye gesture. In embodiments of the invention the controller is further configured to store in memory a correct selection data set comprising a brain signal, a stimulus parameter and one or more of a determined stimulus parameter signal and, a selected label, an arrangement of the plurality of labels and an arrangement of the plurality of stimuli. In embodiments of the invention the controller is further configured to use a correct selection data set to train a training-based SSVEP detection algorithm. In embodiments of the invention, the system comprises a brain signal electrode configured to continuously receive brain signals. The brain signal electrode may be non-invasive. The brain signal electrode may particularly be configured to receive EEG signals. In use, the brain signal electrode may be positioned against the user’s head, and in particular the brain signal electrode may be positioned close to the occipital region of the user’s brain. In some embodiments of the invention the system may comprise a plurality of brain signal electrodes which may comprise one or more sets of brain signal electrodes. In embodiments of the invention, the system comprises an eye movement electrode configured to continuously receive eye movement signals. The eye movement electrode may be non-invasive. In use, the eye movement electrode may be positioned against the user’s head, and in particular the eye movement electrode may be positioned close to one of the user’s eye sockets. In some embodiments of the invention the system may comprise a plurality of eye movement electrodes which may comprise one or more sets of eye movement electrodes. In embodiments of the invention, the controller is operatively connectable to external software or hardware. In such embodiments the controller may cause an external device to perform an action by transmitting a signal to the device. Therefore, the system allows a user to control a device such as a motorised bed or a television with eye movements and brain signals only. In embodiments of the invention, the system further comprises a headcap adapted to fit a user’s head. The headcap may be any shape suitable to be worn on the head. For example, the head cap may be in the form of a hat or a headband. The headcap may be coupled to a brain signal electrode and an eye movement electrode, and the headcap may be adapted to position the electrodes against the user’s head. The headcap may be adapted to be one or more of: lightweight, comfortable, breathable and discreet so that the user may wear the headcap for long period of time without it causing irritation. In embodiments of the invention, the system further comprises a pre-processing box configured to optimise the brain signals and the eye movement signals. In such embodiments of the invention, brain signals and / or eye movement signals received from the user may be transmitted to the pre-processing box. The pre-processing box may be configured to pre-process a received brain signal or eye movement signal by filtering the relevant data, removing artefacts and reducing noise. In embodiments of the invention, the system comprises a gaze tracker. The gaze tracker may be configured to track the user’s point of gaze such that a region of interest based on the point of gaze can be determined. The controller may limit the number of stimuli that are displayed such that the number of stimuli corresponds to the number of labels positioned within the region of interest only. Limiting the number of flickering stimuli that are displayed to the number of labels that are positioned in a region of the display is advantageous because it reduces the amount of irritation that may be caused by having several stimuli flickering with different parameters. Further, it means that fewer stimuli are required so the BMI system may select stimuli parameters which are best suited for the particular user and thereby improve the accuracy with which the user is able to operate the BMI system. In embodiments of the invention, the gaze tracker is operatively coupled to the eye movement electrode and receives electrooculography (EOG) data. In such embodiments the gaze tracker may be configured to determine the user’s point of gaze from the EOG data. Further, the eye movement electrode may be a plurality of eye movement electrodes wherein at least one of the eye movement electrodes is configured to receive EOG data. This electrode may be positioned close to the prefrontal part of the brain. An advantage of this is that the gaze tracker may be discreet as no cameras are required to monitor the user’s eye movements. Further, the use of a single eye movement electrode to receive EOG data is more discreet than known EOG measuring devices which include a 4-channel configuration, with 2 electrodes at the lateral canthi, one on top of an eye and one below the eye. In embodiments of the invention, the gaze tracker comprises a camera and receives videooculography (VOG) data. In such embodiments the gaze tracker may be configured to determine the user’s point of gaze from video footage of one or both of the user’s eyes. An advantage of this is that the gaze tracker may accurately track an exact point on a display device at which the user is looking. The invention will now be described by way of example only with reference to the accompanying drawings in which: Figure 1 is a schematic representation of a BMI system according to an embodiment of the invention; Figure 2 is a schematic representation of a display forming part of the BMI system shown in Figure 1; Figure 3 is a schematic representation of a method according to an embodiment of the invention; Figure 4 is a schematic representation of method steps forming part of a method according to an embodiment of the invention; Figure 5 is a schematic representation of method steps forming part of a method according to an embodiment of the invention; Figure 6 is a schematic representation of method steps forming part of a method according to an embodiment of the invention; Figure 7 is a schematic representation of method steps forming part of a method according to an embodiment of the invention; Figure 8 is a schematic representation of method steps forming part of a method according to an embodiment of the invention; Figure 9 is a schematic representation of the display shown in Figure 2 before and after the method steps shown in Figure 8 are performed; Figure 10 is a schematic representation of method steps which are similar to those shown in Figure 6 but with additional steps included; Figure 11 is a schematic representation of method steps which are similar to those shown in Figure 6 but with a step removed; Figure 12 is a graphical plot of EEG data representative of an eye gesture, specifically a double blink; and Figure 13 is a graphical plot of EEG data representative of an SSVEP triggered by a stimulus flickering at 10.25 Hz. Referring initially to Figure 1, a BMI system according to an embodiment of the invention is generally defined by the reference numeral 2. The BMI system 2 comprises a controller 10, a display 20, a wearable device 30, a brain signal sensor 40, an SSVEP detector 50, a label determiner 60, an eye movement sensor 70 and an eye movement processor 80. The controller 10 is configured to determine a label-stimuli configuration wherein each of a plurality of stimuli is associated with a different one of a plurality of labels. Further, the controller 10 is coupled to the display 20 and configured to determine a display signal 11, and particularly a menu display signal, and transmit the menu display signal to the display 20. The menu display signal may be configured to cause the display 20 to display a plurality of labels 22 and a plurality of stimuli according to the label-stimulus configuration. Referring briefly to Figure 2, the display 20 is shown displaying the plurality of labels 22 wherein each label 22 is associated with a function (F_A, F_B, F_C, F_D, F_E and F_F) and a stimulus (S1, S2, S3, S4, S5 and S6) according to the label-stimulus configuration. Each function may be a function or action that the user of the BMI system 2 might wish to perform such as controlling a motorised bed, controlling an air conditioning unit, controlling a television or switching the display into a standby mode. Each stimulus may involve a region of the screen flickering with a different stimulus parameter. The stimulus parameter may comprise one or more predetermined characteristics, such as frequency, pattern or colour. That is, each stimulus may flicker at a different frequency. For example each stimulus may flicker at a different frequency within the range of 6 Hz to 40 Hz and the stimulus parameter corresponding to each stimulus may be considered as the specific flicker frequency of the stimulus, 20 Hz for example. Each stimulus parameter may be configured such that, when the user 4 looks at the stimulus, the user’s retinas are excited in a particular way to generate an SSVEP. Therefore, when an SSVEP is detected, it is possible to determine a stimulus parameter signal based on the SSVEP (the stimulus parameter signal being representative of a hypothetical stimulus parameter that caused the SSVEP) and compare that determined stimulus parameter signal with the stimulus parameters corresponding to each of the plurality of stimuli. In this example there are six labels associated with six functions and six stimuli. However, the display 20 may display any suitable number of labels associated with a corresponding number of functions. The number of stimuli may be equal to the number of labels, as is the case in this example. However there are also embodiments of the invention where the number of stimuli displayed is lower than the total number of labels displayed, i.e. not all labels are associated with a stimulus at one time. In order for such SSVEPs to be sensed and analysed the user 4 may wear the wearable device 30, which may be any suitable device that is wearable on the head, such as a headband, head cap or hat. The wearable device 30 is coupled to the brain signal sensor 40 which may comprise a brain signal electrode, or a plurality of brain signal electrodes, (not shown) such that the brain signal electrode may be positioned against the head of the user 4. Further, the brain signal electrode may be positioned close to the occipital region of the user’s brain. The brain signal electrode may receive a brain signal 12 from the user 4. The brain signal sensor 40 may further comprise a pre-processing box (not shown) which is configured to pre-process a received brain signal 12 by filtering the relevant data, removing artefacts and reducing noise. The brain signal sensor 40 is coupled to the SSVEP detector 50 and is configured to transmit the brain signal 12 to the SSVEP detector 50. The SSVEP detector 50 is configured to apply an SSVEP detection algorithm to the brain signal 12. The SSVEP detection algorithm may be configured to detect whether the data forming the brain signal indicates that an SSVEP has been induced in the occipital region of the user’s brain by the user looking at a label 22 and associated stimulus displayed on the display 20. The SSVEP detection algorithm is further configured to determine a stimulus parameter signal 13 that is representative of a hypothetical stimulus parameter that caused the SSVEP. The SSVEP detector 50 is coupled to the label determiner 60 and configured to transmit the stimulus parameter signal 13 to the label determiner 60. The label determiner 60 is in turn coupled to the controller 10 and configured to receive a label-stimuli signal 14 from the controller 10 that indicates the label-stimulus configuration displayed to the user 4 via the display 20. The label determiner 60 is further configured to determine which of the plurality of stimuli corresponds to a stimulus parameter which is closest to the stimulus parameter signal 13 and therefore determine which of the plurality of labels 22 the user 4 most likely looked at to induce the SSVEP. This label may be identified as the selected label. The label determiner 60 is configured to transmit a selected label signal 15 to the controller 10 that indicates which of the plurality of labels 22 has been determined as the label selected by the user 4. Upon receiving the selected label signal 15, the controller 10 may determine a new display signal 11, and particularly a label display signal, wherein the label display signal causes the display 20 to indicate the selected label to the user 4. For example, the selected label may be highlighted amongst the plurality of labels (such as with a change of colour, size or boldness) or all labels except the determined label L may be hidden. The controller 10 may transmit the label display signal to cause the display 20 to indicate the selected label to the user 4 for a predetermined period of time, during which the user 4 may assess whether the selected label was determined correctly. In other words, the user 4 can check whether the selected label determined by the BMI system 2 is the label that the user intended to select when looking at the display 20. If the selected label is correct the user is not required to perform any further action and, once the time for indicating the selected label has elapsed, the controller 10 may transmit an output signal 18 that causes a function or action corresponding to the selected label to be performed. The output signal 18 may be transmitted to the display 20 and cause display 20 to switch to an off / standby mode or change a display setting, such as brightness or colour scheme, or display a sub-menu which comprises a new plurality of labels. The output signal 18 may alternatively be transmitted to an external device in communication with the BMI system 2. For example, the BMI system 2 could be in communication with a motorised bed occupied by the user 4 and the output signal 18 could be a command to raise the angle of the bed to move the user 4 into a sitting position. Any suitable devices may be controlled by the user 4 with the BMI system 2. By way of example, such devices include devices that affect the user’s environment (such as air conditioning systems, lights or motorised blinds) and devices for the user’s entertainment (such as televisions, radios or smart speakers). If the selected label is not correct, the user may perform a predetermined cancel eye gesture, such as a double blink or a wink. The wearable device 30 is further coupled to the eye movement sensor 70. The eye movement sensor 70 may comprise an eye movement electrode, or a plurality of eye movement electrodes, (not shown) such that the eye movement electrode may be positioned against the head of the user 4. Further, the eye movement electrode may be positioned close to the eyes of the user 4. The eye movement electrode may receive an eye movement signal 16 from the user 4. The eye movement sensor 70 may further comprise a pre-processing box (not shown) which is configured to pre-process a received eye movement signal 16 by filtering the relevant data, removing artefacts and reducing noise. The eye movement sensor 70 is coupled to the SSVEP determiner 50 and is configured to transmit the eye movement signal 16 to the eye movement processor 80. The eye movement processor 80 is configured to determine whether an eye movement signal 16 corresponds to the predetermined cancel eye gesture. The eye movement processor 80 is coupled to the controller 10 and, upon determining a predetermined cancel eye gesture, the eye movement processor 80 may transmit an eye gesture signal 17 to the controller 10. If the controller 10 receives an eye gesture signal 17 from the eye movement processor 80 during the time that the selected label is indicated to the user 4, the controller 10 will not issue an output signal 18. Instead, the controller 10 will store a record of the cancel eye gesture and selected label, count the number of consecutive cancel eye gestures that have been determined and determine whether the counted number is equal to a predetermined rearrangement number. If the counted number is less than the predetermined rearrangement number, the controller may reissue the menu display signal 11 referred to earlier. However, if the counted number is equal to the predetermined rearrangement number the controller may determine a rearranged label-stimulus configuration wherein each of the plurality of stimuli are associated with a different one of a plurality of labels that was the case for the previous label-stimulus configuration. The controller 10 may also reset the count. Further, the controller 10 will determine a new display signal 11, and particularly a rearranged menu display signal, and transmit the rearranged menu display signal to the display 20. The rearranged menu display signal may be configured to cause the display 20 to display the plurality of labels 22 and the plurality of stimuli according to the rearranged label-stimulus configuration. Also, the controller 10 may issue a new label-stimulus signal 14 to the label determiner 60 that indicates the rearranged label-stimulus configuration so that the label determiner 60 is able to associate a stimulus with a label 22 according to the rearranged configuration rather than the previous configuration. The BMI system 2 is then ready for the user 4 to attempt a new label selection. If the label-stimulus configuration was kept the same after a selected label was incorrectly determined, it is possible that the user 4 could attempt to select the same label as before and fail to do so again because the parameters of the stimulus associated with that label (on both attempts) are unsuitable for the user 4. For example, the flicker frequency of that stimulus may not excite the user’s retinas suitably to generate an SSVEP, instead the determined SSVEP signal might have been caused by a stimulus in the user’s peripheral vision. Alternatively, the stimulus parameters might not have the expected effect on that particular user which causes the BMI system 2 to mis-identify the resulting SSVEP signal 13. Therefore, an advantage of rearranging the label-stimulus configuration after a predetermined number of incorrect determinations of a selected label is that, if the user 4 attempts to select the same label a further time, the label will be associated with a new stimulus and it is likely that the user will be able to select the desired label. Further, the more this process is repeated, the more likely it is that the user 4 will succeed in selecting the desired label. Whereas, if there was no rearrangement of the label stimulus configuration, it could be impossible for the user 4 to succeed in selecting the desired label. As set out above, the BMI system 2 may display a plurality of labels and a plurality of stimuli to the user 4 according to a label-stimulus configuration. The brain signal sensor 40, SSVEP determiner 50 and label determiner 60 may then receive brain signals 12 from the user 4, identify data indicative of an SSVEP within a brain signal 12 and determine which stimulus and associated label the user 4 was most likely looking at to cause the SSVEP. The controller 10 may cause the display to indicate that label to the user 4 so that the user 4 may either confirm the label was determined correctly by performing no predetermined cancel eye gesture, or indicate the label was determined incorrectly by performing a predetermined cancel eye gesture. If the user 4 confirms that the label was determined correctly, the controller 10 may cause an action to be performed that corresponds to that label by transmitting an output signal 18. If the user 4 performs the predetermined cancel eye gesture, the controller 10 may determine a rearranged labelstimulus configuration and cause the display 20 to display the plurality of labels and the plurality of stimuli according to the rearranged label-stimulus configuration so that the user 4 may reattempt selection of a label in order to perform a desired action. Further to this functionality, the BMI system may be configured to perform further functions. The BMI system 2 may be configured to switch from a standby mode wherein the display 20 is turned off to an active mode wherein the display 20 is turned on. If the user 4 is wearing the wearable device 30 when the BMI system 2 is in standby mode, the user 4 may perform a predetermined activate eye gesture, such as a double blink, wink or eye roll in order to switch the BMI system 2 into active mode. When the BMI system is in standby mode, the eye movement sensor 70 may continuously receive eye movement signals 16 from the user 4 and transmit the eye movement signals 16 to the eye movement processor 80. The eye movement processor 80 may be configured to determine whether an eye movement signal 16 corresponds to the predetermined activate eye gesture. The eye movement processor 80 is coupled to the controller 10 and, upon determining a predetermined cancel eye gesture, the eye movement processor 80 may transmit an eye gesture signal 17 to the controller 10. If the controller 10 receives an eye gesture signal 17 from the eye movement processor 80 when the BMI system is in standby mode, the controller 10 will transmit a display signal 11, and specifically a menu display signal, to the display 20. The display signal may be configured to cause the display 20 to turn on and display the plurality of labels and the plurality of stimuli as set out above, thereby switching the BMI system 2 to active mode. The user 4 may then switch the BMI system 2 back to the standby mode by selecting a label that corresponds to the action of switching the BMI system 2 to the standby mode. In which case, the output signal 18 would be transmitted to the display 20 and would be configured to cause the display to turn off. Being able to switch the BMI system 2 between a standby mode and an active mode is advantageous because the flickering stimuli displayed on the display 20 may be irritating to look at over a long period of time, particularly if the user 4 is not actively interacting with the BMI system 2 but is noticing the flickering in peripheral vision. Being able to switch the BMI system 2 between modes by using a predetermined activate eye gesture and selecting a label associated with switching the BMI system to the standby mode may be particularly beneficial for a user 4 with severely restricted mobility as such a user might be unable to perform the same action manually (by pressing a button for example). Also, a user with restricted mobility might be unable to move the display 20 or even look away from the display 20 to avoid looking at the flickering stimuli. Therefore, being able to turn off the display when it is not needed may be even more useful for a user with restricted mobility. The BMI system 2 may also be configured to learn from instances when a label 22 has been correctly determined. When a user 4 confirms that a label was determined correctly by the BMI system, by performing no predetermined cancel eye gesture, the controller 10 may store in memory a correct selection data set comprising the brain signal, the determined stimulus parameter signal and the stimulus parameter of the stimulus associated with the selected label. The controller 10 may then use correct selection data set to improve the performance of the BMI system 2, for example by training the SSVEP detection algorithm or determining new stimulus parameters for future stimuli displayed by the BMI system 2. The longer that a specific user 4 uses a specific BMI system 2, the more opportunities the BMI system 2 will have to improve its performance in accurately determining the labels selected by the user 4 based on an increasing volume of correct selection data sets. This will allow the controller 10 to determine label-stimulus configurations and stimulus parameter signals wherein the user’s selection of a label is very likely to be determined correctly by the BMI system, no matter which label the user wants to select. Therefore, over time, a user 4 may find that the BMI system 2 becomes very accurate at determining what actions the user 4 is trying to select and operation of the BMI system may become both easy and reliable. The BMI system 2 may also be configured to track a user’s gaze and limit the number of stimuli displayed to the number of labels that are positioned within a particular region of the display 20. The BMI system 2 may further comprise a gaze tracker (not shown). When the display 20 is displaying a plurality of labels the gaze tracker may continuously receive gaze signals from the user 4 and transmit the gaze signals to a gaze processor (not shown). The gaze processor may be configured to determine whether a gaze signal indicates the user’s point of gaze on the display 20. The gaze processor is coupled to the controller 10 and, upon determining the user’s point of gaze on the display 20, the gaze processor may transmit a gaze signal, that indicates the user’s point of gaze, to the controller 10. If the controller 10 receives such a gaze signal from the gaze processor, the controller 10 may determine a region of interest centred around the point of gaze, and transmit a display signal 11, specifically a focused menu display signal, to the display 20. The display signal 11 may be configured to cause the display 20 to display a plurality of stimuli that are associated only with a plurality of labels that are positioned within the region of interest. Limiting the number of flickering stimuli that are displayed to the number of labels that are positioned in a region of the display 20 is advantageous because it reduces the amount of irritation that may be caused by having several stimuli flickering with different parameters. Further, it means that fewer stimuli are required so the BMI system may select stimuli which are best suited for the particular user and thereby improve the accuracy with which the user is able to operate the BMI system. Referring now to Figure 3, a method according to an embodiment of the invention is generally defined by the reference numeral 102. The method 102 is a method for performing an action based on a user’s brain activity. In order for the method 102 to be carried out, a BMI system may be fitted to a person and turned on. The BMI system may be a BMI system according to an embodiment of the invention, such as the BMI system 2 shown in Figure 1, although any suitable BMI system may be used provided it is capable of carrying out the method shown in Figure 3. The method 102 comprises the steps: “Display labels” 106, “Display stimuli” 108, “Receive a brain signal” 110, “Detect an SSVEP” 112, “Determine a selected label” 114, “Indicate selected label” 115, “Receive an eye movement signal” 116, “Eye gesture?” 118, “Store eye gesture” 120 and “Perform action” 122. Each step is described below. “Display labels” 106 involves displaying a plurality of labels to a user of the BMI system. For example, the plurality of labels may be displayed as is demonstrated in Figure 2. “Display stimuli” 108 involves displaying a plurality of stimuli to the user of the BMI system wherein each stimulus is associated with a different label. “Receive a brain signal” 110 involves receiving a brain signal while the user is exposed to the plurality of stimuli. This step may be carried out by a brain signal sensor such as the brain signal sensor 40 shown in Figure 1. In particular this step may be carried out by a brain signal electrode that is coupled to a wearable device, such as the wearable device 30 shown in Figure 1, and positioned against the head of the user. “Detect an SSVEP” 112 involves detecting an SSVEP from the received brain signal using an SSVEP detection algorithm that forms part of the BMI system. Further, the SSVEP detection algorithm determines a stimulus parameter signal from the SSVEP. This step may be carried out by an SSVEP detector such as the SSVEP detector 50 shown in Figure 1. “Determine a selected label” 114 involves determining which one of the plurality of stimuli corresponds to a stimulus parameter that is closest to the stimulus parameter signal and identifying the label associated with that stimulus as a selected label. “Indicate selected label” 115 involves indicating the selected label to the user so that the user can judge whether the selected label determined by the BMI system is the label that the user intended to select. “Receive an eye movement signal” 116 involves receiving an eye movement signal while the user is exposed to the indication of the selected label. The “Receive an eye movement signal” step 116 may involve receiving eye movement signals over a predetermined period of time parallel to indicating the selected label, which forms part of the “Determine a selected label” step 114. The eye movement signals may comprise data that is representative of eye movements performed by the user of the BMI system during the predetermined time period. Within the predetermined period of time, the user of the BMI system may judge if the selected label is correct, i.e. the label that the user intended to select. If the selected label is incorrect, the user may be required to perform a predetermined cancel eye gesture, such as a double blink or a wink. Conversely, if the selected label is correct, the user may perform no further action and, in particular, the user may be required to avoid performing the cancel eye gesture. “Eye gesture?” 118 involves determining whether a received eye movement signal is indicative that the user carried out a cancel eye gesture. Figure 12 shows an example of an eye movement signal 16, in this case formed of EEG data, that comprises a double-peak 302 which is indicative that the user performed a double blink. That is, each peak exceeds a specific pre-defined threshold (determined during training) which indicates a blink gesture and both peaks are within a predetermined time window which implies that a purposeful double blink was performed (rather than the user blinking twice coincidentally). Other eye gestures such as a wink or an eye roll may be identified based on different signal characteristics that are unique to that gesture. There are two possible steps following “Eye gesture?” 118. If the answer to “Eye gesture?” 118 is yes - a cancel eye gesture is identified - then the step “Store eye gesture” 120 may be carried out before repeating the step “Display labels” 106. The “Store eye gesture” step 120 involves adding one tally to a counted number of consecutive cancel eye gestures. In some embodiments of the invention, it may be determined whether the counted number is equal to a predetermined rearrangement number. If the counted number is less than the predetermined rearrangement number the method may proceed to the “Display labels” step 106. However, if the counted number is equal to the predetermined rearrangement number, the method may include the additional steps rearranging the plurality of stimuli such that each stimulus is associated with a different label, resetting the count and only then preceding to the “Display labels” step 106. A reason behind this portion of the method 102 is that, if the cancel eye gesture is identified from the received eye movement signal, the implication is that the user did not intend to choose the selected label and the user wishes to re-attempt the selection process. If the “Store eye gesture” step 120 was not performed and the stimuli were never rearranged before re-displaying the pluralities of labels and stimuli, it is possible that the user could attempt to select the same label as before and fail to do so again because the parameters of the stimulus associated with that label (on both attempts) are unsuitable for the user. For example, the flicker frequency of that stimulus may not excite the user’s retinas suitably to generate an SSVEP, instead the SSVEP might have been caused by a stimulus in the user’s peripheral vision. Alternatively, the stimulus parameter might not have the expected effect on that particular user which causes the BMI system to determine a stimulus parameter signal representative of a stimulus parameter that is quite different to the true stimulus parameter which caused the SSVEP. To avoid a situation where the user repeatedly fails to select a desired label because the associated stimulus is unsuitable, the plurality of stimuli may be rearranged after each incorrect label selection, or after a certain number of incorrect label selections related to the same label, so that each label is associated with a different stimulus. Therefore, if the user attempts to select the same label, the label will be associated with a new stimulus and it is likely that the user will be able to select the desired label. Further, the more times this rearranging process is repeated, the more likely it is that the user will succeed in selecting the desired label. Now referring back to the “Eye gesture?” step 118. If the answer to “Eye gesture?” 118 is no - a cancel eye gesture is not identified - then the step “Perform action” 122 may be carried out and the method 102 for performing an action based on a user’s brain activity is completed. “Perform action" 122 may involve an action being performed by the BMI system, such as switching the screen to an off / standby mode or changing a display setting, such as brightness or colour scheme, or entering a sub-menu which displays a new plurality of labels. “Perform action” 122 may also involve an action signal being issued to an external device in communication with the BMI system. For example, the BMI system could be in communication with a motorised bed occupied by the user of the BMI system and action signal could be a command to raise the angle of the bed to move the user into a sitting position. Any suitable devices may be controlled by the user with the BMI system and the method set out in Figure 3. By way of example such devices include devices that affect the user’s environment (such as air conditioning systems, lights or motorised blinds) and devices for the user’s entertainment (such as televisions, radios or smart speakers). Further to the desired action being performed, if the cancel eye gesture is not identified from the received eye movement signal, the implication is that the user intended to choose the selected label and the BMI system correctly determined the label based on the user’s brain activity. Accordingly, a correct selection data set comprising the brain signal, the determined stimulus parameter signal and the stimulus parameter of the stimulus associated with the selected label may be stored in memory and used to improve the accuracy of the SSVEP detection algorithm in future. Referring now to Figures 4 to 8, a method according to an embodiment of the invention is shown in five stages. The method is similar to the method 102 set out in Figure 3 except it includes some additional steps. In order for the method to be carried out, a BMI system may be fitted to a person and turned on. The BMI system may be a BMI system according to an embodiment of the invention, such as the BMI system 2 shown in Figure 1, although any suitable BMI system may be used provided it is capable of carrying out the method shown in Figures 4 to 8. First, second, third, fourth and fifth stages of the method are generally defined by the reference numerals 201,202, 203, 204 and 205 respectively and shown in Figures 4, 5, 6, 7 and 8 respectively. Referring now to Figure 4, the first stage 201 of the method comprises the steps: “Recording of data” 210 and “Pre-processing” 212. Once a BMI system is fitted to a user and turned on, the first step, “Recording of data” 210, involves causing the continuous recording of data to begin and this may continue throughout the method. The data recorded can include brain signals, and particularly EEG signals, received by a non-invasive electrode positioned at the back of the head, close to the occipital region of the brain. The recorded data can also include eye movement signals, and particularly EOG signals, received by a non-invasive electrode positioned close to the eye. The recorded data then undergoes pre-processing as part of the “Pre-processing” step 212. Pre-processing may involve filtering, artefact removal and noise reduction of the brain signals and eye movement signals to generate pre-processes signals. For example, filtering may be carried out using a Butterworth or Chebyshev infinite impulse response filter which are designed to keep only signals within a specific frequency band of interest (such as 6 Hz to 40 Hz). Further, power line noise at 50 Hz or 60 Hz (depending on the country) can be removed by designing a specific notch filter. Such pre-processing can enable important information, such as SSVEPs, to be more easily identified from the brain signals. Once the recorded data has been pre-processed, the method may progress to the second stage 202. Referring now to Figure 5, the second stage 202 of the method comprises the steps: “Double blink (DB) detector” 220 and “DB = True?” 222. In this embodiment of the invention the pre-processed eye movements signals are analysed by a double blink detection algorithm. However, in other embodiments of the invention the detection algorithm may be purposed with detecting a different predetermined eye gesture, such as a wink. The “Double blink (DB) detector” step 220 involves running the double blink detection algorithm to determine whether a pre-processed eye movement signal indicates that the user has performed a double blink, such as the eye movement signal 16 shown in Figure 12. If a double blink is detected then the double blink detection algorithm issues a DB signal. In embodiments of the invention, the purpose of the user performing a double blink, or other predetermined gesture, is to activate the display / graphical user interface (GUI) so that the user may interact with the BMI system to perform an action. The next step of the method is a decision step: “DB = True?” 222. This step involves determining whether a DB signal has been issued by the double blink detection algorithm as part of the “Double blink (DB) detector” step 220. If the determination is “Yes” - a DB signal has been issued - the method may progress to the third stage 203 wherein the BMI system display, or GUI, is activated. If the determination is “No” - a DB signal has not been issued - the “Double blink (DB) detector” step 220 is repeated and therefore this loop will repeat until a double blink is detected and a DB signal issued. Referring now to Figure 6, the third stage 203 is complex and comprises several steps with a number of different possible outcomes. Further, the third stage comprises a first path 207 and a second path 208 wherein the steps belonging to each path may be performed in parallel with one another. The second path 208 comprises an initial step, “GUI: Flicker mode” 260, which may be triggered by the DB signal issued in the second stage 202. “GUI: Flicker mode" 260 may involve activating the graphical user interface (GUI) which displays aspects of a BMI system application and, in particular, displays a plurality of labels and a plurality of stimuli to the user via the GUI wherein each stimulus is associated with one of the plurality of labels. Each stimulus may comprise an image or pattern that flickers between two colours. The two colours may be any two colours that are suitable to cause excitation of the user’s retina when flickering, for example the two colours may be contrasting colours such as black and white. Further, each stimulus may flicker at a specific frequency, within a range such as 6 Hz to 40 Hz or 6 Hz to 20 Hz (frequencies higher than 20 Hz may suffer from lower classification accuracy), and at a different phase to the other stimuli. Also, each stimulus may be of different maximum intensity and the intensity may temporally follow a square, sinusoidal, or other periodic waveform. The stimulus parameter for each stimulus may therefore comprise a combination of the above features such that each stimulus parameter is unique. If the user looks at a particular label displayed on the GUI, the associated stimulus flickering according to the unique stimulus parameter will cause excitation of the user’s retinas based on the combined features of that stimulus parameter. The first path 207 comprises an “SSVEP detection algorithm” step 230 which determines whether the data forming the pre-processed brain signal indicates that the user exhibited an SSVEP. The “SSVEP detection algorithm” 230 is carried out continuously and simultaneously to the “GUI: Flicker mode” 260. Therefore, if the user looks at a label displayed on the GUI and the user’s retinas are excited such that an SSVEP is generated, the “SSVEP detection algorithm” 230 will detect the SSVEP and determine a stimulus parameter signal from the SSVEP. The first path 207 further comprises a decision step: “SSVEP detected?” 232 which determines whether an SSVEP has been detected by the SSVEP detection algorithm. If the determination is “No” - an SSVEP has not been detected - the “SSVEP detection algorithm” step 230 is repeated and this loop will repeat until an SSVEP is detected and an SSVEP signal is issued. Further, the “GUI: Flicker mode” 260 will continue until an SSVEP is detected. If the determination is “Yes” - an SSVEP has been detected - then the resulting stimulus parameter signal may be analysed as part of a “Determine label L” step 236 wherein label L is the label which the user is looking at. This portion of the method allows the BMI system to be used asynchronously as it involves an idle state wherein the SSVEP detection algorithm is continuously operating to detect an SSVEP until an SSVEP is detected. The user may therefore interact with the BMI system, by looking at a label to generate an SSVEP, at any time and the system will respond accordingly. If the system does not detect an SSVEP, it will determine that the user does not want to perform an action and will simply continue attempting to detect an SSVEP until it is successful in doing so, or it is turned off manually. Each of the plurality of stimuli displayed via the GUI has a corresponding stimulus parameter, including features such as flicker frequency, phase and intensity, that define how the user’s retinas will be excited when they look at that stimulus. The variation in retina excitation caused by the different stimuli results in variation between the SSVEPs that are generated. The stimulus parameter signal determined by the SSVEP detection algorithm may be indicative of a hypothetical stimulus parameter that could have caused the SSVEP to be generated. In other words, the stimulus parameter signal may be indicative of the particular variation of SSVEP which is generated and, therefore, the stimulus parameter signal may be analysed to determine which of the plurality of stimuli most likely caused the SSVEP. Accordingly the “Determine label L” step 236 involves analysing the stimulus parameter signal to determine which of the plurality of stimuli is most likely to have induced the SSVEP and therefore which of the plurality of labels the user was looking at - label L. For example, Figure 13 shows an example of a brain signal 12, in this case formed of EEG data, that comprises a primary peak 304 at 10.25 Hz and auxiliary peaks 306 at multiples of 10.25 Hz (20.5 Hz and 30.75 Hz). The primary peak 304 and auxiliary peaks 306 are indicative of an SSVEP caused by a stimulus flickering with a frequency 10.25 Hz. The SSVEP detection algorithm may therefore determine a stimulus parameter signal comprising a flicker frequency of 10.25 Hz which would be compared against the stimulus parameters of the plurality of labels. In this example, the label with an associated stimulus parameter comprising a flicker frequency closest to 10.25 Hz may be determined as label L. (It is to be understood that this example represents a simplified version of the invention and that other features of the brain signal and stimulus parameters, such as phase and intensity, would be involved in the determination of label L.) Once label L has been determined, a “Communicate label L” step 261 forming part of the second path 207 may be performed wherein the determined label is communicated to the user so that the user may judge whether the determination is correct. The “Communicate label L” step 261 may involve causing the GUI to adjust what is displayed to the user such that it is clear to the user which label has been determined by the BMI system as the label the user selected. For example, the determined label L may be highlighted amongst the plurality of labels (such as with a change of colour, size or boldness) or all labels except the determined label L may be hidden. Alternatively, the “Communicate label L” step 261 may involve any suitable action to communicate which label has been determined to the user of the BMI system. For example, the action may be to play a sound such as a voice recording to indicate the determined label. Or, if the label is associated with an external device to be controlled then that external device may be caused to flash, beep or otherwise indicate that it is ready to be controlled, provided the determined label is correct. In this embodiment of the invention, if the user agrees that the determined label L is the label that they were looking at and intending to select then they are not required to take further action. However, if the user did not intend to select determined label L then they may perform a double blink. However, in other embodiments of the invention the double blink may be replaced by any suitable predetermined eye gesture, such as a wink. The first path 207 further comprises a “Double blink (DB) detector” step 238 which may be carried out in parallel with the “Communicate label L” step 261. Similarly to the “Double blink (DB) detector” step 220 shown in Figure 5, “Double blink (DB) detector" step 238 involves running a double blink detection algorithm to determine whether data recorded indicates that the user has performed a double blink. In this case, the data analysed may particularly be eye movement signals recorded since the label L was first communicated to the user. If the double blink detection algorithm detects data indicative of a double blink from the recorded eye movement signals, it issues a DB signal representative of the relevant data. In other embodiments of the invention the double blink detection algorithm may be replaced by any suitable algorithm for detecting a predetermined eye gesture from recorded data. Similarly, in other embodiments of the invention, the “Double blink (DB) detector” step 238 may be adapted for detection of any suitable predetermined eye gesture. Following the “Double blink (DB) detector”, the first path 207 further comprises a decision step: “DB = True?” 240. This step involves determining whether a DB signal has been issued by the double blink detection algorithm as part of the “Double blink (DB) detector” step 238. If the determination is “Yes” - a DB signal has been issued - the method may progress to an “L = Null” step 242. If the determination is “No” - a DB signal has not been issued - the method may simultaneously progress to the fourth stage 204 (for more information refer to Figure 7) and a decision step “L = Quit?” 244. As set out above, “L = Null” step 242 is reached when a DB signal is issued due to a double blink being detected and “L = Null” 242 is therefore associated with an indication that the user judged the determined label L as being incorrect, i.e. not what the user intended to choose. Hence, it can be assumed that the user does not want the action associated with the incorrectly determined label L to be performed. Accordingly, the “L = Null” step 242 may involve issuing a nullify signal which prevents the action associated with the incorrectly determined label from being performed and influences an “L = Null?” decision step 262. The “L = Null?” decision step 262 forms part of the second path 208 and involves determining whether the nullify signal has been issued. If the determination is “Yes” - a nullify signal has been issued - the method may progress to the fifth stage 205 (for more information refer to Figure 8). If the determination is “No” - a nullify signal has not been issued - the method may progress to a decision step “Off = True?” 264. The “L = Quit?” step 244 is reached when no DB signal is issued due to no double blink being detected and “L = Quit” 244 may therefore be associated with an indication that the user judged the determined label L as being correct, i.e. the user intended to choose the determined label L. Hence it can be assumed that the user wants the action associated with correctly determined label L to be performed. The “L = Quit?” step 244 involves determining whether the correctly determined label is associated with an action to quit the GUI application. If the determination is “Yes” - the label L is associated with quitting the GUI application -the method may progress to an “Off = True” step 248. If the determination is “No” - the label L is not associated with quitting the GUI application - the method may progress to a “Perform action related to L” step 246. The “Off = True” step 248 involves issuing an off signal which causes the method to return to the second stage 202 and repeat the “Double blink (DB) detector” step 220, and also influences an “Off = True?” decision step 264. The “Off = True?” decision step 264 involves determining whether the off signal has been issued during the Off = True” step 248. If the determination is “Yes” - an off signal has been issued - the method may progress to a “GUI: Off mode” step 266 which may involve switching the GUI display into an off / standby mode. (The user may then re-activate the GUI display by performing a double blink in accordance with the second stage 202 of the method.) If the determination is “No” - an off signal has not been issued - the method may progress to a “GUI: Feedback mode” 268 described in further detail below. Now referring back to the “Perform action related to L” step 246, this involves issuing a command for the action, associated with the label L, to be performed. For example, the action may be changing the position of a motorized bed, adjusting the temperature setting of an air conditioning system or turning on a television. Once the action has been performed the “SSVEP detection algorithm” step 230 is repeated so that the user can select another label in order to trigger a corresponding command. In parallel to the “Perform action related to L” step 246, the “GUI: feedback mode” step 268 may also be performed as set out above. The “GUI: feedback mode” step 268 involves causing the GUI to display feedback to the user, for example the GUI might display a graphic of the new motorized bed position or the new temperature setting that the air conditioning system has been set to. Once the feedback has been displayed for a predetermined period of time, 2 seconds for example, the method repeats the “GUI: Flicker mode” step 260 so that the user is able to look at another label in order to issue a new command by triggering an SSVEP which may be detected by the SSVEP detection algorithm as part of the “SSVEP detection algorithm” step 230. Referring now to Figure 7, the fourth stage 204 of the method comprises steps that allow the BMI system to learn from the way that the user interacts with it. If a label L has been determined according to the third stage 203 of the method and confirmed by the user via decision step “DB = True?” 240, the stimulus parameter associated with the label L are proved to be successful in causing an excitation of the user’s retinas sufficient to generate a detectable SSVEP from which stimulus features may be accurately determined. In order that the BMI system may learn from this event and hopefully repeat its success in the future, when a label L has been correctly determined (the determination forming part of the “DB = True?” step 240 is “No”) a next step of the method is the “Train / Re-Train with new data” step 270. The “new data” may be a correct selection data set and may include the brain signal, the determined stimulus parameter signal and the stimulus parameter of the stimulus associated with label L. The method step itself may involve using the new data stored in memory in combination with existing data (i.e. previously stored correct selection data sets), if it exists, in order to train / re-train the SSVEP detection algorithm and improve its performance in classifying features of brain signals in relation to stimulus parameters. This is needed as SSVEP detection algorithms are known to perform better if they have examples of brain signals (such as EEG data) corresponding to each stimulus parameter for each specific user. That way the algorithm will know what is the expected SSVEP response in each case and be able to identify similarities across responses to learn common activity and remove it, hence enhancing the relative differences across the different responses and obtain higher classification performance. The next step of the method is “Update stimuli parameters if necessary” 272 before the method 202 reaches its end point 206, marking the completion of one training cycle. As the user continues to use the BMI system and completes more training cycles, the BMI system will be able to collect more data and improve the classification performance of the SSVEP detection algorithm. If certain stimuli are being classified poorly, then the stimulus parameters associated with those stimuli may be updated, as part of the step “Update stimuli parameters if necessary” 272, to increase the chance of correct labelling. For example, if a specific stimulus is often misclassified (absent from the correct selection data sets and / or associated with repeated cancel eye gestures) then the stimulus parameter may be updated to comprise a frequency that has been proved to be successful in relation to a different stimulus, but with a different phase. Other updates may also be implemented, such as changing the colours used or the intensity of the flickering. Further, aspects such as the organisation of stimuli may be updated in order to ensure that two or more stimuli with similar frequencies are not positioned next to one another on the display to reduce the risk of misclassification between them. Referring now to Figure 8, a fifth stage 205 of the method comprises steps that allow the BMI system to adjust its approach in response to determining label L incorrectly. If a label L has been determined by the BMI system and cancelled by the user via the decision step “DB = True?” 240, the parameters of the stimulus associated with the label which the user intended to select may be unsuitable. Further, if the selection process is repeated then the user will likely want to select the same label that they attempted to, unsuccessfully, in the previous trial. If this is the case and a new trial is started with the same stimuli associated with the same labels as the previous trial then it is likely that the user will again be unsuccessful in selecting the desired label. To avoid this situation, when a label L has been nullified by the system following a double blink from the user, the “Rearrange stimuli” step 280 may be performed wherein the stimuli are rearranged so that each label is associated with a stimulus that is different to the stimulus it was associated with in the previous trial. As a result, the user will hopefully be able to successfully select a desired label on the second time of trying, if not the first. Further, as the user performs more trials, the likelihood of stimuli with unsuitable parameters will be reduced by the learning method steps set out in Figure 6 and the likelihood of unsuccessful label selections will also reduce. For demonstration, Figure 9 shows an initial display 20a (demonstrating what may be shown before the “Rearrange stimuli” step 280 has been carried out) and a rearranged display 20b (demonstrating what may be shown after the “Rearrange stimuli” step 280 has been carried out). The initial display 20a comprises a plurality of labels 22 wherein a first label 22a is associated with a function F_A and a stimulus S1, a second label 22b is associated with a function F_B and a stimulus S2, and so on. The rearranged display 20b comprises the same plurality of labels except the plurality of stimuli S1, S2, S3, S4, S5, S6 have been rearranged (rotated anticlockwise in this case) such that the first label 22a is still associated with function F_A but is now associated with stimulus S2 instead of S1. Similarly, the second label is still associated with function F_B but is now associated with stimulus S3 instead of S2, and so on such that each label 22a, 22b, 22c, 22d, 22e, 22f is associated with a different stimulus S1, S2, S3, S4, S5, S6 in the rearranged display 20b than it was in the initial display 20a. As well as meaning that no label is associated with the same stimulus after an incorrect determination of a label selected by the user, the method of stimuli rotation set out above allows labels associated with certain functions to occupy a consistent part of the screen, while the stimuli are shifted. This means that a user can become familiar with the information that is displayed and can pre-empt where certain labels will appear, resulting in faster operation of the BMI system. Referring now to Figure 10, an alternative third stage 203b is shown which is a modified version of the third stage 203 shown in Figure 6. The difference between the two stages is that the alternate third stage 203b comprises additional gaze tracking steps 250. The gaze tracking steps 250 are performed prior to “GUI: Flicker mode” 260 and comprise an “Identify point of gaze” step 252 wherein recorded data, such as recorded eye movement signals, is analysed to determine the direction that the user is gazing and calculate a location on the GUI display that the user is looking at or, in other words, a point of gaze. The calculated point of gaze is then used during a “Select region of interest” step 254 which determines a region of the GUI display that has the point of gaze at its centre and further determined which of the plurality of labels are positioned within that region. The “GUI: Flicker mode” step 260 is then carried out. However, the number of stimuli displayed is limited to the number of labels which are positioned within the region of interest. By doing this, the number of flickering stimuli that need to be displayed at any one time is reduced. This has two advantages, the first is a reduction in irritation to the user that can be caused by looking at a screen with several conflicting stimuli being displayed. The second advantage, particularly after the user has completed a number of label selections, is that reducing the number of stimuli shown at one time allows the BMI system to cherry pick the most suitable stimuli parameters for that user and thereby improve the accuracy of the BMI system. Referring now to Figure 11, a further alternative third stage 203c is shown. The difference between the third stage 203 shown in Figure 6 and this alternative third stage 203c is that the decision step: “SSVEP detected?” 232 of Figure 6 has been removed. This means that the “SSVEP detection algorithm” step 230 detect an SSVEP from the data forming the pre-processed brain signal and determines a stimulus parameter signal from the SSVEP (irrespective of whether the user actually exhibited an SSVEP). The “SSVEP detection algorithm” 230 is carried out after the plurality of labels and stimuli are displayed to the user as part of the “GUI: Flicker mode” step 260. This portion of the method means that the BMI system operates synchronously, rather than asynchronously according to the third stage 203 shown in Figure 6, wherein the SSVEP detection algorithm is used to assess the user’s brain signals recorded over distinct predetermined periods of time. The user may therefore interact with the BMI system, by looking at a label to generate an SSVEP, within predefined time windows only and the system will respond accordingly. If the user does not wish to trigger an action to be performed, he / she is required to select the label associated with switching the BMI system to the standby mode. Once the stimulus parameter signal is determined during the “SSVEP detection algorithm” step 230, the “Determine label L” step 236 involves analysing the stimulus parameter signal to determine which of the plurality of stimuli is most likely to have induced the SSVEP and therefore which of the plurality of labels the user was looking at - label L. The method may then progress according to the remaining steps of the alternative third stage 203c, which correspond to the steps shown in Figure 6. Options for a given aspect, feature or parameter of the invention should, unless the context indicates otherwise, be regarded as having been disclosed in combination with any and all preferences and options for all other aspects, features and parameters of the invention. For example, the first and second stages 201 and 202 of a method according to an embodiment of the invention which are shown in Figures 4 and 5 respectively may be performed prior to the method 102 according to another embodiment of the invention shown in Figure 3. The listing or discussion of an apparently prior published document in this specification should not necessarily be taken as an acknowledgement that the document is part of the state of the art or is common general knowledge.

Claims

1. A method for performing an action based on a user’s brain activity comprising the steps of:(a) displaying a plurality of labels;(b) displaying a plurality of stimuli wherein each stimulus corresponds to a stimulus parameter and is associated with a different label;(c) receiving a brain signal while the user is exposed to the plurality of stimuli;(d) applying a steady state visually evoked potential (SSVEP) detection algorithm to the brain signal in order to detect an SSVEP and determine a stimulus parameter signal from the SSVEP;(e) determining which one of the plurality of stimuli corresponds to a stimulus parameter which is closest to the determined stimulus parameter signal and identifying the label associated with that stimulus as a selected label;(f) indicating the selected label to the user;(g) receiving an eye movement signal while the user is exposed to the indication of the selected label; and(h) determining whether the recorded eye movement signal corresponds to a predetermined cancel eye gesture, wherein:if a cancel eye gesture is determined, performing the further steps of: adding one tally to a counted number of consecutive cancel eye gestures and repeating steps (a) to (h);if a cancel eye gesture is not determined, performing the further step of: performing the action corresponding to the selected label, resetting the counted number of consecutive cancel eye gestures to zero and storing in memory a correct selection data set comprising the brain signal and the stimulus parameter of the stimulus associated with the selected label;wherein, following the step of adding one tally to the counted number of consecutive cancel eye gestures, the method comprises the further steps of:determining whether the counted number is equal to a predetermined rearrangement number wherein:if the counted number is less than the predetermined rearrangement number, proceeding with repeating steps (a) to (h); orif the counted number is equal to the predetermined rearrangement number, rearranging the plurality of stimuli such that each stimulus is associated with a different label, resetting the counted number to zero and proceeding with repeating steps (a) to (h).

2. A method according to claim 1, wherein the method comprises the further step of: using a correct selection data set to train a training-based SSVEP detection algorithm.

3. A method according to claim 1 or claim 2 wherein the correct selection data set further comprises one or more of the determined stimulus parameter signal, the selected label, the arrangement of the plurality of labels at the time the brain signal was received and the arrangement of the plurality of stimuli at the time the brain signal was received.

4. A method according to any preceding claim wherein if the plurality of stimuli are rearranged and there is no cancel eye gesture determined with respect to a selected label determined from a successively received brain signal, the method comprises the further steps of:identifying that the stimulus parameter associated with the selected label before the rearrangement of the plurality of stimuli is problematic;storing in memory a record of the problematic stimulus parameter; and counting the number of times a stimulus parameter is identified as problematic.

5. A method according to claim 4, wherein, if a stimulus parameter is identified as problematic a predetermined number of times, the method comprises the further step of updating the problematic stimulus parameter.

6. A method according to any preceding claim comprising the further steps, prior to step (b), of:tracking the user’s point of gaze;determining a region of interest based on the point of gaze; andlimiting the number of stimuli that are displayed such that the number of stimuli corresponds to the number of labels positioned within the region of interest only, wherein each stimulus displayed in step (b) is associated with a different label within the region of interest.

7. A method according to claim 6 wherein the step of tracking the user’s point of gaze comprises the steps of:receiving a gaze signal comprising electroencephalography (EEG) data, electrooculography (EOG) data and / or video-oculography (VOG) data; anddetermining a point of gaze of the user based on the gaze signal.

8. A method according to any preceding claim comprising the further steps, prior to step (a), of:continuously recording eye movement signals from the user;determining whether a recorded eye movement signal corresponds to a predetermined activate eye gesture; andif an activate eye gesture is determined, activating the display and performing step (a); orif an activate eye gesture is not determined, repeating the activate eye gesture determining step.

9. A method according to claim 8 wherein; if the selected label determined in step (e) corresponds to an “off’ action and a cancel eye gesture is not determined in step (h); then the method comprises the further steps of:deactivating the display; and repeating the steps of claim 8.

10. A method according to any preceding claim wherein: if the selected label determined in step (e) corresponds to an application action and the cancel eye gesture is not determined in step (h); then the step of performing an action corresponding to the selected label comprises issuing a command signal representing a command to perform the application action.

Images