FIELD OF THE INVENTION

The present invention relates to a command detection device and method for detecting a command selected by a user by clenching teeth. The present invention further relates to a command detection assembly comprising such device and a computer program for implementing such method.

BACKGROUND OF THE INVENTION

Conventional command detection and/or control devices are based on conventional communication that can be observed visually (e.g. through gestures) or auditory (e.g. through speech). However, in some cases communication that cannot be visually or auditorily observed can be desirable, thus a communication that ensures secrecy of the communication (inconspicuous communication).

One example of such non-conventional communication is the usage of subtle muscle movement, also called electromyography (EMG), for communication, in particular for command detection and/control. One possibility of such EMG communication is using teeth clenching. When the user clenches his/her teeth, electrical activity produced by muscle tension in a specific area of the head of the user can be measured by means of an

electromyogram (EMG).

US 7,783,391 B2 discloses an apparatus for controlling a vehicle by a teeth clenching motion made by a user. The apparatus comprises an electromyogram signal obtaining unit including electromyogram sensors disposed at both sides for generating an electromyogram signal according to a predetermined muscle moved when a disabled person clenches teeth, and a ground electrode connected to a body of the disabled persons for providing a reference voltage. The apparatus further comprises a vehicle driving unit including a control command generating unit for generating a vehicle driving command according to the electromyogram signal by classifying the electromyogram signal based on a side of teeth clenched, a duration time for clenching teeth and a sequence of teeth clenching motions made by the disabled person, a control command interface for generating a predetermined level of voltage according to the created vehicle driving command, and a vehicle driving unit for driving the vehicle according to the generated voltage. The control command generating unit is configured to divide the electromyogram signal by a

predetermined time, and to obtain a difference of absolute means value (DAMV) of each channel as feature value from the divided electromyogram signals obtained from two channels. The control command generating unit is configured to classify the electromyogram signals into four basic patterns by comparing the obtained feature values of the

electromyogram signals with the predetermined threshold value, classifying the

electromyogram signals into an ON-state if the feature value is greater than the

predetermined threshold value and classifying the electromyogram signals into an OFF-state if the feature value is smaller than the predetermined threshold value. The control command generating unit is configured to generate the vehicle driving command based on a classified pattern by setting a previous motion of vehicle as a reference value and analyzing the classified pattern based on the reference value.

The problem with such apparatus is that due to the usage of only four basic patterns, the complex way how control commands are generated and/or the required relation to the previous state, only low communication speed can be achieved or the communication speed might not be sufficient.

SUMMARY OF THE INVENTION

It is an object of the present invention to provide a command detection device and method with increased communication speed. It is a further object of the present invention to provide a command detection assembly comprising such command detection device and a computer program for implementing such method.

In a first aspect of the present invention a command detection device for detecting a command selected by a user by clenching teeth is presented, the device comprising an interface for receiving an electromyogram signal indicative of electrical activity produced by muscle tension, caused by teeth clenching of the user, in a specific area of the head of the user. The device further comprises a signal processing unit for processing the electromyogram signal. The signal processing unit is configured to provide a time window of the electromyogram signal, wherein a size of the time window is sufficiently large to detect at least two consecutive teeth clenching events in the time window of the electromyogram signal, each of the teeth clenching events indicating a single teeth clenching action by the user, detect at least two sub-segments in the time window, each sub-segment indicating one of the at least two teeth clenching events, and detect the command selected by the user based on the at least two sub-segments in the time window.

In a further aspect of the present invention a command detection assembly for detecting a command selected by a user by clenching teeth is presented, the assembly comprising such command detection device The assembly further comprises a wearable device wearable by the user comprising at least one electromyogram sensor providing at least one electromyogram signals.

In a further aspect of the present invention a method for detecting a command selected by a user by clenching teeth is presented, the method comprising receiving an electromyogram signal indicative of electrical activity produced by muscle tension, caused by teeth clenching of the user, in a specific area of the head of the user. The method further comprises processing the electromyogram signal, the processing comprising providing a time window of the electromyogram signal, wherein a size of the time window is sufficiently large to detect at least two consecutive teeth clenching events in the time window of the electromyogram signal, each of the teeth clenching events indicating a single teeth clenching action by the user. The method further comprises detecting at least two sub-segments in the time window, each sub-segment indicating one of the at least two teeth clenching events, and detecting the command selected by the user based on the at least two sub-segments in the time window.

In a further aspect of the present invention a computer program is presented comprising program code means for causing a computer to carry out the steps of such method when said computer program is carried out on the computer.

The basic idea of the invention is to provide a time window of the electromyogram signal (electromyogram values over time), wherein the size of the time window is sufficiently large to detect at least two consecutive teeth clenching events in the time window of the electromyogram signal. Each of the teeth clenching events indicates a single teeth clenching action by the user. At least two sub-segments can be detected in the time window, each sub-segment indicating one of the at least two teeth clenching events. The command selected by the user is then detected based on the at least two sub-segments in the time window. The term command and symbol can be used interchangeably. A command or symbol represents an elementary communication measurement unit. Since at least two consecutive teeth clenching events (at least "double" teeth clenching) are detectable in the time window, communication speed is increased. The basic idea of the present invention is thus that a teeth clenching combination or sequence of at least two consecutive teeth clenching events or sub-segments is used in one time window. For example, also at least three consecutive teeth clenching events (at least "triple" teeth clenching) can be used.

Further, for example, additionally the duration of the sub-segment in the time window, and/or other parameters used for the clenching command or symbol detection can be used. In this way a very high communication speed is enabled.

Preferred embodiments of the invention are defined in the dependent claims. It shall be understood that the claimed method, assembly or computer program has similar and/or identical preferred embodiments as the claimed command detection device and as defined in the dependent claims.

In one embodiment each teeth clenching event is a teeth clenching event selected from the group comprising at least a first teeth clenching event indicating a teeth clenching action of a first time duration and a second teeth clenching event indicating a teeth clenching action of a second time duration, the second time duration being greater than the first time duration. In this embodiment a distinction or classification based on the duration of a sub-segment, thus teeth clenching event, can be made. In this way it can be distinguished or classified between short teeth clenching event (first teeth clenching event) and long teeth clenching event (second teeth clenching event).

In a variant of this embodiment the signal processing unit is configured to determine the duration of each sub-segment and to determine if the sub-segment is of the first or the second duration. In this way, a distinction or classification based on the sub-segment duration can be made, and the command can be detected also based on the duration of the sub-segment (e.g. short or long). For example, it can be determined that the sub-segment is of the first time duration if the duration of sub-segment is below a predefined threshold, and it can be determined that the sub-segment is of a second time duration if the time duration of the sub-segment is above the predefined threshold. For example, the predefined threshold can be 0.5 seconds. This enables to distinguish between short and long teeth clenching events.

In a further variant of this embodiment the size of the time window is sufficiently large to detect at least two consecutive teeth clenching events of the first time duration. In this way at least two consecutive short teeth clenching events ("double" short teeth clenching) can be detected in one time window, because the time window is sufficiently large.

In a further variant of this embodiment the group further comprises at least a third teeth clenching event indicating a teeth clenching action of a third time duration and/or a fourth teeth clenching event indicating a teeth clenching action of a fourth time duration, the third time duration being smaller than the first time duration and/or the fourth time duration being greater than the third time duration. In this way a classification not only into a short teeth clenching event and a long teeth clenching event is made, but additionally other classification types, such as a very short teeth clenching event and/or a very long teeth clenching event, are added. This further increases the communication speed. For example, a very short (or brief) teeth clenching event can be determined if the time duration of the sub- segment is below 0.3 s, a short teeth clenching event when the time duration of the sub- segment is between 0.3 and 0.5 s, a long clenching event if the time duration is between 0.5 and 0.8 s, and a very long teeth clenching event if the time duration is greater than 0.8 s.

In a further embodiment the size of the time window is between 0.6 and 1.6 s, in particular about 1 s. The size of this time window is sufficiently large to detect at least two consecutive teeth clenching events in the time window, in particular at least two consecutive (short) teeth clenching events of the first (short) time duration.

In a further embodiment the signal processing unit is further configured to determine a spectral power of the electromyogram signal in the time window, and to detect one of the sub-segments based on the spectral power. In this way the sub-segment indicative of the teeth clenching event can be reliably detected. The spectral power can in particular be the spectral power density (PSD).

In a variant of this embodiment the signal processing unit is configured to provide the time window at a start time of the time -window when the spectral power exceeds a predefined threshold. In this variant the time window is provided only when a teeth clenching event will be detected. This reduces computing time.

In a further variant of this embodiment the signal processing unit is configured to detect one of the sub-segments between a start time of the sub-segment when the spectral power exceeds a predefined threshold and an end time of the sub-segment when the spectral power falls below the predefined threshold. In this variant a reliable detection of the duration of the sub-segment is provided.

In a further embodiment the signal processing unit is configured to detect the command based on at least one feature selected from the group comprising a total amount of spectral power in the time window, a duration of at least one or each of the sub-segments, a duration between the two consecutive sub-segments, a number of sub-segments, a start time of at least one or each of the sub-segments, and an end time of at least one or each of the sub- segments. This enables a combination of different classification types of clenching events in a single time window. Consequently, the number of commands or symbols in a time window can be increased.

In a further embodiment the interface can receive a first electromyogram signal from a first electromyogram sensor, the first electromyogram signal indicative of electrical activity produced by muscle tension caused by teeth clenching on the right side of the head of the user, and a second electromyogram signal from a second electromyogram sensor, the second electromyogram signal indicative of electrical activity produced by muscle tension caused by teeth clenching on the left side of the head of the user. In this embodiment a distinction or classification between a right teeth clenching event or a left teeth clenching event or both can be made.

In a variant of this embodiment the processing unit is configured to process the first electromyogram signal and the second electromyogram signal independently from each other, and to detect the command selected by the user based on the sub-segments determined in both the first and the second electromyogram signal. In this variant the communication can be further increased by detecting sub-segments in both the first (right) and the second (left) electromyogram signal.

In a further embodiment the at least one electromyogram sensor comprises an array of dry pin electrodes. Since dry electrodes do not require any preparation and cleaning they are superior to traditional wet electrodes. Additionally, the pin-structure of the electrodes enables the penetration through the hair and/or positioning of the dry electrodes on an arbitrary location on the user's head. Thus, the performance is further increased. Also, the electrodes can be seamlessly integrated into the wearable device, such as for example headphones or a headset.

In a further embodiment interface is further adapted for receiving an electroencephalogram signal indicative of electrical brain activity of the user, and the signal processing unit configured for processing the electroencephalogram signal. In this way, additionally to the electromyogram signal (EMG signal), also an electroencephalogram signal (EEG signal) can be received and processed. Thus, a combination of EMG measurement and EEG measurement is achieved. Such combination can yield better performance and/or higher stability.

In a variant of this embodiment, the signal processing unit is configured to detect the command selected by the user also based on the electroencephalogram signal. In this way, the communication and/or control can be complemented. In another variant of this embodiment, the signal processing unit is configured to monitor brain activity of the user based on the electroencephalogram signal. In this way, additionally the brain activity of the user can be monitored. BRIEF DESCRIPTION OF THE DRAWINGS

These and other aspects of the invention will be apparent from and elucidated with reference to the embodiment(s) described hereinafter. In the following drawings

Fig. 1 shows a schematic block diagram of a command detection assembly according to an embodiment,

Fig. 2 shows a first (right) electromyogram signal,

Fig. 3 shows a second (left) electromyogram signal,

Fig. 4a-c each show an electromyogram signal of two consecutive short teeth clenching events on the right side,

Fig. 5a-c each show an electromyogram signal of two consecutive short teeth clenching events on the left side,

Fig. 6a-c each show an electromyogram signal of two consecutive short teeth clenching event on both the right and the left side,

Fig. 7a-c each show an electromyogram signal of a single short teeth clenching event on the right side,

Fig. 8a-c each show an electromyogram signal of a single short teeth clenching event on the left side,

Fig. 9a-c each show an electromyogram signal of a single short teeth clenching event on both the right and left side,

Fig. lOa-c each show an electromyogram signal of a single long teeth clenching event on the right side,

Fig. 1 la-c each show an electromyogram signal of a single long teeth clenching event on the left side,

Fig. 12a-c each show a electromyogram signal of a single long teeth clenching event on both the right and left side, and

Fig. 13 shows a front view of a wearable device according to an embodiment.

DETAILED DESCRIPTION OF THE INVENTION

Fig. 1 shows a schematic block diagram of a command detection assembly 30 according to one embodiment. The command detection assembly 30 comprises a command detection device 10 for detecting a command selected by a user by clenching teeth, and a wearable device 20 wearable by the user. The wearable device 20 comprises at least one, in particular at least two, electromyogram sensors 40a, 40b, each providing an electromyogram signal. Each of the electromyogram sensors can comprise an array of dry electrodes. A first electromyogram sensor 40a provides a first electromyogram signal indicative of electrical activity produced by muscle tension caused by teeth clenching on the right side of the head of the user, and a second electromyogram sensor 40b provides a second electromyogram signal indicative of electrical activity produced by muscle tension caused by teeth clenching on the left side of the head of the user. In particular, the wearable device can comprises four electromyogram sensors, the first and second electromyogram sensors 40a, 40b or its electrodes for providing the electromyogram signals, a third electromyogram sensor 40c or electrodes as reference sensor or electrodes, and a fourth electromyogram sensor 40d as ground sensor or electrodes.

The electromyogram sensors 40a, 40b can also be used to detect an electroencephalogram (EEG) signal. In particular, dry electrodes and the accompanying electronic circuits (EEG sensor) designed to detect an electrical brain signal, i.e. an electroencephalogram (EEG), having an amplitude in the order of up to ΙΟΟμν, can also be used to detect the electromyogram (EMG) signal stemming from muscle movements on the head. The EMG signal is in the frequency range of EEG (brain) signals (i.e. covering the frequency range of 10 tolOOHz), but the amplitude of the EMG signal is much greater (in the order of few mV) than the amplitude of the EEG signal. The EMG signal has a quite distinctive spectral content than can be distinguished from the EEG spectral content, e.g. using a suitable method such as independent component analysis, canonical correlation analysis, or the like. For example, the EMG signal or peaks can be extracted or removed from the EEG signal. Also, for example, at least one time segment that contains an EMG signal or peaks can be extracted or removed from the signal for EEG analysis, in order to prevent contamination between EEG analysis and EMG analysis. Thus, the command detection device 10 can perform EMG analysis if the presence of an EMG signal is detected, and can also analyze the spectral content and the features of the EEG signal.

To reliably monitor the EMG signal produced by teeth (i.e., jaw) clenching on the left and/or right side of the head, while also monitoring the hemispheric EEG activity, at least four sensors or electrode arrays can be used (minimum electrode configuration). Two sensors (first and second sensors 40a, 40b in Fig. 1) of the four sensors or electrode arrays are positioned at the contralateral sides of the head such that they acquire the teeth clenching action or activity (e.g., around temporal position on the head), one sensor (third sensor 40d in Fig. 1) at a position that has less or no influence of teeth clenching action (reference) and one sensor (fourth sensor 40d in Fig. 1) used as ground electrode positioned further away from measuring sensors or electrode arrays.

The number of sensors or arrays of dry electrodes are integrated into the wearable device 20, which is a headset or headphones in the embodiment of Fig. 1. The wearable device 20 further comprises a hardware component 24 for signal amplification (including battery) and a transmission unit 22 for wirelessly transmitting the sensor data (EMG signal). In the embodiment of Fig. 1, the command detection device 10 is a separate device comprising a receiver unit 12 for wirelessly receiving the sensor data (EMG signal) transmitted by the transmitter unit 22 of the wearable device 20. In the embodiment of Fig. 1, the receiver element 12 functions as an interface for receiving the EMG signal indicative of electrical activity produced by muscle tension, caused by teeth clenching of the user, in a specific area of the head of the user. The command detection device 10 further comprises a signal processing unit 14 (e.g. processor or microprocessor) for processing the EMG signal (and/or the EEG signal). It will be understood that the signal processing unit 14 can also be formed by a plurality of signal processing units (e.g. processors or microprocessors). The command detection device 10 further comprises a command execution unit 16 for executing the detected command. The command execution unit 16 can for example be a display which displays the command or an actuator for actuating an element. In the embodiment of Fig. 1, the command execution unit 10 is part of the command detection device 10. Alternatively, the command execution unit 10 can also be a separate device comprising a receiver unit 12 for wirelessly receiving the command.

In the embodiment of Fig. 1, the wearable device 20 (e.g. headset or headphones) does not include the command detection device or signal processing unit for detecting a command, but just the sensor part (the part that is responsible for acquiring the EEG/EMG signal and transmitting it to the separate external command detection device 20). In an alternative embodiment, the command detection device or signal processing unit for detecting a command can be integrated into the wearable device 20 (e.g. headset or headphones). In this case the commands or symbols are extracted directly within the signal processing unit integrated in the wearable device 20 and the commands are transmitted directly to a separate external command execution device that receives and executes the commands. Fig. 2 shows a first (right) electromyogram signal and Fig. 3 shows a second (left) electromyogram signal. In each of Fig. 2 and Fig. 3 a spectral power (EMG power) of the electromyogram is determined, in particular a spectral power density (PSD). The spectral power density can be estimated using the Welch method, in particular (short time) Fourier transform. The spectral power can be computed in a selected frequency domain (e.g. 12 to 100 Hz). Fig. 2 shows the spectral power of the first (right) electromyogram signal (indicated by circles) received from the first (right) electromyogram sensor 40a in Fig. 1, the signal being indicative of electrical activity produced by muscle tension caused by teeth clenching on the right side of the head of the user. Fig. 3 shows the spectral power of the second (left) electromyogram signal (indicated by crosses) received from the second (left) electromyogram sensor 40b of Fig. 1, the signal being indicative of electrical activity produced by muscle tension caused by teeth clenching on the left side of the head of the user.

In each of Fig. 2 and Fig. 3 a time window of size T of the electromyogram signal is provided. The size T of the time window is sufficiently large to detect at least two consecutive teeth clenching events, also called "double" teeth clenching, in the time window of the electromyogram signal. Each of the teeth clenching events indicates a single teeth clenching action by the user. At least two sub-segments (s _al , s _a2 in Fig. 2, or SM , Sb ₂ in Fig. 3) are detected in the time window of size T. Each sub-segment (s _al , s _a2 in Fig. 2, or SM , Sb ₂ in Fig. 3) indicates one of the at least two teeth clenching events ("double" teeth clenching). The command selected by the user is detected based on the at least two sub-segments segment s _al , s _a2 or SM , Sb ₂ in the time window of size T.

In each of the examples of Fig. 2 and Fig. 3, a sub-segment s _al , s _a2 or SM , Sb ₂ is detected based on the spectral power (EMG power) of the EMG signal. The first (right) electromyogram signal (see e.g. example of Fig. 2) and the second (left) electromyogram signal (see e.g. example of Fig. 3) are processed independently from each other (also called right channel and left channel). However, the command selected by the user is then detected based on the sub-segments determined in both the first (right) and the second (left) electromyogram signal.

As can be seen in each of Fig. 2 and Fig. 3 the time window of size T is provided at a start time T _star t of the time window when the spectral power exceeds a predefined threshold TH (TH _a in Fig. 2 or T¾ in Fig. 3). One of the sub-segments s _al , s _a2 or Sbi , Sb ₂ is detected between a start time t _star t of the sub-segment when the spectral power exceeds a predefined threshold, in particular the predefined threshold TH _a or T¾ mentioned above, and an end time t _en d of the sub-segment when the spectral power falls below the predefined threshold. In this way also the duration t of the sub-segment can be determined, in particular between the start time t _star t and the end time t _en d of the sub-segment (duration t _al between t _star tai and t _en dai for the first sub-segment s _al of Fig. 2 and duration t _a2 between t _star ta2 and t _en da2 for the second sub-segment s _a2 of Fig. 2; duration tbi between t _star tbi and t _en dbi for the first sub-segment SM of Fig. 3 and duration ½ between t _s tartb2 and t _en db2 for the second sub- segment Sb2 of Fig.3). Further, the duration t; between the two consecutive sub-segments can be determined, in particular between the end time of the first sub-segment and the start time of the second, consecutive sub-segment (between t _en dai and t _sta rta2 in Fig. 2; between t _en dbi and tstartb2 in Fig. 3).

The command can be detected based on at least one feature selected from the group comprising a total amount of spectral power in the time window, a duration of at least one or each of the sub-segments s _al , s _a2 , SM , Sb2, a duration t; between the two consecutive sub-segments, a number of sub-segments N, a start time t _star t of at least one or each of the sub-segments, and an end time t _en d of at least one or each of the sub-segments.

When comparing the start time t _star t of one of the sub-segments (either s _al or s _a2 ) of the first (right) EMG signal in Fig. 2 with the start time t _star t of the corresponding sub- segment (either s _al or s _a2 ) of the second (left) EMG signal in Fig. 3, it can be determined that the corresponding right clenching event and the corresponding left clenching event are synchronized. If the right clenching event and the left clenching event are synchronized, a clenching event on both the right and left side is present (both sides). In particular, it can be determined that the clenching events are synchronized when the time between the two start times is below a predefined threshold (e.g. a threshold between 30 and 70 ms, in particular about 50ms). As an example, since the time between the start time t _star t _a iin Fig. 2 and the start time tstartbi in Fig. 3 is below a threshold of 50ms (in other words within 50ms), the clenching events are synchronized (clenching on both the right and left side). However, if the time between the start time t _s tartai in Fig. 2 and the start time t _s tartbi in Fig. 3 would be much greater than the threshold of 50ms (e.g. 100ms), it would be determined that a left clenching event followed by a right clenching event, or the other way round, is present.

The signal processing unit is responsible for determining the spectral power of the signal and for detecting the command/symbol issued by the user. The detection of the spectral power parameters for a single time window is exemplary illustrated in each of Fig. 2 and Fig. 3. The detection can be based on one or more (or all) of the following features or parameters: First, the total amount of spectral power above the threshold (indicated by the hatched area in each of Fig. 2 and Fig. 3) for each or both channels for the duration of the time window, (which is about O. lmV ² s in Fig. 2 or Fig. 3; size of the area above the threshold TH and below the EMG signal or power line/curve on a time segment of Is), which is the indication that there is either one long clenching event or at least two short ones.

Second, the number of segments where the spectral power is above the threshold, and optionally where the duration of each sub-segment is larger than a predefined minimal segment duration (e.g. 125ms), (the number being two in Fig. 2 or Fig. 3) for each or both channels. Third, the start time t _stait of the each sub-segment with respect to the start time T _star t of the time window where the spectral power is above the threshold, and optionally where the duration is larger than the predefined minimal segment duration, for each or both channels and for each of the two consecutive clenching actions or events (the start time being within 50ms in Fig. 2 and Fig. 3, indicating that the clenching actions or events are synchronized, as explained above). Fourth, the duration and/or the power above the threshold for each sub- segment that is above a threshold (e.g. 0.0 lmV ² s) for both clenching actions or events at both sides (which is around 0.05mV ² s in Fig. 2 and Fig. 3), which would indicate that these clenching actions can be classified as events (two short clenching events on the right side in Fig. 2 and two short clenching events on the left side in Fig. 3). Fifth, the duration t; of the section between the two consecutive sub-segments being larger than a threshold (e.g. 25ms), indicating or confirming that these sub-segments are separate clenching actions or events.

Now, with respect to Figs. 4 to 12 the detection or classification of short and long teeth clenching events is described. Thus, each teeth clenching event is a teeth clenching event selected from the group comprising at least a first (short) teeth clenching event indicating a teeth clenching action of a first time duration and a second (long) teeth clenching event indicating a teeth clenching action of a second time duration, the second time duration being greater than the first time duration. The size T of the time window is sufficiently large to detect at least two consecutive first (short) teeth clenching events of the first time duration. Assuming a first (short) time duration of less than 0.5 seconds, the size of the time window can be about 1 second. In the examples shown in Figs. 2 to 12 the size T of the time window is exactly 1 second.

Even though only a short and a long teeth clenching event are detected with reference to Figs. 4 to 12, it will be understood that further teeth clenching events of different durations can be detected or classified. For example, each teeth clenching event can be a first (short) or second (long) teeth clenching event as explained above, and the group can further comprise at least a third teeth (very short) clenching event indicating a teeth clenching action of a third time duration and/or a fourth (very long) teeth clenching event indicating a teeth clenching action of a fourth time duration, the third time duration being smaller than the first time duration and/or the fourth time duration being greater than the third time duration.

Figs. 4a, 5a, 6a, 7a, 8a, 9a, 10a, 11 a, 12a each show an electromyogram signal indicative of electrical activity produced by muscle tension cause by teeth clenching on the left side of the head of the user (left channel), in particular measured with the second EMG sensor 40b of Fig. 1. Figs. 4b, 5b, 6b, 7b, 8b, 9b, 10b, l i b, 12b each show a electromyogram signal indicative of electrical activity produced by muscle tension cause by teeth clenching on the right side of the head of the user (right channel), in particular measured with the first EMG sensor 40a of Fig. 1. Figs. 4c, 5c, 6c, 7c, 8c, 9c, 10c, 11 c, 12c each show the spectral power of both the EMG signal of the right channel (indicated by circles) and the EMG signal of the left channel (indicated by crosses).

For obtaining the EMG signals of Fig. 4 to Fig. 12, the user was asked to repeatedly perform a specific teeth clenching action. Fig. 4a-c each shows an

electromyogram signal of (repeatedly) two consecutive short teeth clenching events on the right side. Fig. 5a-c each shows an electromyogram signal of (repeatedly) two consecutive short teeth clenching events on the left side. Fig. 6a-c each show an electromyogram signal of (repeatedly) two consecutive short teeth clenching event on both the right and the left side. Fig. 7a-c each shows an electromyogram signal of (repeatedly) a single short teeth clenching event on the right side. Fig. 8a-c each shows an electromyogram signal of (repeatedly) a single short teeth clenching event on the left side. Fig. 9a-c each shows an electromyogram signal of (repeatedly) a single short teeth clenching event on both the right and left side. Fig. lOa-c each shows an electromyogram signal of (repeatedly) a single long teeth clenching event on the right side. Fig. 1 la-c each shows an electromyogram signal of (repeatedly) a single long teeth clenching event on the left side. Fig. 12a-c each shows an electromyogram signal of (repeatedly) a single long teeth clenching event on both the right and left side. In each of Figs. 4c, 5c, 6c, 7c, 8c, 9c, 10c, 11c, 12c the specific teeth clenching action is performed in the time window of size T of 1 second, as explained above. For simplification purposes the time window of size T is only explicitly indicated in Fig. 4c and Fig. 5c.

However, it shall be understood that such time window of size T is used in the same way also for each of Figs. 6c, 7c, 8c, 9c, 10c, 11c, 12c.

With reference to Figs. 4 to 12 the detection of short and long teeth clenching, on the left, right and at both sides, and double clenching is illustrated, measured with dry electrodes integrated in a wearable device, in particular the wearable device 20 of Fig. 1 in form of headphones. The first and second sensors 40a, 40b or its dry electrodes were positioned at temporal locations (around T7 and T8 positions in the International 10-20 System for positioning EEG electrodes) and the third and fourth sensors 40c, 40d or its dry electrodes at central locations (around C3 and C4). The amplification was performed using small-scale amplifier on the two channels that measure the difference in the electrical signal obtained with the electrodes at the position C3 and T7 for the first channel and C4 and T8 for the second channel. The (dry) ground electrode was positioned at the earlobe, below the T8 position. The recorded electrical signal was amplified and transmitted to the command detection device, such as a PC, for signal analysis.

Thus, with reference to Figs. 4 to 12, to enable fast communication using teeth clenching a number of teeth clenching event combinations are proposed to be used. These combinations are based on the hemispheric teeth clenching action (i.e., left, right, and both left and right side teeth clenching), temporal dimension (i.e., longer and shorter teeth clenching duration), and the sequence of teeth clenching actions or events in a time window (e.g. two short clenching actions following each other, thus "double clenching"). For example, a combination of left side teeth clenching followed by a right side and both side clenching action can be performed within the Is time window. Reliable detection of these activities enables the communication speed of 54 symbols per one time window, which can lead to a bit rate of more than 250 bits per minute (see below). These values are much higher as compared to, for example, the communication speed achieved with brain-computer- interface (BCI) applications (up to 125bpm). Therefore, the device or method described herein can be used to replace or to complement a BCI application which is designed to capture primarily brain activity. For such replacement, only EMG signals are used (instead of EEG signals). For such complementation, both EMG signals and EEG signals are used, which combined can yield better performance and/or higher stability.

The detection of the symbol or command issued by the user is based on the changes in the spectral power in the frequency band of 12 to 100Hz. A number of parameters are extracted from the spectral power measured at electrodes positioned at the contralateral hemisphere, based on the pre-determined thresholds for the EMG signal of each channel: the amount of spectral power above the threshold, number of segments where the spectral power is above the threshold, the starting time of the segment where the spectral power is above the threshold, duration of each segment where the spectral power is above the threshold, and duration of segments that are below the threshold. These parameters are estimated within the time window of around Is that starts when the spectral power of the EMG signal in one of the channels becomes larger than the threshold. Based on the analysis of the parameters a decision is made what is the command or symbol selected by the user.

The method for analyzing the EMG signal and for detecting the command or the symbol selected by the user can be implemented either in a signal processing unit (e.g. dedicated electronic circuit) that is integrated in a wearable device (e.g. headphones or headset) or it can be a signal processing unit (e.g. software component) that is installed on a separate command detection device, in particular receiver unit, (e.g., MP3/MP4/DVD player, TV, or personal computer).

The detection of the command/symbol issued by the user is based on one or more (or all) of the following parameters extracted by analyzing the changes in the spectral power of the band-pass (e.g. 12-100Hz) filtered bio-signal (EMG signal) obtained using a sensor having dry electrodes, and based on the predefined spectral power thresholds for each of the measuring sensors or channels (generated either manually or in a calibration period that precedes the usage of the device): the total amount of spectral power above the threshold in a time window, the number of segments where the spectral power is above the threshold and where the duration of each sub-segment is larger than the pre-specified minimal segment duration (i.e., 125ms) within the time window, the starting time of each sub-segment with respect to the beginning of the time window where the spectral power is above the threshold and where the duration is larger than minimal segment duration, the duration and the spectral power above the threshold for each sub-segment that is above the threshold, and the duration of the section between the sub-segments detected in a time window. All these parameters are estimated on a pre-defined time window, such that the start point of each time window is synchronized to the time point when the spectral power in one of the electrodes reaches the threshold value for that electrode. The pre-defined parameters can be tailored to suit each person in a manual or automatic calibration procedure.

Based on these parameters it can be distinguished how many times the person clenches his teeth in a time window, whether he clenched the teeth at both sides or only on the right or on the left side and what was the combination or sequence of clenching actions or events within the time window. In one specific example, if the size or duration of the time window is set to Is, three types of clenching events (brief, short, and long) are detectable, and if it is assumed that a maximal number of clenching events per time window is three, the following list of teeth clenching actions or events per time window can be formed for this specific example: Long clenching (>0.6s): left, right, both (3 commands)

One short clenching (0.4s): left, right, or synchronized at both sides (3 commands) One brief clenching (0.2-0.3s): left, right, or synchronized at both sides (3 commands)

Two short clenching (9 commands)

a. of the same kind: left, right, or synchronized at both sides (3 commands) b. in a sequence: left-Might, right-Meft, left-M)oth, right-bboth, both-Meft, both- right (6 commands)

Two brief clenching (9 commands)

a. of the same kind: left, right, or synchronized at both sides (3 commands) b. in a sequence: left-Might, right-Meft, left-M)oth, right-bboth, both-Meft, both- right (6 commands)

Three brief clenching (27 commands)

a. of the same kind: left, right, or synchronized at both sides (3 commands) b. in a sequence starting with left: left -Meft-Might, left-Meft-M)oth, left->right->left, left bright bright, left^right^both, left->both->left, left -M)oth- right, left-M)oth-M)oth (8 commands)

c. in a sequence starting with right: right-Meft-Meft, rights left-Might, right^left^both, rights rights left, right^right^both, right^both^left, right -M)oth- right, right -M)oth- both (8 commands)

d. in a sequence starting with both: both-Meft-Meft, both-Meft-Might, both^left^both, both^ rights left, both^ rights both, both^ rights both, both-M)oth-Meft, both-M)oth- right (8 commands).

In total this makes 54 commands/symbols which together with no clenching action can be used for asynchronous control and communication. Assuming that a user can realize each of these commands/symbols in a time frame of Is and that 0.4s is the rest time between two consecutive command executions, the potential bit rate that can be achieved is log_2(54) * 60s/(l+0.4)s = 247bmp. Additional 18 combinations are also possible:

1. One short followed by one brief clenching (9 commands)

a. of the same kind: left, right, or synchronized at both sides (3 commands) b. in a sequence: left-Might, right-Meft, left-M)oth, right-bboth, both-Meft, both- right (6 commands)

2. One short followed by one brief clenching (9 commands)

a. of the same kind: left, right, or synchronized at both sides (3 commands) b. in a sequence: left-Might, right-Meft, left-M)oth, right-bboth, both-Meft, both- right (6 commands)

This would increase the number of symbols to 72 and the transfer rate to 264bpm.

Having the same setup with another very long clenching event (>0.8s) and a time window of 1.4s (command) and 0.4s (rest) would yield 201 symbols (81 four times brief clenching, 27 triple short clenching, 9 double long clenching, 3 single very long clenching, 3 single brief clenching, 3 single short clenching, 3 single long clenching, 9 double brief clenching, 9 double short clenching, 27 triple brief clenching, and 27 triple short clenching actions) and a bit rate of 255bpm. By allowing the mixing of clenching actions with different duration in a single command this would result in 498 symbols (additional 18 brief and short clenching pair, 18 brief and long clenching pair, 18 short and long clenching pair, 81 double brief and single short clenching triplet, 81 single brief and double short clenching triplet, and 81 double brief and single long clenching triplet actions) and a bit rate of 299bpm.

Fig. 13 shows a front view of a wearable device 20 according to an

embodiment, in particular the wearable device 20 as explained with reference to Fig. 1. In Fig. 13 the wearable device 20 is worn by user on the head of the user. The wearable device 20, configured to be put at least partly around the head of a user, comprises a plurality of electromyogram sensors 40a, 40b, 40c, 40d as previously explained, in particular with reference to Fig. 1. Each of the electromyogram sensors 40a, 40b, 40c, 40d comprises an array of dry pin electrodes 109. Further, the wearable device 20 comprises at least one elastic element 111 which is adapted to at least partly follow the curvature of the user's head. Each of the sensors 40a, 40b, 40c, 40d or its electrode array 109 is arranged on the elastic element 111. In particular, the wearable device comprises a plurality of elastic elements 111 , for example two elastic elements 111 as shown in Fig. 13. In an example, the wearable device 20 enables positioning of the electrode arrays 109 on a user's scalp. The wearable device 20 of Fig. 13 is a headset or headphones comprising a headpiece 103 (also referred to in this description by "unit" or "housing"), made of firm but flexible material and having the shape of a headband, and earpieces 105. The device furthermore comprises a positioning arrangement consisting of two projection elements 107 and a positioning strap 113. The projection elements 107 are used for positioning the plurality of sensors (each comprising an electrode array 109 and a (flexible) substrate 110), elastic elements (bands) 111 and the positioning strap 113 at the inner side of the housing 103. The two projection elements 107 project the ends of the positioning strap 113 against the auricles 114 of the user, when the housing is put around the user's head. The ends of the elastic bands 111 are also fixed to the projection elements 107 close to the points at which the ends of the positioning strap 113 are fixed to the projection elements 107. The positioning strap 113 is used for positioning the sensors or electrode arrays at predefined positions on the scalp. To cope with the variety in head sizes and shapes the positioning strap 113 is divided in two halves with a connecting elastic band 119 between them to guarantee good mechanical contact to the scalp all over the circumference.

The electrode position of two of the electrode arrays 109 (sensors 40a, 40b) is set by the spring loaded fixation point of the two halves of the positioning strap as close as possible to the T7 and T8 locations of the International 10/20 System. The two halves of the positioning strap have the ears as mechanical reference via the main headphone-clamp. The electrode arrays 109 (sensors 40c, 40d) positioned at C3 and C4 locations according to the International 10/20 System are fixed at the other ends of both of the positioning strap parts. In this way all electrode arrays 109 (or sensors 40a, 40b, 40c, 40d) are positioned in their right places in one simple action by just putting on the headphones in the usual way by the user. In this regard it is to be noted that the wearable device 20 may be something else than headphones, for example a (Alice) band, cap, helmet, glasses, etc. Due to the stress or strain on the elastic bands 111 caused by the stretching thereof, which at its turn is caused by the insertion of the user's head in the housing (which pushes the positioning strap 113 and the elastic bands 111 upwards), the elastic bands 111 press the arrays of electrodes on the user's scalp, resulting in an effective contact of the electrodes to the scalp.

Dry electrodes, along with the amplification component, are designed to handle low amplitude (EEG) signal which is in the order of up to ΙΟΟμν and the EMG signal in the order of few mV. The sampling rate is sufficient to reliably monitor frequencies of up to 100Hz (i.e., sampling rate should be at least 200Hz) and low enough to preserve independent operation of the electronics for a prolonged period of time. The integration of dry electrodes in the headset is such that it ensures the contact with the skin on the scalp even through the hair, while being comfortable to wear. This, together with the proper amplifier component ensures the signal quality for reliable control and communication.

By using dry electrodes with the flexible pin structure that can be integrated in an off-the-shelf wearable device, such as for example headphone, head cap, helmet, etc, convenience can be provided. The assembly or device can be applied to various use cases ranging from everyday usage of headphones to the usage of helmets in a space shuttle. Since dry electrodes do not require any preparation and cleaning they are superior to the traditional wet electrodes used for bio-signal measurement. Additionally, the pin-structure enables the penetration through the hair and positioning of dry electrodes on an arbitrary location on a person's head, and seamless integration in a headset.

The wearable device can comprise at least four arrays of dry electrodes that are integrated in a headset and designed to be position on a human head. Out of these four arrays of electrodes at least two arrays of electrodes can be positioned at a contralateral hemisphere of a human head such to be close enough to the temporal lobe to be able to acquire the electrical signal stemming from jaw clenching action or activity. The other two arrays of electrodes can be used as a reference and ground electrode, respectively. To enable arbitrary positioning on a human head, electrodes that are intended to be positioned on places where the presence of hair is expected can be designed with flexible pins. The electrodes can be integrated in a headset such that they do not breach the outside appearance of the headset and the comfort of the user wearing it is maintained (i.e., to preserve the look and feel). All the required electronics for amplifying and wirelessly transmitting the signal can be seamlessly integrated into the headset following the previously mentioned requirements, powered by an integrated battery such that it can work autonomously for a larger period of time and without any external connections. The electronic components can be designed to monitor human brain activity, i.e., EEG, as well as electrical activity produced by tension of muscles on a human head that are relatively close to the measuring electrodes.

The command detection device and method disclosed herein can be used in any suitable implementation, in particular for inconspicuous communication where secrecy of communication and/or control is desired. Examples of such implementations are using a bank or credit card, communication and/or control in a "busy hands" situation and/or in a condition where speech cannot be used (e.g. for astronauts, divers, etc.), or controlling an audio/video device (e.g., MP3/MP4/DVD player, a TV, etc.). In another example a user with a disability, who cannot use conventional means of communication, can use such command detection device and method.

The invention can be used as a mean of communication and/or control for people that are not able to use the standard communication means (such as keyboards and pointers) or standard controls (such as joysticks and steering wheels). The communication and control inability can for example be caused by disability of people to use their extremities caused by illness or accidents, or by situational impairment, such as impairments due to low pressure (e.g., astronauts) or high pressure (e.g., divers), or "busy hands" (e.g., surgeons).

The invention provides very fast communication and at the same time prevents stigmatization of people due to the inconspicuous nature of the teeth clenching action or activity. For the latter, this might be the only way to provide fast communication, especially in critical situations.

In one exemplary application, the invention can be used as an additional device in the healthcare field (e.g. in assisting surgeons, radiologists, etc.), such as an additional device for an MR system, CT system, or other healthcare system. In another exemplary application, the invention can be used in an everyday life device. One aspect is the usage of the invention where secrecy of communication and/or control is required, such as to provide privacy when communicating confidential information (e.g., entering passwords, sending confidential information using a PC - by replacing keyboard typing). Another aspect is the usage of the invention to replace the traditional typing and voice control of devices (such as controlling electronic devices at home and on-the-go). The device to be controlled can be for example be an MP3 player, an MP4 player, a DVD player, or a TV. Such a device can be controlled while for example riding a bicycle or driving in the car.

The invention described herein can be used for communication and/or control of environments using electromyogram (EMG) signals stemming from teeth clenching. The advantage of the presented invention is that communication achieved by using various combinations/sequences and durations of teeth clenching actions or events can lead to a very high transfer rates (around 1600 symbols or 250 bits per minute), and that it be realized in such a way that the communication and/or control process is inconspicuous, i.e., it can be applied in situations where secrecy of communication and/or control can be achieved.

The size of the time window corresponds to a time required to detect or select a single command or a single symbol and that allows a combination of at least two clenching sub-segments of the same and/or different types (i.e., laterality (right or left) and/or duration). In particular, a combination of the clenching combination/sequence of clenching events and the duration of clenching events in the pre-specified time window can be used for the command/symbol detection or selection. Optionally additional features or parameters can be used for the clenching command/symbol detection or selection. In this way, a very high communication speed can be achieved. In particular, the additional features or parameters can be: the total power spectral density (PSD) in a time -window above the threshold, number and duration of sub-segments (e.g. larger than a minimal segment duration, e.g. >125ms) with PSD above the threshold, the start times t _sta rt and end times t _en d of each sub-segment and the duration t; or distance between them. In this way a theoretically very high communication speed (of more than 250 bits per minute) that can be increased with the length of the time window (i.e., number of possible symbols) can be provided.

While the invention has been illustrated and described in detail in the drawings and foregoing description, such illustration and description are to be considered illustrative or exemplary and not restrictive; the invention is not limited to the disclosed embodiments. Other variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure, and the appended claims.

In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. A single element or other unit may fulfill the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

A computer program may be stored/distributed on a suitable medium, such as an optical storage medium or a solid-state medium supplied together with or as part of other hardware, but may also be distributed in other forms, such as via the Internet or other wired or wireless telecommunication systems.

Any reference signs in the claims should not be construed as limiting the scope.