Method and system for improved electroencephalography (EEG) sensor mounting

Abstract

A method for aligning an electroencephalography (EEG) cap and a system operable to carry out the method. The method comprising capturing an image of a subject's face and an EEG cap, wherein the EEG cap is placed on the subject's head. The method further comprising determining, from the captured image, an alignment of the subject's face. The method further comprising determining, from the captured image, an alignment of the EEG cap relative to the alignment of the subject's face. The method further comprising indicating that the EEG cap is aligned correctly if the alignment of the EEG cap relative to the alignment of the subject's face is within a predetermined threshold alignment.

Description

[0001] METHOD AND SYSTEM FOR IMPROVED ELECTROENCEPHALOGRAPHY (EEG) SENSOR MOUNTING

[0002] The present invention relates to a method and a system of determining alignments of an electroencephalography, EEG, cap and on improvements of aligning an EEG cap.

[0003] BACKGROUND OF THE INVENTION

[0004] Electroencephalography (EEG) is a method to record an electrogram of the spontaneous electrical activity of the brain. Biosignals recorded with EEG represent the total post synaptic potential of pyramidal neurons of the cerebral cortex and are used for functional assessment of brain function in a non-invasive manner. The electroencephalogram is typically recorded by placing one or more electrodes on a user’ s head. This can be done, for example, with the means of an electroencephalogram (EEG) cap. As a part of the work on Neuroscience proj ects, a large number of psychophysiological experiments may be carried out. Experiments involve the recording of a large number of biological signals (such as electroencephalogram (EEG), electrocardiogram, eye movements, respiration rate, photoplethysmography and others) and the use of a large amount of technically sophisticated equipment.

[0005] Quality installation of such a large number of sensors in a highly compliant manner requires a lot of effort from technicians. The mounting procedure of the EEG cap is labor- intensive, because it requires proper positioning of the electrodes on the scalp surface, degreasing the mounting site to reduce impedance, filling the electrodes with gel, and ensuring that the quality of electrode-skin contact is uniform over the entire scalp surface. Electrode positioning is one of the key characteristics of installation quality, since further data analysis relies on the assumption that the electrodes are positioned in the correct regions. Incorrect positioning can lead to errors in data interpretation.

[0006] Therefore, it is important on the one hand to help technicians to signal an error before starting an experiment, and on the other hand to be able to detect already committed errors post factum to exclude such experiments from the analysis. The gold standard for EEG mounting is electrode placement using the 10-20 system. Quality control of the installation is performed by the technician who performs the installation. This involves manual review of the experiment recordings to assess the quality of the assembly which is time-consuming and inserts a margin of error which can obscure results from an electrogram. It is an aim of this invention to address the above-mentioned issues, for example, by reducing human errors during installation and to automate quality control in el ectroencephal ography .

[0007] SUMMARY

[0008] To achieve the above objectives, the invention sets out a method and a system for aligning an electroencephalography (EEG) cap as set out in the claims below.

[0009] In a preferred embodiment, a method for aligning an electroencephalography (EEG) cap is provided. The method includes capturing an image of a subject’s face and an EEG cap, wherein the EEG cap is placed on the subject’s head. The method further includes determining, from the captured image, an alignment of the subject’s face. The method further includes determining, from the captured image, an alignment of the EEG cap relative to the alignment of the subject’s face. The method further includes indicating that the EEG cap is aligned correctly if the alignment of the EEG cap relative to the alignment of the subject’s face is within a predetermined threshold alignment. By determining the alignment of both the subject’s face and the EEG cap from an image taken of the subject’s face (as opposed to, for example, by a human manually adjusting the subject’s face and / or the EEG cap without the help of an image), errors in alignment can be determined at a much higher accuracy. This has the added benefit of informing an EEG system (for example, by displaying the misalignment on a computer screen) of misalignments on a smaller scale (for example, on the order of millimeters and fractions of a millimeter rather than centimeters) which would not be visible to a human analysing the alignment of the subject’s face and / or the EEG cap. Accordingly, human errors during installation of the EEG cap are reduced. It is imperative for electrodes within an EEG cap to be placed accurately on a subject’s head to ensure that a useful EEG reading is taken. By improving the detection accuracy of misalignment, further occurrences of misalignment can be detected before an EEG reading is taken. A user can be informed, by the system, by what amount the subject’s head and / or the EEG cap have to be aligned for an ideal EEG reading to be taken. Thus, the percentage of false or meaningless EEG readings (due to misalignment) can be drastically reduced.

[0010] In an embodiment, capturing the image may include capturing the image with an image capturing device and storing the captured image on memory. Advantageously, the method can be performed by a simple image capturing device without the need of additional complex and expensive physical components. This has the additional benefit of providing a robust, fast and accessible EEG alignment detection method.

[0011] In an embodiment, the method may further include identifying a first set of features on the subj ecf s face of the captured image, the first set including at least two areas on the subj ecf s face corresponding to horizontal features of the subject’s face. The method may further include identifying a second set of features on the subject’s face of the captured image, the second set including at least two areas on the subject’s face corresponding to vertical features of the subject’s face. Advantageously, a unique alignment of the subject’s face, and a corresponding unique alignment of the EEG cap based on the subject’s face, can be determined depending on the specific features of that subject’s facial features. Different alignment can be determined and / or different misalignments can be identified depending on the subject’s facial features. For example, one subject might have a crooked nose whereas a further subject may have a different separation between their eyes in any dimension compared to another subjects. By identifying the horizontal and vertical features of the subject’s face, any variations in the subject’s face can be taken into account when identifying misalignments of the EEG cap.

[0012] In an embodiment, the first set of features may include the subject’s right temple and left temple, left comer of left eye and right comer of right eye, right comer of left eye and left corner of right eye, or any combination of the above. The second set of features may include at least two of the subject’s highest point of the nose, point of the nose tip, lowest point of the nose, or the point of the chin. Advantageously, different pairs of features of a subject’s face can be taken into account when identifying misalignments of the EEG cap and / or determining how to align the EEG cap relative to the subject’s face. For example, this may be advantageous because an example subject might have more pronounced eyes (thus providing a more accurate result when determining the horizontal features based on their eyes) whereas a different subject may have more pronounced temples (thus providing a more accurate result when determining the horizontal features based on their temples).

[0013] In an embodiment, determining the alignment of the subject’s face may include determining a horizontal value from the first set of features on the subject’s face of the captured image and determining a vertical value from the second set of features on the subject’s face of the captured image. Determining the alignment of the subject’s face may also include comparing the horizontal value to a predetermined horizontal value, comparing the vertical value to a predetermined vertical value, and determining an alignment value of the subject’s face of the captured image based on the comparison of the horizontal value to the predetermined horizontal value and the vertical value to the predetermined vertical value. Advantageously, the current alignment of the subject’s face can be compared to a ‘base value’ where the ‘base value’ is an ‘ideal’ alignment of a “sample face” (for example, the subject looking directly at the camera) at which more accurate measurements of the EEG cap relative to the subject’s face can be taken. By determining how aligned the subject’s face is to the “sample face”, an indication can be provided (for example, a prompt on a screen) to align the subject’s face in the horizontal and / or vertical axis to come as close as possible to the alignment of the “sample face”.

[0014] In an embodiment, it may be determined that the subject’s face is aligned when the alignment value is below a pre-determined alignment threshold.

[0015] In an embodiment, the EEG cap may include an alignment marker and determining an alignment of the EEG cap relative to the alignment of the subject’s face may include determining a horizontal value of the alignment marker relative to the horizontal features of the subject’s face. Determining an alignment of the EEG cap relative to the alignment of the subject’s face may further include determining a vertical value of the alignment marker relative to the vertical features of the subject’s face.

[0016] In an embodiment, the vertical value of the alignment marker may include a vertical distance, wherein the vertical distance is a shortest distance between a point of the alignment marker that is on a bottom line and is equidistant to the lowermost left corner and the lowermost right comer of the alignment marker, wherein the bottom line is a relatively horizontal line from the lowermost left corner and the lowermost right comer of the alignment marker, and a central most point on a line drawn from a first of the first set of features to a second of the first set of features. Additionally or alternatively, the vertical value may include a vertical angle between the bottom line and the line drawn from the first of the first set of features to the second of the first set of features. Advantageously, the method can determine the vertical alignment value with a simple formula which provides a more robust, simple and relatively easy method of indicating an alignment / misalignment of the EEG cap relative to the subject’s face.

[0017] In an embodiment, the horizontal value of the alignment marker may include a horizontal distance, wherein the horizontal distance is a shortest distance between a vertical line that is perpendicular to the bottom line and passes through a point of the alignment marker that is on the bottom line and is equidistant to the lowermost left comer and the lowermost right corner of the alignment marker, and a line drawn from a first of the second set of features and a second of the second set of features. Additionally or alternatively, the horizontal value may include a horizontal angle between the vertical line and the line drawn from the first of the second set of features and the second of the second set of features. Advantageously, the method can determine the horizontal alignment value with a simple formula which provides a more robust, simple and relatively easy method of indicating an alignment / misalignment of the EEG cap relative to the subject’s face.

[0018] In an embodiment, the predetermined threshold alignment may include the vertical distance being within a predetermined threshold. Additionally or alternatively, the predetermined threshold alignment may include the vertical angle being within a predetermined threshold. Advantageously, the alignment / misalignment of the EEG cap can be adjusted (in the vertical direction) depending on the necessity of alignment. Some EEG readings require more alignment than others. Thus, the method can be applied to systems that require a higher degree of alignment as well as those that require a lower degree of alignment.

[0019] In an embodiment, the predetermined threshold of the vertical distance may be greater than or equal to 0.9 and less than or equal to 1.9 when the vertical distance is a dimensionless measure, wherein the dimensionless measure of the vertical distance is the vertical distance divided by the distance of the line drawn from the first of the first set of features to the second of the first set of features. This value is determined to represent a further high level of alignment accuracy based on an experimental finding 159 EEG experiments.

[0020] In an embodiment, the predetermined threshold of the vertical angle may be greater than or equal to 0 degrees and less than or equal to + / - 5 degrees (or 5 degrees as an absolute value). This value is determined to represent a further high level of alignment accuracy based on an experimental finding 159 EEG experiments.

[0021] In an embodiment, the predetermined threshold alignment may include the horizontal distance being within a predetermined threshold. Additionally or alternatively, the predetermined threshold alignment may include the horizontal angle being within a predetermined threshold. Advantageously, the alignment / misalignment of the EEG cap can be adjusted (in the horizontal direction) depending on the necessity of alignment. Some EEG readings require more alignment than others. Thus, the method can be applied to systems that require a higher degree of alignment as well as those that require a lower degree of alignment.

[0022] In an embodiment, the predetermined threshold of the horizontal distance may be greater than or equal to -0.3 and less than or equal to 0.3 when the horizontal distance is a dimensionless measure, wherein the dimensionless measure of the horizontal distance is the horizontal distance divided by the length of the bottom line. This value is determined to represent a high level of alignment accuracy based on an experimental finding 159 EEG experiments.

[0023] In an embodiment, the predetermined threshold of the horizontal angle may be greater than or equal to 0 degrees and less than or equal to + / - 5 degrees (or 5 degrees as an absolute value)D. This value is determined to represent a further high level of alignment accuracy based on an experimental finding 159 EEG experiments.

[0024] In an embodiment, the method as described above may be computer-implemented. Advantageously, the alignment and / or misalignment values can be determined without the help of a user. This can be done, for example, with the use of neural networks and / or other machinelearning algorithms. This drastically improves the automation aspect of determining alignment / misalignment of an EEG cap relative to a subject’s face which, in turn, further increases the robustness and simplicity of the method.

[0025] In a preferred embodiment, a system for aligning an electroencephalography (EEG) cap is provided. The system includes an electroencephalography (EEG) cap, an image capturing device, and a computer or computing device (wherein the computer or computing device may include a memory, and a processor). The image capturing device is operable to capture an image of a subject’s face and an EEG cap, wherein the EEG cap is placed on the subject’s head. The computing device is operable to determine, from the captured image, an alignment of the subject’s face. The computing device is further operable to determine, from the captured image, an alignment of the EEG cap relative to the alignment of the subject’s face. The computing device is further operable to indicate that the EEG cap is aligned correctly if the alignment of the EEG cap relative to the alignment of the subject’s face is within a predetermined threshold alignment.

[0026] In an embodiment, the system may further include an alignment marker placed on the EEG cap. Advantageously, horizontal and vertical alignments can be measured specifically without having to take into account the varying shape of the EEG cap. The alignment marker may be any one of a barcode, a QR code, an ArUco marker, or a combination of the above. Advantageously, the horizontal and vertical alignments can be measured with any suitable alignment marker depending on what is available. This increases the useability of the system and allows the system to be future-proofed and used in markets throughout the globe. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The features, objects, and advantages of the present disclosure will become more apparent from the detailed description set forth below when taken in conjunction with the drawings in which like reference numerals refer to similar elements.

[0028] Figure 1 shows an example of a system for determining alignment of an EEG cap according to the invention;

[0029] Figure 2 shows an image of a subject and an EEG cap captured by the system of Figure 1 and depicts the detection of facial features of the subject;

[0030] Figure 3 shows a plurality of images of a subject and an EEG cap captured by the system of Figure 1 and depicts examples of detection of an alignment of the subject’s face;

[0031] Figure 4 shows a further image of a subject and an EEG cap captured by the system of Figure 1 and depicts the detection of facial features of the subject;

[0032] Figure 5 shows an image of a subject and an EEG cap captured by the system of Figure 1 and depicts horizontal and vertical directions of an alignment marker and the spatial positioning of the EEG cap;

[0033] Figure 6 shows a plurality of images of a subject and an EEG cap captured by the system of Figure 1 and depicts examples of detection of an alignment of the EEG cap relative to the alignment of the subject’s face;

[0034] Figure 7 shows graphs that depict distributions of metrics used to determined threshold alignment values;

[0035] Figure 8 shows images of a subject and an EEG cap captured by the system of Figure 1 and depicts close to optimal positions of an alignment marker of the EEG cap compared to a target position of the alignment marker, in accordance with an embodiment of the invention;

[0036] Figure 9 shows images of a subject and an EEG cap indicating an aligned but with examples where the EEG cap is significantly higher or lower; and

[0037] Figure 10 depicts a flow chart of a method according to the invention.

[0038] DETAILED DESCRIPTION

[0039] Figure 1 shows a system 100 for determining an alignment of an electroencephalography (EEG) cap. The system includes an electroencephalography (EEG) cap 102, an image capturing device 106, and a computer or computing device. A memory and processor may form part of the computer or computing device which may be coupled to the system 100 by a wireless or wired means. The image capturing device (for example, a video or image camera) is operable to capture an image of a subject’s face and an EEG cap, wherein the EEG cap is placed on the subject’s head 108. Examples of captured images are shown in Figures 2 to 6, 8 and 9 below. The computing device is operable to receive the captured image and to determine, from the captured image, an alignment of the subject’s face. The computing device is further operable to determine, from the captured image, an alignment of the EEG cap relative to the alignment of the subject’s face. The computing device is further operable to indicate that the EEG cap is aligned correctly if the alignment of the EEG cap relative to the alignment of the subject’s face is within a predetermined threshold alignment.

[0040] By determining the alignment of both the subject’s face and the EEG cap 102 from an image taken of the subject’s face (as opposed to, for example, by a human manually adjusting the subject’s face and / or the EEG cap without the help of an image), errors in alignment can be determined at a much higher accuracy. This has the added benefit of informing an EEG system 100 (for example, by displaying the misalignment on a computer screen) of misalignments on a smaller scale (for example, on the order of millimeters and fractions of a millimeter rather than centimeters) which would not be visible to a human analysing the alignment of the subject’s face and / or the EEG cap. Accordingly, human errors during installation of the EEG cap 102 are reduced. It is imperative for electrodes within an EEG cap to be placed accurately on a subject’s head to ensure that a useful EEG reading is taken. By improving the detection accuracy of misalignment, further occurrences of misalignment can be detected before an EEG reading is taken. A user can be informed, by the system, by what amount the subject’s head and / or the EEG cap have to be aligned for an ideal EEG reading to be taken. Thus, the percentage of false or meaningless EEG readings (due to misalignment) can be drastically reduced.

[0041] The system 100 includes an alignment marker 104 placed on the EEG cap. This marker may be printed on fabric of the EEG cap, glued on to the EEG cap, engraved within the EEG cap, or attached in any other suitable way. Advantageously, horizontal and vertical alignments can be measured specifically without having to take into account the varying shape of the EEG cap. The alignment marker 104 may be a 2D binary-encoded fiducial pattern, such as a barcode, a QR code, an ArUco marker, or a combination of the above. The advantages of utilizing a 2D binary-encoded fiducial pattern for alignment purposes is that such patterns can easily be detected by standardized computer programs (such as OpenCV, an open-source computer vision library) and the coordinates of such patterns can easily be determined by such programs. This provides a robust, fast and simple alignment option for the EEG cap. The system 100 is not limited to a certain type of 2D binary-encoded fiducial pattern and may include any other suitable pattern (such as a barcode, QR code, ArUco marker, but also any other type of 2D binary-encoded fiducial pattern). Advantageously, the horizontal and vertical alignments can be measured with any suitable alignment marker 104 depending on what is available. This increases the useability of the system and allows the system to be future-proofed and used in markets throughout the globe.

[0042] The alignment marker 104 may be attached (for example, sewn, stuck or any other method of attachment) at a center position of the EEG cap 102 such that the alignment marker 104 is forward-facing (towards the image capturing device 106) area of the EEG cap 102. This is exemplified in Figure 2.

[0043] Figures 2 to 6, 8 and 9 show images of a subject and an EEG cap 102 captured by the system 100 of Figure 1. In Figure 2, an EEG cap 102 is shown including the alignment marker 104. Figures 3 to 6, 8 and 9 also show the EEG cap 102 including the alignment marker 104. Figures 2 to 9 indicate a method for aligning an EEG cap 102 which may be carried out by the system 100 of figure 1. The method may be carried out by a computer (i.e., be a computer- implemented method), the computer comprising a processor and memory as described above in Figure 1. Advantageously, the alignment and / or misalignment values can be determined without the help of a user. This can be done, for example, with the use of neural networks and / or other machine-learning algorithms. This drastically improves the automation aspect of determining alignment / misalignment of an EEG cap relative to a subject’s face which, in turn, further increases the robustness and simplicity of the method. In addition to Figures 2 to 9, further features of the method are shown and described below in relation to Figure 10.

[0044] The method includes capturing an image of a subject’s face and an EEG cap 102, wherein the EEG cap 102 is placed on the subject’s body or head 108. Capturing the image may include capturing the image with an image capturing device 106 and storing the captured image on memory, as described above in relation to Figure 1. The image capturing device 106 may be a film camera, a video camera, an image camera or any other type of suitable image capturing device. Advantageously, the method can be performed by a simple image capturing device without the need of additional complex and expensive physical components. This has the additional benefit of providing a robust, fast and accessible EEG alignment detection method. Subsequent to capturing the image of the subject’s face and the EEG cap 102 (or subsequently to receiving the captured image), an alignment of the subject’s face is determined from the captured image. This may include detecting the presence of a face by retrieving facial features or facial landmarks from the captured image by processing the captured image with a neural network (such as, but not limited to, a convolutional neural network) or any other suitable machine learning algorithm.

[0045] Detection of the face may include identifying a plurality of features on the subject’s face and grouping those features into ‘sets’ of features. There may be one, two, three or more ‘sets’ of features. In one example, a first set of features on the subject’s face of the captured image may be identified, wherein the first set includes at least two areas (denoted as circles on the subject’s face in captured image 200) on the subject’s face corresponding to horizontal features of the subject’s face, as represented by reference numeral 202 in Figure 2. In another example, a second set of features on the subject’s face of the captured image may be identified, wherein the second set of features includes at least two areas (denoted as circles on the subject’s face in captured image 200) on the subject’s face corresponding to vertical features of the subject’s face, as represented by reference numeral 204 in Figure 2. Additional sets of features may be identified as well, for example, to include areas on the subject’s face corresponding to horizontal or other features of the subject’s face. Advantageously, a unique alignment of the subject’s face, and a corresponding unique alignment of the EEG cap 102 based on the subject’s face, can be determined depending on the specific features of that subject’s facial features. Different alignment can be determined and / or different misalignments can be identified depending on the subject’s facial features. For example, one subject might have a crooked nose whereas a further subject may have a different separation between their eyes in any dimension compared to another subjects. By identifying the horizontal and vertical features of the subject’s face, any variations in the subject’s face can be taken into account when identifying misalignments of the EEG cap.

[0046] The first set of features 202 and the second set of features 204 must each include at least two areas in order to determine a horizontal value (for example, as shown by 304a - 304f in Figure 3) and a vertical value (for example, as shown by 302a - 302f in Figure 3). Accordingly, the first set of features may include the subject’s right temple and left temple, the subject’s left corner of left eye and right comer of right eye, the subject’s right comer of left eye and left corner of right eye, or any combination of the above. A line may be drawn (for example, by a computer, a neural network, a machine learning algorithm, etc.) from a first of the at least two areas to a second of the at least two areas to represent a horizontal value of the subject’s face. The second set of features may include at least two of the subject’s highest point of the nose, point of the nose tip (i.e. the closest point of the subject’s face to the image capturing device 106), lowest point of the nose (i.e. the point between the subject’s nostrils where the noise meets the philtrum), or the point of the chin. A line may be drawn (for example, by a computer, a neural network, a machine learning algorithm, etc.) from a first of the at least two areas to a second of the at least two areas to represent a vertical value of the subject’s face.

[0047] In an embodiment, the first set of features 202 is not limited to the subject’s temple and / or eyes but can also include areas relating to the subject’s mouth as indicated by 206 in captured image 200.

[0048] Advantageously, different pairs of features of a subject’s face can be taken into account when identifying misalignments of the EEG cap and / or determining how to align the EEG cap relative to the subject’s face. For example, this may be advantageous because an example subject might have more pronounced eyes (thus providing a more accurate result when determining the horizontal features based on their eyes) whereas a different subject may have more pronounced temples (thus providing a more accurate result when determining the horizontal features based on their temples).

[0049] As mentioned above, determining the alignment of the subject’s face may include determining the horizontal value from the first set of features 202 on the subject’s face of the captured image and determining a vertical value from the second set of features on the subject’s face of the captured image. Figure 3 shows a series of six example captured images 300 where respective horizontal values 304a - 304f and vertical values 302a - 302f for each of the respective six captured images 300 are depicted. The horizontal value is determined as an alignment metric by connecting each of the at least two areas of the first set of features 202, 206 and by taking a mean angle between each of the lines connecting each of the respective areas. As shown in Figure 3, this corresponds to a substantially horizontal line for each example captured image. The vertical value is determined as an alignment metric by connecting each of the at least two areas of the second set of features 204 and by taking a mean angle between each of the lines connecting each of the respective areas. As shown in Figure 3, this corresponds to a substantially vertical line for each example captured image.

[0050] Determining the alignment of the subject’s face may further include comparing the horizontal value to a predetermined horizontal value (for example, a horizontal line) and comparing the vertical value to a predetermined vertical value (for example, a vertical line), and determining an alignment value of the subject’s face of the captured image based on the comparison of the horizontal value to the predetermined horizontal value and the vertical value to the predetermined vertical value. Advantageously, the current alignment of the subject’s face can be compared to a ‘base value’ where the ‘base value’ is an ‘ideal’ alignment of a “sample face” (for example, the subject looking directly at the camera) at which more accurate measurements of the EEG cap relative to the subject’s face can be taken. By determining how aligned the subject’s face is to the “sample face”, an indication can be provided (for example, a prompt on a screen) to align the subject’s face in the horizontal and / or vertical axis to come as close as possible to the alignment of the “sample face”.

[0051] In an embodiment it may be advantageous to ensure that the alignment of the subject’s face is within a pre-determined threshold (i.e. when the subject’s face is aligned straight / orthogonal to the camera) before the alignment of the EEG cap 102 is measured. That way the alignment procedure can be fine-tuned to that of the EEG cap 102 without having to take into account issues of parallax or other calculations due to the subject’s face facing in a direction that is not substantially orthogonal. The pre-determined threshold may be met when it has been determined that the subject is looking straight ahead towards the image capturing device 106. This can be determined by comparing the horizontal value (of the captured image) to the predetermined horizontal value and by comparing the vertical value (of the captured image) to the predetermined vertical value. If the horizontal value is parallel (or within a threshold level of parallelism) to the predetermined horizontal value, and if the vertical value is parallel (or within a threshold level of parallelism) to the vertical horizontal value, then it may be determined that the subject’s face is aligned.

[0052] In Figure 3, each of the six example captured images (numbered 1 through 6) have an alignment metric which corresponds to a “mean angle of angles” between the vertical line and the horizontal line of each of the six example captured images. A lower calculated alignment corresponds to a higher degree of alignment of the subject’s face. For example, example image number 6 is the most aligned because it has an alignment metric of 1.02, whereas example image number 1 is the least aligned because it has an alignment metric of 6.10.

[0053] To assist in determining (from the captured image) an alignment of the EEG cap relative to the alignment of the subject’s face, face position features may be determined (as outlined in the captured image 400 of Figure 4) and EEG cap position features may be determined (as outlined in the captured image 500 of Figure 5).

[0054] The face position features may be derived from the first set of features and the second set of features as described above. The face position features may overlap from the horizontal value and the vertical value as described above. The purpose of the horizontal value and the vertical value is to determine the alignment of the subject’s face. The purpose of the face position features is to help with the determination of the EEG cap 102 alignment relative to the subject’s face. The face position features comprises horizontal face position features 404 and a vertical face position features 402 as shown in captured image 400.

[0055] A horizontal face position feature 404 is a line drawn from a first of the first set of features to a second of the first set of features. In the example feature shown in Figure 4, this corresponds to a line (herein referred to as the “eye line” or leye) which passes through the inner corners of the eyes (i.e. the right comer of the left eye and the left comer of the right eye). However, this horizontal face position feature 404 may be a line drawn from any set of the first set of features, as described above. The set may, for example, be the right temple and left temple, the left corner of left eye and right corner of right eye, the right comer of left eye and left corner of right eye, or the left and right comers of the mouth.

[0056] A further horizontal face position feature 404 is the distance of the line drawn from the first of the first set of features to the second of the first set of features (for example, the distance between the inner comers of the eyes, or other). This will be referred to herein as deye. A further horizontal face position feature 404 is the central most point on the line drawn from the first of the first set of features to the second of the first set of features.

[0057] A vertical face position 402 is a line drawn from a first of the second set of features and a second of the second set of features. In the example feature shown in Figure 4, this corresponds to a line (herein referred to as the “nose line” or ln0Se) which is perpendicular to the “eye line” and crosses the “eye line” in the center (e.g. at the central most point of the “eye line”). However, this vertical face position feature 402 may be a line drawn from any set of the second set of features, as described above. The set may, for example, be any two of the highest point of the nose, the point of the nose tip, the lowest point of the nose, and the point of the chin.

[0058] A further vertical face position 402 is the distance of the line drawn from the first of the second set of features and the second of the second set of features (for example, the distance between the highest point of the nose and the lowest point of the nose, or other). This will be referred to herein as dnOse.

[0059] The EEG cap features may be derived from the alignment marker 104 as shown in the captured image 500 of Figure 5. The EEG cap features comprise a relatively horizontal line 504 (herein referred to as Imarker bottom) that passes through the bottom edge of the alignment marker 104 (i.e., passing through the lowermost left comer and the lowermost right corner of the alignment marker 104). The term “relatively horizontal” is used herein to refer to represent a horizontal feature, even if the line is not perfectly horizontal. That is because the “relatively horizontal line” may be horizontal if, for example, the EEG cap is perfectly aligned. However, it may not be perfectly horizontal but still horizontal relative to the normal position of the EEG cap when the EEG cap is not aligned. Therefore, “relatively” horizontal may include a line that is up to and possibly more than 45° skewed from the ideal horizontal line, but still represents a line passing through the bottom edge of the alignment marker 104.

[0060] The EEG cap features further comprise a marker bottom line 506 (herein referred to as marker bottom- The marker bottom line (also referred herein as “bottom line”) is a relatively horizontal line from the lowermost left comer to the lowermost right corner of the alignment marker 104. The EEG cap features further comprise the center of the bottom line 506 (i.e. a point on the bottom line 506 that is equidistant to the lowermost left corner and the lowermost right corner of the alignment marker 104) which is referred to herein as pmarker bottom.

[0061] The EEG cap features further comprise a relatively vertical line 502 (herein referred to aS Imarker vert ). The vertical line 502 is perpendicular to the relatively horizontal line 504 (and, because the bottom line 506 is part of the relatively horizontal line 504, it is also perpendicular to the bottom line 506) and crosses the relatively horizontal line 504 at the center of the bottom line 506 (i.e. at pmarker bottom). Similarly to above, the term “relatively vertical” is used herein to refer to represent a vertical feature, even if the line is not perfectly vertical. That is because the “relatively vertical line” may be vertical if, for example, the EEG cap is perfectly aligned. However, it may not be perfectly vertical but still vertical relative to the normal position of the EEG cap when the EEG cap is not aligned. Therefore, “relatively” vertical may include a line that is up to and possibly more than 45° skewed from the ideal vertical line, but still represents a line passing through the bottom edge of the alignment marker 104.

[0062] To accurately determine an alignment of the EEG cap 102 relative to the alignment of the subject’s face, the determination may include determining a horizontal value of the alignment marker relative to the horizontal features of the subject’s face. Determining an alignment of the EEG cap relative to the alignment of the subject’s face may further include determining a vertical value of the alignment marker relative to the vertical features of the subject’s face. Due to the convex shape of the human head, it may be advantageous to not only take into account the horizontal and / or vertical distances of the EEG cap relative to a threshold value, but also the horizontal and / or vertical angular rotation of the EEG cap relative to a threshold value. Accordingly, the vertical value of the alignment marker 104 may include a vertical distance, herein referred to as dvert. The vertical distance is a shortest distance between a point of the alignment marker that is on the bottom line 506 and is equidistant to the lowermost left corner and the lowermost right corner of the alignment marker 104. In an example, dvert may therefore correspond to the vertical distance between pmarker bottom and the center of leye. Additionally or alternatively, the vertical value of the alignment marker may include a vertical angle (herein referred to as avert) between the bottom line 506 and the line drawn from the first of the first set of features 202, 206 to the second of the first set of features 202, 206. Advantageously, the method can determine the vertical alignment value with a simple formula which provides a more robust, simple and relatively easy method of indicating an alignment / misalignment of the EEG cap relative to the subject’s face.

[0063] The horizontal value of the alignment marker 104 may include a horizontal distance, herein referred to as dhor. The horizontal distance is a shortest distance between a vertical line 502 that is perpendicular to the bottom line 506 and passes through a point of the alignment marker (that is on the bottom line and is equidistant to the lowermost left corner and the lowermost right comer of the alignment marker - for example, pmarker bottom), and a line drawn from a first of the second set of features and a second of the second set of features. In an example, dhor may therefore correspond to the horizontal distance between Inose and I marker vertical • Additionally or alternatively, the horizontal value may include a horizontal angle (herein referred to as cihor) between the vertical line 502 and the line 402 drawn from the first of the second set of features 204 and the second of the second set of features 204. Advantageously, the method can determine the horizontal alignment value with a simple formula which provides a more robust, simple and relatively easy method of indicating an alignment / misalignment of the EEG cap relative to the subject’s face.

[0064] Distances (both horizontal and vertical) may vary depending on how far the subject sits from the image capturing device 106. Furthermore, vertical distances can be different for different sizes of heads. Therefore, it is advantageous to convert dvert and dhor into dimensionless metrics to avoid these variations which may occur during alignment calculations. A dimensionless metric for dvert (corresponding to a vertical distance relative to eye distance) may be formulated as dvert= . A dimensionless metric for dhor (corresponding to a horizontal eye distance relative to a width of the alignment marker 104) may be formulated as dhor=

[0065] All of the above measures of the alignment of the EEG cap 102 relative to thewmarkerho ttom subject’s face will have an improved accuracy when the measurements are taken after it is determined that the subject’s face is aligned, as described above.

[0066] Figure 6 shows a series of six example captured images 600 where the alignment of the EEG cap 102 relative to the subject’s face was measured. In each of the six example captured images, the respective identified relatively vertical line 502 (Imarkervert) is shown by 502a - 502f and the respective identified relatively horizontal line 504 (Imarker bottom) is shown by 504a - 504f. Figure 6 also shows the horizontal value of the subject’s face (from the subject’s face alignment) 506a - 506f and the vertical value of the subject’s face (from the subject’s face alignment) 508a - 508f. When the EEG cap aligned relative to the subject’s face, the vertical lines (i.e. the vertical line 502* of the EEG cap and the vertical line 508* of the subject’s face) and the horizontal lines (i.e. the horizontal line 504* of the EEG cap and the horizontal line 506* of the subject’s face) should overlap as can be seen, for example, in example image 6 of Figure 6. In the example images 1 to 5 there are more distinct differences between the horizontal lines 504* and 506* indicating that a horizontal misalignment is present (which can be measured by dhor and etho as described above). In the example images 1 to 5 there are also varying differences between the vertical lines 502* and 508* indicating that a vertical misalignment is present (which can be measured by dvert and avert as described above). In this paragraph, ‘a’ corresponds to the first image, ‘f corresponds to the sixth image, and ‘b’ - ‘e’ correspond to the second to fifth images, respectively.

[0067] Figure 7 depicts a first graph 702 corresponding to a distribution of threshold values for aVert, a second graph 704 corresponding to a distribution of threshold values for dvert(the dimensionless metric for dvert), a third graph 706 corresponding to a distribution of threshold values for ahor, and a fourth graph 708 corresponding to a distribution of threshold values for dhor(the dimensionless metric for dhor). The threshold values in the graphs of Figure 7 are based on statistics taken from 159 experiments, where the alignment was manually measured by a test person (i.e. with an operator mounting an EEG cap 102 onto a subject, the operator sending a captured image to a separate person, who validates the alignment of the subject’s face and also validates the alignment of the EEG cap 102) and entered into a data collection system. Based on the statistics collected a deviation for each of the above-mentioned horizontal and vertical values can be measured and, more specifically, thresholds for EEG cap alignment relative to a subject’s face alignment could be determined (i.e. thresholds for determining that an EEG cap 102 is aligned and / or is not aligned). The thresholds in the graphs 702, 704, 706 and 708 are based on best possible values. They are not too strict and are instead, “achievable”. They are also not too coarse, to maintain accurate EEG readings.

[0068] Accordingly, an EEG cap 102 is determined to be aligned correctly if the alignment of the EEG cap 102 relative to the alignment of the subject’s face is within a predetermined threshold alignment. The predetermined threshold alignment may include at least one of, and any combination of the vertical distance (either dvert or the dimensionless metric) being within a predetermined threshold, the vertical angle (avert) being within a predetermined threshold, the horizontal distance (either dhor or the dimensionless metric) being within a predetermined threshold, and the horizontal angle (cthor) being within a predetermined threshold. Advantageously, the alignment / misalignment of the EEG cap can be adjusted (in the horizontal direction and / or in the vertical direction) depending on the necessity of alignment. Some EEG readings require more alignment than others. Thus, the method can be applied to systems that require a higher degree of alignment as well as those that require a lower degree of alignment.

[0069] The graphs in Figure 7 show a metric value on the x-axis and the count of patient numbers (out of 159) on the y-axis. A higher number of patients for a certain threshold value indicates a likely value at which a sufficient alignment of the EEG cap relative the subject’s face can be achieved. Accordingly, and using standard deviation of the distribution, a threshold range and a threshold value can be determined from the collected data which is displayed above each graph for each value.

[0070] Graph 702 depicts a threshold value for avert, indicating that the predetermined threshold of the vertical angle may be greater than or equal to 0 degrees and less than or equal to + / -5 degrees (or 5 degrees as an absolute value). A more accurate predetermined threshold may be greater than or equal to 0 degrees and less than or equal to 2.25 degrees. An ideal predetermined threshold may be at 0 degrees.

[0071] Graph 704 depicts the dimensionless metric of dvertindicating that the predetermined threshold of the vertical distance may be greater than or equal to 0.9 and less than or equal to 1.9 when the vertical distance is a dimensionless measure. A more accurate predetermined threshold may be greater than or equal to 1.24 and less than or equal to 1.50. An ideal predetermined threshold may be 1.35.

[0072] Graph 706 depicts a threshold value for cthor indicating that the predetermined threshold of the horizontal angle may be greater than or equal to 0 degrees and less than or equal to + / -5 degrees (or 5 degrees as an absolute value). A more accurate predetermined threshold may be greater than or equal to 0 degrees and less than or equal to 2.0 degrees. An ideal predetermined threshold may be 0 degrees.

[0073] Graph 708 depicts a threshold value for the dimensionless metric of dhor indicating that the predetermined threshold of the horizontal distance may be greater than or equal to -0.3 and less than or equal to 0.3 when the horizontal distance is a dimensionless measure. A more accurate predetermined threshold may be greater than or equal to -0.09 and less than or equal to 0.06. An ideal predetermined threshold may be 0.

[0074] Figure 8 shows two examples 802 and 804 of suboptimal positions of the EEG cap 102 relative to the subject’s face (i.e. within the thresholds of Figure 7). As can be seen from those examples, some overlap in the horizontal lines 502x and 502y and the vertical lines 504x and 504y may be acceptable. This is especially true when compensating for individual physical characteristics of a subject’s face.

[0075] Figure 9 shows two examples 902a, 902b where the EEG cap 102 is placed relatively aligned to the subject’s face but is significantly lower than expected. This may occur due to a large EEG cap 102 being placed on a relatively small subject head. Such a reading may determine that a different sized EEG cap is required to take an accurate EEG reading. Alternatively, alignment of the EEG cap can be determined solely based on the horizontal and vertical angles, rather than the horizontal and vertical distances as described above. Figure 9 also shows two examples 904a, 904b where the EEG cap 102 is placed relatively aligned to the subject’s face but significantly higher than expected. As with the examples 902a and 902b, this may occur due to the EEG cap 102 being a wrong size for the subject. An accurate alignment of the EEG cap 102 relative to the subject’s face can still be determined using the predetermined threshold data described above with regard to Figure 7 to determine that (as with examples 904a and 904b) the EEG cap 102 is aligned with the subject’s face. Accordingly, there is a benefit in being able to measure both the angles and the distances (in both horizontal and vertical alignments) to accommodate for unexpected scenarios.

[0076] The arrangements and methods described above (in relation to figures 1 to 9) can be implemented in an automated quality control system. The various advantages described above can also be used to control the quality of EEG cap 102 mounting by casual users of neuroheadsets, as well as neuroheadsets combined with virtual and / or augmented reality helmets, either in combination with a separate image capturing device or with an image capturing device that is built into the virtual and / or augmented reality headsets. Figure 10 depicts a flow-chart of a method 1000 for aligning an electroencephalography (EEG) cap according to the arrangements as described above with relation to figures 1 to 9 and can be carried out by the system of Figure 1.

[0077] At step 1002 an image of a subject’s face ad an EEG cap is captured, wherein the EEG cap is placed on the subject’s head. At step 1010, a first set of features (as described above) on the subject’s face of the captured image may be identified, wherein the first set comprising at least two areas on the subject’s face corresponding to horizontal features of the subject’s face (as described above). At step 1012 a second set of features on the subject’s face of the captured image may be identified, the second set comprising at least two areas on the subject’s face corresponding to vertical features of the subject’s face (as described above).

[0078] At step 1004 an alignment of the subject’s face is determined from the captured image. At step 1014 a horizontal value from the first set of features on the subject’s face of the captured image may be determined (as described above). At step 1016 a vertical value from the second set of features on the subject’s face of the captured image may be determined (as described above). The horizontal value may be compared to a predetermined horizontal value. The vertical value may be compared to a predetermined vertical value. At step 1018, an alignment value may be determined of the subject’s face of the captured image based on the comparison of the horizontal value to the predetermined horizontal value and the vertical value to the predetermined vertical value (as described above). At step 1020 the subject’s face may be determined to be aligned when the alignment value is below a pre-determined alignment threshold (as described above).

[0079] At step 1006 an alignment of the EEG cap relative to the alignment of the subject’s face is determining from the captured image. At step 1022, a horizontal value of an alignment marker relative to the horizontal features of the subject’s face may be determined (as described above). At step 1024 a vertical value of the alignment marker relative to the vertical features of the subject’s face may be determined (as described above).

[0080] At step 1008 the EEG cap is indicated to be aligned correctly if the alignment of the EEG cap relative to the alignment of the subject’s face is within a predetermined threshold alignment.

Claims

CLAIMS1. A method (1000) for aligning an electroencephalography, EEG, cap comprising: capturing (1002) an image of a subject’s face and an EEG cap, wherein the EEG cap is placed on the subject’s head; determining (1004), from the captured image, an alignment of the subject’s face; determining (1006), from the captured image, an alignment of the EEG cap relative to the alignment of the subject’s face; and indicating (1008) that the EEG cap is aligned correctly if the alignment of the EEG cap relative to the alignment of the subject’s face is within a predetermined threshold alignment.

2. The method of claim 1, wherein capturing the image comprises: capturing the image with an image capturing device; and storing the captured image on memory.

3. The method of claims 1 or 2, further comprising: identifying (1010) a first set of features on the subject’s face of the captured image, the first set comprising at least two areas on the subject’s face corresponding to horizontal features of the subject’s face; and identifying (1012) a second set of features on the subject’s face of the captured image, the second set comprising at least two areas on the subject’s face corresponding to vertical features of the subject’s face.

4. The method of claim 3, wherein: the first set of features comprise the subject’s: right temple and left temple, left corner of left eye and right comer of right eye, right corner of left eye and left comer of right eye, or any combination of the above; and the second set of features comprise at least two of the subject’s: highest point of the nose, point of the nose tip,lowest point of the nose, or point of the chin.

5. The method of claims 3 or 4, wherein determining the alignment of the subject’s face comprises: determining (1014) a horizontal value from the first set of features on the subject’s face of the captured image; determining (1016) a vertical value from the second set of features on the subject’s face of the captured image; comparing the horizontal value to a predetermined horizontal value; comparing the vertical value to a predetermined vertical value; and determining (1018) an alignment value of the subject’s face of the captured image based on the comparison of the horizontal value to the predetermined horizontal value and the vertical value to the predetermined vertical value.

6. The method of claim 5, further comprising determining (1020) that the subject’s face is aligned when the alignment value is below a pre-determined alignment threshold.

7. The method of claims 3 to 6, wherein the EEG cap comprises an alignment marker and wherein determining an alignment of the EEG cap relative to the alignment of the subject’s face comprises: determining (1022) a horizontal value of the alignment marker relative to the horizontal features of the subject’s face; and determining (1024) a vertical value of the alignment marker relative to the vertical features of the subject’s face.

8. The method of claim 7, wherein the vertical value comprises: a vertical distance, wherein the vertical distance is a shortest distance between: a point of the alignment marker that is on a bottom line and is equidistant to the lowermost left corner and the lowermost right corner of the alignment marker, wherein the bottom line is a relatively horizontal line from the lowermost left comer and the lowermost right comer of the alignment marker, anda central most point on a line drawn from a first of the first set of features to a second of the first set of features; and / or a vertical angle between the bottom line and the line drawn from the first of the first set of features to the second of the first set of features.

9. The method of claims 7 or 8, wherein the horizontal value comprises: a horizontal distance, wherein the horizontal distance is a shortest distance between: a vertical line that is perpendicular to the bottom line and passes through a point of the alignment marker that is on the bottom line and is equidistant to the lowermost left corner and the lowermost right corner of the alignment marker, and a line drawn from a first of the second set of features and a second of the second set of features; and / or a horizontal angle between the vertical line and the line drawn from the first of the second set of features and the second of the second set of features.

10. The method of claim 8, wherein the predetermined threshold alignment comprises: the vertical distance being within a predetermined threshold; and / or the vertical angle being within a predetermined threshold11. The method of claim 10, wherein the predetermined threshold of the vertical distance is greater than or equal to 0.9 and less than or equal to 1.9 when the vertical distance is a dimensionless measure, wherein the dimensionless measure of the vertical distance is: the vertical distance divided by the distance of the line drawn from the first of the first set of features to the second of the first set of features.

12. The method of claims 10 or 11, wherein the predetermined threshold of the vertical angle is greater than or equal to 0 degrees and less than or equal to 5 degrees.

13. The method of claims 8 or 10 to 12, wherein the predetermined threshold alignment comprises: the horizontal distance being within a predetermined threshold; and / or the horizontal angle being within a predetermined threshold.

14. The method of claim 13, wherein the predetermined threshold of the horizontal distance is greater than or equal to -0.3 and less than or equal to 0.3 when the horizontal distance is a dimensionless measure, wherein the dimensionless measure of the horizontal distance is: the horizontal distance divided by the length of the bottom line.

15. The method of claims 13 or 14, wherein the predetermined threshold of the horizontal angle is greater than or equal to 0 degrees and less than or equal to 5 degrees.

16. A computer-implemented method comprising the method of claims 1 to 15.

17. A system (100) comprising: an electroencephalography, EEG, cap (102); an image capturing device (106); and computing device operable to carry out the method of claims 1 to 15.

18. The system of claim 17, further comprising an alignment marker (108) placed on the EEG cap.

19. The system of claim 18, wherein the alignment marker is any one of: a barcode; a QR code; an AruCo marker; or a combination of the above.

Images