EP3713475A1 - System for real-time measurement of the activity of a cognitive function and method for calibrating such a system - Google Patents
System for real-time measurement of the activity of a cognitive function and method for calibrating such a systemInfo
- Publication number
- EP3713475A1 EP3713475A1 EP18811176.9A EP18811176A EP3713475A1 EP 3713475 A1 EP3713475 A1 EP 3713475A1 EP 18811176 A EP18811176 A EP 18811176A EP 3713475 A1 EP3713475 A1 EP 3713475A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- subject
- activity
- cognitive function
- representative
- task
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000000694 effects Effects 0.000 title claims abstract description 153
- 230000003920 cognitive function Effects 0.000 title claims abstract description 118
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000005259 measurement Methods 0.000 title abstract description 20
- 238000012360 testing method Methods 0.000 claims abstract description 63
- 230000001537 neural effect Effects 0.000 claims abstract description 44
- 239000003550 marker Substances 0.000 claims abstract description 37
- 238000010801 machine learning Methods 0.000 claims abstract description 6
- 230000003936 working memory Effects 0.000 claims description 51
- 238000012545 processing Methods 0.000 claims description 9
- 230000003595 spectral effect Effects 0.000 claims description 9
- 230000006870 function Effects 0.000 claims description 6
- 238000001228 spectrum Methods 0.000 claims description 4
- 230000001149 cognitive effect Effects 0.000 description 11
- 210000004556 brain Anatomy 0.000 description 5
- 230000002459 sustained effect Effects 0.000 description 4
- 108010076504 Protein Sorting Signals Proteins 0.000 description 3
- 230000004397 blinking Effects 0.000 description 3
- 238000010276 construction Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 230000005284 excitation Effects 0.000 description 3
- 238000002599 functional magnetic resonance imaging Methods 0.000 description 3
- 238000003384 imaging method Methods 0.000 description 3
- 230000035945 sensitivity Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000001419 dependent effect Effects 0.000 description 2
- 238000002567 electromyography Methods 0.000 description 2
- 238000002570 electrooculography Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 210000003205 muscle Anatomy 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 208000019914 Mental Fatigue Diseases 0.000 description 1
- 241001465754 Metazoa Species 0.000 description 1
- 238000004497 NIR spectroscopy Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000004872 arterial blood pressure Effects 0.000 description 1
- 238000003556 assay Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000008094 contradictory effect Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 208000035475 disorder Diseases 0.000 description 1
- 238000002565 electrocardiography Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000004424 eye movement Effects 0.000 description 1
- 238000012880 independent component analysis Methods 0.000 description 1
- 238000002582 magnetoencephalography Methods 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 230000003446 memory effect Effects 0.000 description 1
- 230000003340 mental effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008450 motivation Effects 0.000 description 1
- 230000004118 muscle contraction Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000003071 parasitic effect Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000029058 respiratory gaseous exchange Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000010187 selection method Methods 0.000 description 1
- 238000002604 ultrasonography Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/369—Electroencephalography [EEG]
- A61B5/372—Analysis of electroencephalograms
- A61B5/374—Detecting the frequency distribution of signals, e.g. detecting delta, theta, alpha, beta or gamma waves
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/369—Electroencephalography [EEG]
- A61B5/377—Electroencephalography [EEG] using evoked responses
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7246—Details of waveform analysis using correlation, e.g. template matching or determination of similarity
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/011—Arrangements for interaction with the human body, e.g. for user immersion in virtual reality
- G06F3/015—Input arrangements based on nervous system activity detection, e.g. brain waves [EEG] detection, electromyograms [EMG] detection, electrodermal response detection
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/369—Electroencephalography [EEG]
- A61B5/377—Electroencephalography [EEG] using evoked responses
- A61B5/378—Visual stimuli
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/40—Detecting, measuring or recording for evaluating the nervous system
- A61B5/4058—Detecting, measuring or recording for evaluating the nervous system for evaluating the central nervous system
- A61B5/4064—Evaluating the brain
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7221—Determining signal validity, reliability or quality
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
Definitions
- the invention relates to a brain-machine interface system and more particularly to a system for measuring in real time the activity of a cognitive function, as well as a method for calibrating such a system.
- Brain-Machine Interface (CIM) systems allow communication between the brain and its environment. These systems are used in known ways to allow an individual to interact with his environment from a reading and interpretation of brain waves of a subject. Brain-machine interface systems have been more recently used to read or measure cognitive function characteristics.
- Mora Sànchez et al. (Mora Sanchez, AM, Gaume, A., Dreyfus, G., & Vialatte, FB, 2015, September, A cognitive brain-computer interface prototype for the continuous monitoring of visual working memory load, 2015 IEEE 25th International Workshop on Machine Learning for Signal Processing (MLSP), pp. 1-5, IEEE) describe a system for evaluating the activity of working memory from electrophysiological signals recorded in a database. Calibration of such a system comprises the steps:
- Reading biases such as blinking, manually or automatically by an independent component analysis
- extraction of marker values depending on a state of high or low activity of the working memory such as, for example, the power measured in a given frequency range, on a given acquisition channel, from the signals electrophysiological;
- the system is tested and the activity of the working memory is evaluated from prerecorded electrophysiological signals processed by the classifier. These signals are acquired beforehand during a known task performed by a subject, resulting in low and / or high activity states of the working memory: it is then possible to test the sensitivity or specificity of the classifier thus constructed.
- the system described does not make it possible to measure in real time the activity of a cognitive function of a subject, in particular the working memory.
- the signal measured at the output of the classifier does not allow an accurate evaluation of the activity of the working memory.
- the signal measured at the output of the classifier may depend on the activity of other cognitive functions of the subject during the execution of a task, such as attention or excitation, for example.
- the activity of the working memory thus evaluated may vary and / or present contradictory values as a function of the electrophysiological signals tested.
- An object of the invention is to propose a solution for increasing the accuracy of the measurement of the activity of a function cognitive of a subject a test.
- Another object of the invention is to measure the activity of a cognitive function, such as working memory, in real time.
- each reference electrical signal being representative of the neuronal activity of a reference subject of a first reference population during the execution of the first task by the reference subject; the values of the markers being representative of a state of activity of the cognitive function of the subject to be tested;
- step c) generating a plurality of copies of marker values calculated in step b) and adding noise to the generated copies; d) constructing a classifier by machine learning from the marker values calculated in step b) and noisy copies calculated in step c), the classifier being adapted to measure the cognitive function of the subject to be tested by calculating a value representative of a probability that an electrical signal representative of the neuronal activity of the test subject results from a state of predetermined activity of the cognitive function of the subject to be tested. Since noisy copies of the marker values calculated in step b) are generated in step c), it is possible to control the proportion of signals representative of the neuronal activity of the test subject in the set of signals. in order to reduce the measurement error related to the classifier.
- the variation distribution of the signals can be parameterized to improve the statistical learning of the classifier.
- the cognitive function is the working memory
- the values of the markers are representative of a state of low activity or a state of high activity of the cognitive function of a reference subject
- the markers are ordered according to their correlation with the states of the activity of the cognitive function, determined from the values of the markers and noisy copies of the marker values, then selecting some of the markers ordered according to their rank, step d) being implemented from the values of the selected markers;
- step d) is implemented solely from the values of the selected markers or only from the values of the selected markers and the noisy copies of the values of the selected markers;
- the first task is configured so that its execution by a subject alternately entails at least two states of activity different from the cognitive function of the subject; the first task is configured so that its execution by a subject alternately causes a state of low activity and a state of high activity of the cognitive function of the subject;
- a second task is configured so that its execution by the subject results in simultaneous states of low activity of the cognitive function and high activity of a confusion function, the method comprising steps of:
- the second task is configured so that its execution by a subject alternately causes at least two different states of activity of the cognitive function of the subject;
- the second task is configured so that its execution by a subject alternately causes a state of low activity and a state of high activity of the cognitive function of the subject;
- the value of one of the markers is a value representative of a spectral power of an electrical signal, calculated on at least a part of the frequency spectrum of the signal;
- the part of the frequency spectrum of the signal is chosen from the range a, the range 6, the range g and the range Q;
- the electrical signals representative of the neuronal activity of the subject to be tested are acquired by means of electrodes arranged at the positions Fp1 and / or Cz and / or Oz and / or CP5 of the 10-20 system of the international standard for placement of electrodes.
- Another subject of the invention is a method for measuring in real time the activity of a cognitive function of a test subject comprising a step of acquiring electrical signals representative of the neuronal activity of the subject to be tested and a step of measuring in real time the activity of the cognitive function of the subject to be tested by calculating a value representative of the probability that an electrical signal representative of the neuronal activity of the test subject results from a state of activity predetermined cognitive function, using a system for measuring in real time the cognitive function of the test subject, the system having been previously calibrated according to a calibration method as defined above.
- Another object of the invention is a system for measuring in real time the activity of a cognitive function of a test subject comprising:
- each reference electrical signal being representative of the neuronal activity of a subject of reference of a first reference population during the execution of the first task by the reference subject and the values of the markers being representative of a state of activity of the cognitive function of the subject to be tested;
- step c) generating a plurality of copies of marker values calculated in step b) and adding noise to the generated copies; d) constructing a classifier by machine learning from the marker values calculated in step b) and noisy copies calculated in step c), the classifier being adapted to measure the cognitive function of the subject to testing by calculating a value representative of a probability that an electrical signal representative of the neuronal activity of the test subject results from a state of predetermined activity of the cognitive function of the subject to be tested.
- the cognitive function measured by the system is the working memory.
- FIG. 1 illustrates the execution of a task specific to the working memory
- FIG. 2 illustrates a system for measuring in real time the activity of a cognitive function
- FIG. 3 illustrates a method of calibrating a real-time measurement system of the activity of a cognitive function of a subject to be tested according to one embodiment of the invention
- FIG. 4 schematically illustrates domains of the activity of working memory, the activity of cognitive functions driven by different tasks and the activity of cognitive functions caused by confounding factors;
- FIG. 5 illustrates a method for measuring in real time the activity of the cognitive function
- FIG. 6 illustrates the performance characteristic of the classifier corresponding to the execution of the task specific to the real-time working memory
- FIG. 7 illustrates the performance characteristics of the classifier during the execution of a task by a subject in the presence of a state of high activity of confusion factor
- FIG. 8 illustrates the evolution of the probability that an electrical signal representative of the neuronal activity of the test subject results from a state of high activity of the cognitive function of the subject to be tested over time
- FIG. 9 illustrates the difference between the value of markers between a low activity state and a high activity state for different channels.
- working memory denotes a cognitive function responsible for the temporary information available for the processing of information. It is described by Baddeley et al. (Baddeley, AD, & Hitch, G., 1974, Working memory, Psychology of learning and motivation, 8, 47-89) as a cognitive model, the activity of which can be confirmed for example by resonance imaging measurements. Magnetic (from Esposito, M., Aguirre, GK, Zarahn, E., Ballard, D., Shin, RK, & Lease, J., 1998, Functional MRI studies of spatial and nonspatial working memory, Cognitive Brain Research, 7 (1), 1 -13). The working memory is dependent on the ability to hold short-term information, a few seconds or minutes, to perform cognitive operations on this information. A subject may have different levels or states of activity (or loads) of the working memory depending on the nature of the tasks he performs. DETAILED DESCRIPTION OF AN EMBODIMENT
- Figure 1 illustrates the execution of a first task, specific to a particular cognitive function, in this case the working memory.
- a task specific to the working memory is performed on the one hand by subjects of a first reference population, and on the other hand by the subject to be tested during the calibration of the system 1 and / or during a measurement of the activity of the cognitive function.
- the different subjects are placed in front of a computer screen, on which a collection of figures is displayed, the figures being used during the task to be executed.
- the subjects are asked to give a short name to each of the figures in order to become familiar with all the figures.
- Different sets of figures are presented, and each set corresponds to different semantic fields, such as animals or geometric shapes.
- a target in a state of low activity of the working memory, a target consists of two displayed figures and in a state of high activity of the working memory, a target consists of five or six figures displayed.
- a target corresponding to one of the states (or one of the conditions) is presented to the subject.
- the subject is asked to memorize the target.
- the target then disappears, and a sequence of figures of the same set slides from the right of the screen to the left of the screen.
- the frame rate is 222 pixels per second.
- the subject is asked to press a button when he finds the target in the sequence of figures, which is considered a test. If the subject presses the button before the target appears, or if the target is missed, the test ends and is not analyzed.
- a test lasts an average of 25 seconds.
- Figure 1 illustrates an example of a trial in a low activity state of the working memory.
- Panel A of the Figure 1 illustrates a target, consisting of two figures: a triangle and a rhombus.
- Panel B of Figure 1 illustrates the scrolling of a sequence of figures during a test.
- a target corresponding to the other state of the working memory is then presented to the subject, and a test is performed. Both conditions are alternated.
- the verbalization of the figures makes it possible to use a simple method of storage / retrieval of information: an internal repetition of the names of the elements of the target, using a phonological loop, allows the subject to compare the target with the sliding elements. The various subjects are asked to perform this internal repetition so that the encoding of the information is homogeneous between the subjects. Each subject performs 10 essays of four different semantic fields.
- the design of the first task makes it possible to vary the activity of sub-functions of the working memory, such as storing, maintaining and / or processing elements.
- the sequence of figures comprises distractors, that is to say sets of figures whose composition is close to a target.
- a distractor may be formed by the sequence of figures of the target in which a figure is changed from the third figure in the order of appearance on the screen.
- Distractors prevent a subject from memorizing only part of the target to recognize it.
- the scrolling of the figures is programmed so that a distractor appears with the same probability as a target. This programming prevents the subject from learning and expecting the appearance of a target as a result of a distractor.
- the duration of each test is programmed randomly between 15 and 30 seconds so as to avoid the subject to learn the duration of a test.
- the size of the part of the screen in which the figures scroll is 100 pixels by 300 pixels and the size of each figure is 100 pixels by 100 pixels. This small size prevents eye movements subjects, causing parasitic electrophysiological signals unrelated to cognitive activity.
- the part of the screen used, the size of the figures and the scrolling speed are adapted so that the subject can only have one figure at the same time.
- the distance between the subject and the screen is 60 cm.
- the first task is configured so that its execution by a subject alternately leads to a state of low activity and a state of high activity of the cognitive function of the subject: thus, a drift of an electrical signal, independent of the cognitive function studied, can be reduced or avoided.
- a transverse task, different from the first task, can also be performed by a subject.
- the transversal task is used to induce cognitive function activity on the subject to be tested and designed to be more representative of the tasks performed by subjects in their everyday environment.
- the transversal task corresponds for example to a series of mental calculations. It is for example presented to the subject to test a series of digits ranging from di to d n .
- the transversal task consists, for example, in multiplying di to d 2 , storing the result, then using the result by multiplying it to d 3 , and so on until d n .
- a second task is also designed, configured to cause a low activity state of the cognitive function and a high activity state of a confusion function, caused by confounding factors.
- the second task may consist, for example, in the execution of the first task adapted to cause a state of low activity of the cognitive function in a subject, in which a red dot, or a fly, is added along a random trajectory for a period of time. from 1 to 2 seconds.
- This modification with respect to the first task makes it possible to cause a high activity state of a confounding factor of the working memory, such as for example attention. Acquisition of electrophysiological signals
- Electrical signals representative of neuronal activity of a test subject or subject of a reference population may be electrophysiological signals.
- the electrical signals representative of a neuronal activity of a subject to be tested or of a subject of a reference population can be electrical signals derived from optical, ultrasonic or magnetic neuronal imaging, such as, for example, the functional magnetic resonance imaging (fMRI), functional brain ultrasonography, positron emission imaging, and / or near infra-red spectroscopy.
- Electrophysiological signals are recorded on the test subject or on subjects of one or more reference populations, using an EEG device (Brain Products V-Amp, registered trademark) at a sampling frequency of 500 Hz.
- the electrical signals representative of a neuronal activity of a subject to be tested can be electrophysiological signals measured by electrocardiography (ECG), by electromyography (EMG, representative measurement of the activity of the muscles), by electro-oculography (EOG, representative measurement of a difference in electric potential in the eye), magnetoencephalography (MEG) and / or by an arterial pressure sensor, and / or by a respiration sensor.
- ECG electrocardiography
- EMG electromyography
- EOG electro-oculography
- MEG magnetoencephalography
- the electrical signals representative of a neuronal activity of a subject to be tested may be derived from a combination (s) of the electrical signals representative of a neuronal activity of a test subject previously described.
- FIG. 2 illustrates a system 1 for measuring, in real time, the activity of a cognitive function of a test subject comprising an electrophysiological signal acquisition subsystem 2, and a processing unit 3.
- acquisition system 2 comprises, for example, an EEG headset, EEG electrodes and connectors for connecting the EEG electrodes to the treatment unit 3.
- Figure 2 also illustrates the arrangement of the EEG electrodes around the cranial box of the test subject, according to the international standard of the system 10-20.
- the electrical signals of a reference population are acquired from 20 healthy subjects, aged between 21 and 31, including 10 men and 10 women.
- the electrical signals for the real-time measurement of the activity of a cognitive function are also acquired on this population.
- Electrical signals were also acquired on 6 subjects performing a transversal task. As a result of each task, the mental fatigue of each subject is collected and the tests are stopped if the response is positive.
- the analyzed EEG signal sequences have a duration of 2.5 seconds. 1744 distinct sequences are used as electrical reference signals during the calibration of the real-time measurement system 1 and 90 distinct sequences are used as electrical signals representative of a subject to be tested during the calibration of the real-time measurement system 1 , for each subject.
- the electrical signals representative of the neuronal activity of the test subject consist of a continuous stream of EEG signals acquired in real time.
- a set of parameters Pi, P 2 , P3, P4 and P5 can for example characterize the design of the brain-machine interface.
- the duration Pi of the electrical signal sequences of reference equal to 2.5 seconds is chosen.
- the number P 2 of electrical signal sequences representative of a subject to be tested is selected, during calibration, equal to 90 (45 sequences representative of a high activity state and 45 sequences representative of a state of activity low).
- the proportion P 3 of the electrical calibration signals from the test subject is equal to 65%.
- the standard deviation of a Gaussian noise, of zero average, is equal to P 4 multiplied by the standard deviation of the marker considered, where P 4 is for example equal to 1, 5.
- the number of markers P 5 is equal to 8.
- the standard deviation of a marker considered is for example calculated from the values of a marker, themselves calculated from electrical signals corresponding to different measurement sequences in time, during the execution of a spot.
- FIG. 3 illustrates a method of calibrating a system 1 for measuring in real time the activity of a cognitive function of a subject to be tested according to one embodiment of the invention.
- the cognitive function whose activity is measured is advantageously the working memory.
- a step 101 of the method implements the acquisition of electrical signals representative of the cognitive activity of a test subject during the execution of the first task by the subject to be tested.
- the first task is a task specific to the cognitive function measured by the system 1: it can be for example a task specific to the working memory.
- Each reference electrical signal is representative of the neural activity of a reference subject of a first reference population during the performance of the first task by the reference subject.
- the task specific to the working memory described above causes, when executed by a subject, according to the tests, a low activity state of the working memory or a high activity state of the working memory. from subject.
- the frequencies of signals acquired below 1 Hz and above 45 Hz are erased EEG signals using a Butterworth filter of the 3rd order.
- the acquired EEG signals are then segmented into several sequences of one duration equal to Pi seconds. Each sequence is inspected visually and any sequences with too much noise or muscle bias are not taken into account. In particular, sequences including characteristics of blinking are not taken into account.
- the marker values of the activity of the cognitive function are calculated from the signals representative of a neuronal activity of the subject to be tested, and from electrical reference signals. Each signal is segmented into sequences. The value of a spectral marker is calculated for each sequence using the Welsh method, with a window duration of 0.5 seconds. The values of the spectral markers are calculated in absolute power and in relative power in each of the following frequency ranges: ⁇ (from 1 to 4 Hz), Q (from 4 to 8 Hz), a (from 8 to 12 Hz), b bass (12 to 20 Hz) and b high (20 to 30 Hz).
- the relative power in a frequency range corresponds to the ratio of the power in a frequency range to the power for all the frequencies.
- the use of relative power as a marker makes it possible to compare the markers between different subjects in a more relevant way than the use of absolute power would allow. For each sequence, a total of 192 markers is obtained from 16 acquisition channels, two markers per frequency range and 6 frequency ranges.
- at least two types of electrical signals can be used to calculate the marker values, as implemented during a step 102 of the method. signals representative of the neuronal activity of a subject to be tested and electrical reference signals representative of the neuronal activity of a reference subject of a first reference population. Subsequently, a classifier is trained with the values of markers and noisy copies potentially derived from the two types of signals.
- the proportion of the marker values resulting from signals representative of a first reference population with regard to all the values of the markers it is for example possible to limit its proportion.
- the addition of noise to the marker values from signals representative of the subject to be tested makes it possible to simulate a variation distribution of the markers, compatible with the statistical learning of the classifier.
- the noise added to each of the copies is, for example, a Gaussian noise, of zero average, and whose standard deviation is equal to P 4 times the standard deviation of a marker considered.
- it is possible to minimize the output error of a classifier by adjusting the proportion of marker values derived from the signals representative of the neuronal activity of a test subject with regard to all the signals used (controlled by the parameter P 3 ), and adjusting the noise added to the different copies (eg controlled by parameter P 4 ).
- the markers used in step 102 are ordered according to their correlation with the state or states of the activity of the cognitive function predefined by the first task, determined from the values of the markers and noisy copies of marker values. For example, a Gram-Schmidt orthogonalization (or O.F.R. for orthogonal forward regression in English) or generally supervised variable selection methods for ordering the markers may be used.
- the first marker, after classification is the marker whose value (s) have the strongest correlation with the state of the activity of the cognitive function.
- the second marker, after classification is the marker whose value or values have the strongest correlation with the state of the activity of the cognitive function, after having discarded the data associated with the first marker, and so on.
- step 105 select a number of markers among the most relevant in the order defined above. It is possible to test the error of a classifier constructed with a given set of markers as described in step 105: it is possible to select a pre-defined number of markers and to optimize the number of elements to be selected from the most relevant by testing the error of the classifier so as to reduce his error.
- a classifier is constructed by automatic learning from the values of the calculated markers and the noisy copies generated.
- the classifier is constructed from at least the values of the markers selected in step 104, and preferably only from the values of the markers selected in step 104 and the noisy copies of the values of the selected markers during step 104. of step 104 of the process.
- Such a classifier can for example be a linear analysis classifier discriminant. Learning the classifier is also done with the state of activity of the cognitive function that is associated with each of the selected patterns (low or high activity state for example).
- the classifier may, after learning, present an output signal making it possible to measure the activity of the cognitive function of the subject to be tested by calculating a probability PA that an electrical signal representative of the neuronal activity of the test subject results from a high activity state of the cognitive function of the test subject, or more generally, a value representative of the probability PA, such as a binary result calculated from the probability PA. It is also possible to construct a classifier with n classes, where n is a natural integer, to characterize n states. In this case, patterns and pattern values corresponding to each of the n activity states of the cognitive function will have been calculated beforehand, and the first task is configured so that its execution by the subject results in a state of activity of the cognitive function. the cognitive function of the subject of a first reference population among the n possible states. The markers are thus used to predict the state or states of a cognitive function of a subject to be tested by means of an algorithm using the values of these markers.
- Figure 4 schematically illustrates finite number sets of markers for measuring activity states of cognitive functions when performing a predetermined task.
- the task specific to the measured cognitive function is designed so as to cause in a subject two distinct states of activity of the cognitive function, in this case a state of low activity and a state of high activity.
- measured cognitive function such as working memory
- the different signals acquired may have been driven by confounding factors. These confounders can be Cognitive: Attention, excitement and / or frustration can be confusing for working memory. These confounding factors can also be the result of motor events, such as blinking, sub-vocalization and muscle contractions.
- the area delimited by the ellipse (a) schematically illustrates all the markers representative of the execution of a first task, specific to the cognitive function.
- the area delimited by the ellipse (b) schematically represents all the markers representative of the execution of the transverse task.
- the area delimited by the ellipse (c) schematically represents the set of markers that make it possible to measure the state of activity of the cognitive function, in this case the working memory.
- the area delimited by the ellipse (d) schematically represents all the markers that make it possible to measure the state of activity caused by motor confusion factors
- the area delimited by the ellipse (e) represents schematically all the markers that measure the state of activity caused by cognitive confusion factors.
- the overlap area of the ellipses (a) and (b) schematically illustrates a non-empty set of markers representative of the execution of the first task and the transverse task.
- the overlap area of the ellipses (a), (b) and (c) (noted area (f) in the figure) schematically illustrates a non-empty set of markers representative of the execution of the first task, the transverse task and working memory.
- This set is not empty results from the fact that the first task and the transversal task have been designed so as to involve the working memory.
- the overlap area of the ellipses (a), (b) and (e) (noted area (g) in the figure) schematically represents a non-empty set of markers representative of the execution of the first task, the transversal task , and markers that measure the state activity caused by cognitive confusion factors.
- the markers belonging to the latter set are not taken into account during the construction of a classifier adapted to measure the state of cognitive activity caused by the execution of the first task or the transverse spot.
- the probability PA of an electrical signal representative of the neuronal activity of a subject d is calculated with the classifier, for example constructed during step 105 of the method, a second reference population, performing the second task, results from a state of high activity of the cognitive function of the test subject (the electrical signal representative of a neuronal activity of a subject of a second reference population acquired during the execution of the second task by the subject of the second reference population).
- the electrical signals of a second reference population executing the second task may, for example, be acquired beforehand during a step 106.
- a step 108 of the method it is then possible to compare the probability PA obtained with a threshold value V s previously recorded or determined by a user.
- the probability PA is greater than 0.5, preferably greater than 0.6, and preferentially greater than 0.7
- the classifier informs a state of high activity of the cognitive function while the task is specifically designed. to cause a low activity state of the cognitive function.
- This test shows a construction of the classifier that does not discriminate a high activity state of cognitive function and confounding factor.
- This test can be followed, for example, by a new step of acquiring the subject's electrical signals, so as to lead to the construction of a new classifier adapted to discriminating the cognitive function of the subject. confounding factors, as shown in figure 3 for the condition PA> Vs. This test can also be followed by stopping the measurement.
- Figure 5 illustrates a method for real-time measurement of cognitive function.
- a method for measuring in real time the activity of the cognitive function is preceded by a calibration of the measurement system 1, for example according to a method comprising steps 101 to 106 of the calibration method: a classifier is constructed in such a way as to be adapted to measure the activity of the cognitive function of the subject to be tested by calculating a probability PA that an electrical signal representative of the neuronal activity of the test subject results from a state of high activity of the cognitive function of the subject to test.
- electrical signals representative of the neuronal activity of the subject to be tested are acquired.
- a continuous flow of electrical signal is transmitted from the acquisition subsystem 2 to the processing unit 3 and analyzed by the processing unit 3. For example, the signals included in a sliding window with a duration of 2 are used. , 5 seconds.
- the probability PA is calculated using a system 1 for measuring in real time the activity of a cognitive function of the subject to be tested.
- a measure of the activity of the cognitive function implementing the calculation of the probability PA in months of 10 seconds, and preferably in less than 5 seconds.
- the representative value of the probability PA may for example be a binary prediction calculated from the probability PA.
- FIG. 6 illustrates the performance characteristic of the classifier, or ROC curve (for operating characteristic of the receiver in English) corresponding to the execution of the first task, specific to the real-time working memory.
- the value of the probability PA is measured continuously during the duration of each trial of the first task.
- the required sustained activity time is the time during which cognitive function activity is greater than a threshold value to classify an assay as having a high activity state of cognitive function.
- the values of the required sustained activity time are different according to the tests, but on average, an optimum value of this time is between 2 seconds and 10 seconds, preferably between 4 seconds and 6 seconds and preferably substantially equal to 5 seconds.
- the mean area under the curve of the classifier (or AUC for Area Under Curve in English) in real time is 0.78, higher, with p ⁇ 0.0001, at the value of 0.5 corresponding to a random classifier, illustrated by the black line in Figure 6.
- Figure 7 illustrates the performance characteristics of the classifier when performing a task by a subject in the presence of a high activity state of a confounding factor (in this case excitation).
- Curve (a) corresponds to the performance characteristic of the classifier when performing a task by a subject in the presence of a high activity state of confusion factor and curve (b) corresponds to the same characteristic corrected, after decorrelating the information included in the markers of the excitation.
- the area of curve (b) is 7% lower than the area of curve (a), which does not represent a significant difference.
- Figure 8 illustrates the evolution of the probability PA in the time corresponding to the activity of the working memory. The average of the probability PA measured over 20 tests during the execution of a transverse task is illustrated as a function of time.
- the probability PA usually shows a decrease after 10 seconds. This variation is consistent in view of the switching of the activity state of the working memory, from high to low, to 8.5 seconds and the delay introduced by the system 1 of substantially 2.5 seconds.
- the measure of the probability PA is an overall measure of the activity of the working memory during the 2.5 seconds which preceding the time at which the probability PA is measured.
- the probability PA is again increasing, without reaching values as high as during the period corresponding to a high activity state of the working memory.
- Figure 9 illustrates the difference in marker values between a high activity state and a low activity state for different frequency ranges.
- Panel A of FIG. 9 illustrates the difference between, on the one hand, an average of the spectral power in the frequency range a, for all the channels, when performing a task corresponding to a state high activity of the working memory, and, on the other hand, the corresponding average in a low activity state.
- Panel B of FIG. 9 illustrates the difference between, on the one hand, an average of the spectral power in the low frequency range y, for all the channels, when performing a task corresponding to a high activity state of the working memory, and, on the other hand, the corresponding average in a low activity state.
- FIG. 9 illustrates the difference between, on the one hand, an average of the spectral power in the low frequency range B, for all the channels, when performing a task corresponding to a high activity state of the working memory, and, on the other hand, the corresponding average in a low activity state.
- Panel D of FIG. 9 illustrates the difference between, on the one hand, an average of the spectral power in the high frequency range B, for all the channels, when performing a task corresponding to a high activity state of the working memory, and, on the other hand, the corresponding average in a low activity state.
- the average of the spectral powers measured is calculated by integrating the powers for 10 seconds.
- the illustrated frequency ranges correspond to the ranges of selected markers frequencies, for example by a method described by step 105 of the calibration method illustrated in Figure 3.
- the selected markers may be preferably selected from the relative power of the low range 6 acquired by an electrode arranged at the Fp1 position , as described by the system standard 10-20, the relative power of the low range 6 acquired by an electrode arranged at the position Cz, the relative power of the low range g acquired by an electrode arranged at the position Fp1, the relative power of the high range 6 acquired by an electrode arranged at the position Cz, the relative power of the range acquired by an electrode arranged at the position Oz and the relative power of the range acquired by an electrode arranged at the position CP5 .
- the system 1 proposed, as well as the methods for calibrating the system 1 and measuring in real time the cognitive function advantageously find their application in:
Abstract
Description
Claims
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
FR1760970A FR3073727B1 (en) | 2017-11-21 | 2017-11-21 | SYSTEM FOR REAL-TIME MEASUREMENT OF THE ACTIVITY OF A COGNITIVE FUNCTION AND METHOD FOR CALIBRATION OF SUCH A SYSTEM |
PCT/EP2018/082109 WO2019101807A1 (en) | 2017-11-21 | 2018-11-21 | System for real-time measurement of the activity of a cognitive function and method for calibrating such a system |
Publications (1)
Publication Number | Publication Date |
---|---|
EP3713475A1 true EP3713475A1 (en) | 2020-09-30 |
Family
ID=61258373
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
EP18811176.9A Pending EP3713475A1 (en) | 2017-11-21 | 2018-11-21 | System for real-time measurement of the activity of a cognitive function and method for calibrating such a system |
Country Status (5)
Country | Link |
---|---|
US (1) | US11950919B2 (en) |
EP (1) | EP3713475A1 (en) |
JP (1) | JP7206287B2 (en) |
FR (1) | FR3073727B1 (en) |
WO (1) | WO2019101807A1 (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114428555B (en) * | 2022-03-04 | 2024-03-22 | 广州大鱼创福科技有限公司 | Electroencephalogram movement intention recognition method and system based on cortex source signals |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP5476137B2 (en) | 2010-01-19 | 2014-04-23 | 株式会社日立製作所 | Human interface based on biological and brain function measurement |
EP3241489A1 (en) * | 2016-05-04 | 2017-11-08 | Mensia Technologies | Predictive neuromarkers of alzheimer's disease |
-
2017
- 2017-11-21 FR FR1760970A patent/FR3073727B1/en active Active
-
2018
- 2018-11-21 US US16/766,087 patent/US11950919B2/en active Active
- 2018-11-21 EP EP18811176.9A patent/EP3713475A1/en active Pending
- 2018-11-21 WO PCT/EP2018/082109 patent/WO2019101807A1/en unknown
- 2018-11-21 JP JP2020544978A patent/JP7206287B2/en active Active
Also Published As
Publication number | Publication date |
---|---|
JP2021504068A (en) | 2021-02-15 |
WO2019101807A1 (en) | 2019-05-31 |
US11950919B2 (en) | 2024-04-09 |
US20210290142A1 (en) | 2021-09-23 |
FR3073727A1 (en) | 2019-05-24 |
JP7206287B2 (en) | 2023-01-17 |
FR3073727B1 (en) | 2019-10-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP2124736B1 (en) | Electrophysiological analysis system | |
JP2593625B2 (en) | Biological information automatic identification device | |
EP2679151B1 (en) | Method for locating brain activity | |
JP7373555B2 (en) | Quantification of motor function using EEG signals | |
FR2887425A1 (en) | ELECTROPHYSIOLOGICAL ANALYSIS SYSTEM | |
EP3586339A1 (en) | Method, device and program for determining at least one cerebral network involved in carrying out a given process | |
FR3073727B1 (en) | SYSTEM FOR REAL-TIME MEASUREMENT OF THE ACTIVITY OF A COGNITIVE FUNCTION AND METHOD FOR CALIBRATION OF SUCH A SYSTEM | |
EP3593709A1 (en) | Method for determining the state of stress of an individual | |
FR3063425A1 (en) | SYSTEM FOR DETERMINING A USER EMOTION | |
CA3092784A1 (en) | Computing device for detecting heart rhythm disorders | |
US11610073B2 (en) | Device and method for analyzing the state of a system in a noisy context | |
CN113143275B (en) | Electroencephalogram fatigue detection method for quantitative evaluation of sample and characteristic quality in combined manner | |
EP4031005B1 (en) | System for determining an emotion of a user | |
RU2704547C1 (en) | Method for remote collection and group treatment of psychophysiological reactions with presentation of various information | |
EP3599988B1 (en) | Biomarker of the colour perception of a mammal subject based on pupil frequency tagging | |
KR102575596B1 (en) | Electronic device for classifying mental state based on machine learning considering daily stress and method thereof | |
WO2022064425A1 (en) | Electrodermal apparatus | |
US20210290145A1 (en) | Information analysis apparatus and information analysis method | |
EP3841964A1 (en) | Method for determining a function of belonging, intended to be applied for estimating a state of stress of a person. | |
EP3017754B1 (en) | Method for automated determination of an index representing the sympathovagal activity of a human being or a competition animal | |
FR3133691A1 (en) | Data labeling process for building a database to configure, validate and/or test an application for monitoring the fatigue level of an individual | |
EP3017757B1 (en) | Automated method for producing a variation curve of an indicator representing the heart-frequency variability of a human being or a competition animal | |
EP3998945A1 (en) | Method and system for testing cognition by processing the reaction of a subject to stimuli | |
CN115399789A (en) | Visual vertigo electroencephalogram signal classification method, device, equipment and medium | |
JP2023136285A (en) | Information processing device, information processing method, and program |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: UNKNOWN |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE |
|
PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE |
|
17P | Request for examination filed |
Effective date: 20200619 |
|
AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR |
|
AX | Request for extension of the european patent |
Extension state: BA ME |
|
DAV | Request for validation of the european patent (deleted) | ||
DAX | Request for extension of the european patent (deleted) | ||
RAP3 | Party data changed (applicant data changed or rights of an application transferred) |
Owner name: SORBONNE UNIVERSITE Owner name: ECOLE SUPERIEURE DE PHYSIQUE ET DE CHIMIE INDUSTRIELLES DE LA VILLE DE PARIS Owner name: CENTRE NATIONAL DE LA RECHERCHE SCIENTIFIQUE (CNRS) Owner name: PARIS SCIENCES ET LETTRES |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: EXAMINATION IS IN PROGRESS |
|
17Q | First examination report despatched |
Effective date: 20240214 |