Method and system for analyzing brain activity

Abstract

A method and system are provided for constructing a representation of the changes in the brain response state of a mammalian subject to multiple repetitive external stimuli. The method and system include acquiring multiple brain activity measurements of the subject over multiple predetermined time periods. Each of the measurements is a brain activity measurement of the subject before and after the presentation of the external stimulus. The method and system further evaluate, via a processor, the variability of the acquired brain activity measurements to the multiple repetitive external stimuli, and generate a report of changes in the brain response state based on the variability of the brain activity measurements over the multiple predetermined time periods.

Description

Technical Field

[0001] The present invention relates to a method and system for analyzing electrical signals detected from brain activity in response to repetitive stimuli. Background Technology

[0002] Hans Berger's 1929 paper describing a device that could measure the brain's electrical activity by placing electrodes on the scalp, amplifying the signal, and plotting voltage changes over time is widely recognized as the first reported human electroencephalography (EEG). Advances in the number and location of electrodes used in EEG, sampling, analysis, and other techniques have contributed to a deeper understanding of brain activity, and the discovery and understanding of the various components of the signal have provided a rich field for further research, including the identification of the first cognitive event-related potential (ERP) component in the early 1960s. Notably, since Berger's initial discovery, advances in surgery have enabled EEG to be recorded directly from the surface of the brain (electrocortical electroencephalography or ECoG) or from electrodes implanted directly in the brain (stereoscopic electroencephalography or sEEG).

[0003] Other techniques have also been developed to observe brain activity over time in many areas, including cognitive studies. These techniques include blood oxygen-dependent (BOLD) functional magnetic resonance imaging (fMRI), functional near-infrared spectroscopy (fNIRS), and magnetoencephalography (MEG).

[0004] Despite the use of specific techniques, many research methods for analyzing brain activity employ similar experimental protocols involving monitoring brain activity before, during, and after the presentation of one or more stimuli (visual, auditory, or tactile / somatosensory), and quantifying spontaneous or stimulus-locked changes over a series of consecutive trials. A common experimental paradigm used is the "oddball paradigm," where two types of stimulus events are presented (rare and frequent / typical), with the type of stimulus determining the behavioral task the subject is required to perform. In analyzing the results of experiments using the oddball paradigm, and indeed other experimental paradigms, electrical activity detected before, during, and after each stimulus is tracked; this detected electrical activity is filtered and typically averaged over multiple trials; and the resulting readings are compared and further evaluated.

[0005] When using these different techniques and methods to study brain activity associated with cognitive processes, typical analytical paradigms reflect a set of assumptions (overt or otherwise) that require different cognitive perceptions corresponding to stimuli (reflected in measured changes in electrical activity) to be the result of invariant and distinct neural processes. However, within each individual session, even for the same individual, the recorded event-related responses are highly variable, despite various types of preprocessing and artifact removal. This variability in recorded activity is typically addressed by temporal ensemble averaging, which is considered to be caused by “neurophysiological noise.” It should be understood that providing further insights into brain activity (particularly into cognitive processes) has a variety of applications, including monitoring anesthesia, detecting individual vigilance (e.g., heavy vehicle operators), and assessing longitudinal changes in individual brain function (diseases).

[0006] Therefore, the purpose of this disclosure is to address some of the problems and shortcomings of previous approaches, and at least to provide the public with other options. Summary of the Invention

[0007] The features and advantages of this disclosure will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the principles disclosed herein. The features and advantages of this disclosure may be realized and obtained by the means and combinations particularly pointed out in the appended claims.

[0008] According to a first aspect of the present invention, a method is provided for constructing a representation of changes in the response state of the brain of a mammalian subject to multiple repetitive external stimuli. The method may include:

[0009] (i) Acquire multiple brain activity measurements of the subject over multiple predetermined time periods, wherein each of the measurements is a brain activity measurement of the subject before and after the presentation of an external stimulus;

[0010] (ii) Using a processor to assess the variability of acquired brain activity measurements to multiple repetitive external stimuli; and

[0011] (iii) Generate a report on changes in brain response state based on the variability of the brain activity measurements over the plurality of predetermined time periods.

[0012] The variability of said brain activity can be assessed using the probability density function of the following brain activity measurements:

[0013]

[0014] The brain activity measurements are related to the random variable X, which is recorded at multiple physical brain locations at time τ = 0, concerning stimulus presentation at time τ = 0.j (τ) Correspondingly, each physical brain location is indexed by an integer j.

[0015] Advantageously, the probability density function can be estimated from the acquired brain activity measurements by using a method selected from the group including the empirical cumulative distribution function method, the proportional histogram estimation and the nuclear density estimation.

[0016] Measurements of brain activity can be obtained from a modality selected from a set of modalities including electrocorticography (ECoG), electroencephalography (EEG), magnetoencephalography (MEG), blood oxygen level dependent (BOLD) functional magnetic resonance imaging (fMRI), and near-infrared spectroscopy (NIRS).

[0017] Optionally, the stimulus may be selected from a group including auditory, visual, olfactory, or tactile stimuli.

[0018] Measurements can be obtained from multiple brain locations over multiple predetermined time periods.

[0019] Advantageously, it allows for continuous temporal estimation based on multiple sessions of a specified subject. Each session includes measurements of brain activity for multiple repetitive stimuli, and these sessions are spaced out at intervals chosen from multiple hours, days, months, or years. For designated subjects... The continuous estimates provide an indication of changes in brain function between sessions.

[0020] Alternatively, brain activity is stimulus-evoked activity estimated by time set, ERP. j (τ), the stimulus-induced activity can be estimated differently as one of the following functions:

[0021] (i)ERP j (τ)=E[X j (τ)], where E[·] is the expectation operator; or

[0022] (ii) ERP j (τ)=median[X j (τ)];or

[0023] (iii) ERP j (τ)=mode[X j (τ)。

[0024] Alternatively, estimations can be made based on empirical data from bandwidth-limited brain activity measurements.

[0025] Advantageously, a brain response state is determined by differential entropy according to the following equation:

[0026]

[0027] Among them, h j (τ) is the differential entropy of the physical brain location specified by the index j at time τ after the stimulus is presented.

[0028] The method may further include: estimating the differential entropy empirically based on a finite number of samples; and defining the change of the differential entropy relative to a baseline reference value.

[0029] Baseline reference values ​​can include h when τ < 0 (before stimulus presentation) j The value of (τ).

[0030] Alternatively, the differential entropy can be estimated using one of the techniques selected from the group including histogram-based bias correction estimation, kernel density estimation, and k-nearest neighbor estimation.

[0031] Advantageously, h is estimated longitudinally based on multiple conversations of a specified subject. j (τ), where each session includes brain activity measurements for multiple repetitive stimuli, and these sessions are separated by intervals chosen from multiple hours, multiple days, multiple months, or multiple years.

[0032] Optionally, h is plotted relative to the physical brain location, indexed by j. j Topographic map of longitudinal estimation of (τ).

[0033] For designated subjects Continuous assessment provides an indication of brain function during cumulative sessions.

[0034] Advantageously, the variability of brain activity is used to obtain a quantitative information theoretical measure representing the state of brain response.

[0035] Optionally, the quantitative information theory measure representing brain function is selected from the group including negative entropy, differential entropy, "spatial average" differential entropy, Kullback-Leibler divergence and negative entropy transfer entropy, mutual information, relative entropy and multiscale entropy.

[0036] Changes in brain response state are independent of changes in the corresponding time set of ERP amplitude.

[0037] Optionally, a quantitative information theory measure representing brain function can be estimated longitudinally from multiple sessions of a specified subject, wherein each session may include brain activity measurements for multiple repetitive stimuli, and these sessions are separated by intervals selected from multiple hours, multiple days, multiple months, or multiple years.

[0038] Advantageously, a quantitative information theoretical measure representing brain function, estimated longitudinally over multiple sessions of a specified subject, can provide an indication of the specified subject's brain function over a cumulative interval.

[0039] A topographic map of longitudinal estimates of quantitative information theoretical measures can be drawn relative to the physical brain location indexed by j.

[0040] According to a second aspect of the invention, a system is provided for representing changes in the brain response state of a mammalian subject to multiple repetitive external stimuli, comprising:

[0041] (i) An acquisition module, the acquisition module including a processor, the acquisition module being configured to acquire multiple brain activity measurements of a subject over multiple predetermined time periods, wherein each of the measurements is a brain activity measurement of the subject before and after the presentation of an external stimulus;

[0042] (ii) an assessment module, the assessment module including a processor, the assessment module being configured to receive the brain activity measurements and assess the variability of the brain activity measurements over the plurality of predetermined time periods; and

[0043] (iii) A determination module, the determination module including a processor, the determination module being configured to determine changes in brain response state based on the variability over the plurality of predetermined time periods and to generate a report based on the variability.

[0044] Optionally, the variability can be evaluated by a processor in an evaluation module configured to utilize a probability density function of the following brain activity:

[0045]

[0046] The brain activity is related to the presentation of stimuli at time τ = 0, and the random variable X recorded at multiple physical brain locations at time τ. j (τ) Correspondingly, each physical brain location is indexed by an integer j.

[0047] The probability density function can be estimated from the acquired brain activity measurements by using a method selected from a group including the empirical cumulative distribution function method, the proportional histogram estimation, and the nuclear density estimation.

[0048] Optionally, estimation can be performed continuously over time based on multiple sessions of a specified subject. Each session includes measurements of brain activity for multiple repetitive stimuli, and these sessions are separated by intervals chosen from multiple hours, days, months, or years.

[0049] For designated subjects Continuous estimation can provide an indication of changes in brain function between sessions.

[0050] Alternatively, a brain response state is calculated by the processor of a defined module using the differential entropy according to the following equation:

[0051]

[0052] Among them, h j (τ) is the differential entropy of the physical brain location specified by index j at time τ after the stimulus is presented.

[0053] According to a third aspect of the invention, a computer-readable medium is provided, comprising program instructions that, when executed by one or more processors, implement the methods discussed above. Attached Figure Description

[0054] To describe the methods by which the above and other advantages and features of this disclosure can be obtained, a more specific description of the principles briefly described above will be presented with reference to specific embodiments of this disclosure illustrated in the accompanying drawings. It should be understood that these drawings depict only exemplary embodiments of this disclosure and are therefore not intended to limit the scope of this disclosure. The principles herein will be described and explained with additional specificity and detail by referring to the accompanying drawings.

[0055] In particular, preferred embodiments of this disclosure will be explained in further detail below by way of example and with reference to the accompanying drawings, in which:

[0056] Figure 1A An exemplary schematic representation of a typical prior art method is depicted, which uses time ensemble averaging (typically for a single electrode) to interpret EEG waveforms to remove variability over a series of periods of a single individual.

[0057] Figure 1B An exemplary schematic representation of a typical prior art method for interpreting EEG waveforms using time-frequency analysis over a series of periods for a single electrode and an individual is depicted.

[0058] Figure 2A An exemplary hardware arrangement in a system for recording event-related responses to visual stimuli is described.

[0059] Figure 2B An exemplary schematic arrangement of an embodiment of the processing system described in this disclosure is shown.

[0060] Figure 3A schematic representation of a method for using probability distribution functions to interpret EEG waveforms to analyze the variability of a single individual across a range of periods at a single electrode / sensor / channel / brain location is depicted.

[0061] Figure 4A A topographic map depicting the change in differential entropy of representative subjects relative to the pre-stimulation baseline during the baseline-uncorrected period according to Example 1, with a post-stimulation latency of 0.39 s for multichannel EEG referenced to the common average electrode mean (common average data); and differential entropy h of channels (labeled relative to the extended 10-20 electrode placement system) for (i) PO5, (ii) PO6, and (iii) PO2. j (τ) and Kullback-Leibler divergence D KY [X j (τ>0)||X j The subgraph of [(τ≤0)].

[0062] Figure 4B A topographic map depicting the variation of ensemble mean ERP amplitude of representative subjects using baseline-uncorrected periods according to Example 1, with a co-mean data post-stimulation latency of 0.39 s; and subplots of ERP amplitude and negative entropy for channels (i) PO5, (ii) PO6 and (iii) POz.

[0063] Figure 4C A topographic map depicting the changes in differential entropy of representative subjects during the baseline-uncorrected period according to Example 1 is presented, with a post-stimulation latency of 0.39 s for multichannel scalp current density (SCD) EEG data (reference Laplace); and differential entropy h for channels (i) PO5, (ii) PO6, and (iii) PO2. j (τ) and Kullback-Leibler divergence D KY [X j (τ>0)||X j The subgraph of [(τ≤0)].

[0064] Figure 4D Topographic maps depicting post-stimulation changes in ensemble mean ERP amplitudes of representative subjects using baseline-uncorrected periods according to Example 1, with a post-stimulation latency of 0.39 s for Laplace reference data; and subplots of the negative entropy of channels (i)PO5, (ii)PO6, and (iii)POz.

[0065] Figure 5A Depicting Figure 4A Topographic maps of differential entropy for representative subjects, using zero-mean data, with a common mean post-stimulus latency of 0.39 s; and differential entropy h of channels (i) PO5, (ii) PO6, and (iii) POz.j (τ) and Kullback-Leibler divergence D KY [X j (τ>0)||X j The subgraph of [(τ≤0)].

[0066] Figure 5B Depicting Figure 4B Topographic maps of the ensemble mean ERP amplitude variation for representative subjects (zero-mean data were used in this study), with a post-stimulation latency of 0.39 s for the co-mean data; and subplots of ERP amplitude and negative entropy for channels (i)PO5, (ii)PO6, and (iii)POz.

[0067] Figure 5C Depicting Figure 4C Topographic maps of differential entropy changes in representative subjects (zero-mean data were used in this study), with a post-stimulus latency of 0.39 s for the Laplace reference data; and differential entropy h of channels (i) PO5, (ii) PO6, and (iii) POz. j (τ) and Kullback-Leibler divergence D KY [X j (τ)||X j The subgraph of (τ0)].

[0068] Figure 5D Depicting Figure 4D Topographic maps of post-stimulation changes in ensemble mean ERP amplitude for representative subjects (zero-mean data were used in this study), with a post-stimulation latency of 0.39 s for Laplace reference data; and subplots of ERP amplitude and negative entropy for channels (i)PO5, (ii)PO6, and (iii)POz.

[0069] Figure 6A Depicting Figures 4A to 4D and Figures 5A to 5B The bias-corrected differential entropy h of three selected channels of the same subject j (τ), which includes the channel (O1) that has the largest entropy change after stimulus presentation.

[0070] Figure 6B The distribution of minimum differential entropy across all channels is depicted.

[0071] Figures 7A to 7H The bias-corrected spatial mean differential entropy H(t) and directional variance (dva) of eight subjects for baseline uncorrected common mean data are plotted.

[0072] Figures 8A to 8H A plot of the bias-corrected spatial mean differential entropy H(t) and directional variance (dva) for the zero-mean common average data of eight subjects was presented.

[0073] Figure 9A The bias-corrected spatial mean (across all electrode / sensor / channel / brain locations) differential entropy and bootstrap-calculated confidence intervals for all participants in the experiment of Example 1 are depicted.

[0074] Figure 9B The directional variance (dva) of all participants in the experiment of Example 1 is depicted, along with the confidence intervals calculated by bootstrap.

[0075] Figure 10A Depicting Figures 4A to 4D and Figures 5A to 5D The correlation between the aggregate mean ERP amplitude and differential entropy across all stimulus latencies and electrode / sensor / channel / brain locations for the same subject.

[0076] Figure 10B Depicting Figures 4A to 4D and Figures 5A to 5D The spatial mean differential entropy and directional variance (dva) of the same subject.

[0077] Figure 11 An exemplary division of electrodes into sagittal and lateral groups, as used in Example 2, is depicted.

[0078] Figure 12 It depicts the results generated from the experiment described in Example 2, based on Figure 11 An exemplary vocabulary neighborhood ERP amplitude diagram for electrode group division.

[0079] Figure 13 An exemplary semantic feature ERP diagram is depicted from the experiment described in Example 2.

[0080] Figure 14 An exemplary lexical neighborhood differential entropy graph is depicted, generated from the experiment described in Example 2.

[0081] Figure 15 The semantic feature differential entropy graph is depicted from the experiment described in Example 2.

[0082] Figure 16 A graph of negative entropy in the lexical neighborhood generated from the experiment described in Example 2 is depicted.

[0083] Figure 17 A negative entropy graph of semantic features generated from the experiment described in Example 2 is depicted. Detailed Implementation

[0084] Various embodiments of this disclosure are discussed in detail below. While specific implementations are discussed, it should be understood that this is for illustrative purposes only. Those skilled in the art will recognize that other components and configurations can be used without departing from the scope of this disclosure.

[0085] In studies of brain activity, one or more stimuli (visual, auditory, or tactile / somatosensory) are presented and the brain's electroencephalogram (EEG) response to these stimuli is recorded.

[0086] Such electrical activity is called event-related brain activity, and it is assumed that this electrical activity is unrelated to ongoing neural activity in the background, and noise artifacts (such as those caused by blinking, eye movements, and power supply interference) can be resolved / extracted by using some type of signal processing (typically averaging responses to multiple stimulus presentations (trials)), as further explained below.

[0087] A standard experimental paradigm involves recording a subject’s response to the same stimulus multiple times at several different time intervals or periods. The final response is typically an average representation of the individual’s response over a predetermined time interval that coincides with the time the stimulus was presented. Figure 1A The diagram depicts a schematic representation of a standard experimental example.

[0088] Such recorded, stimulus-induced electrical activity is classified into two main types: evoked activity and induced activity.

[0089] The evoked activity is the temporal activity from time-locking and phase-locking to the onset of the stimulus; and it is calculated by averaging the responses of individual trials to repeated presentations of the stimulus, wherein the signal-to-noise ratio of the evoked response increases with the square root of the average number of trials. Therefore, with doubling the number of trials, the noise is reduced by approximately 30%; and to reduce the noise by half, it is necessary to conduct four times the number of trials.

[0090] The resulting voltage fluctuations are commonly referred to as event-related potentials (ERPs). The underlying theoretical assumptions supporting this process have been comprehensively described by Rugg and Coles (Event-Related Brain Potentials: Introduction, edited by M.D. Rugg and M.G. Coles, Oxford Psychology Series, Vol. 25, Psychophysiology: Event-Related Brain Potentials and Cognition (pp. 1-26), Oxford University Press, 1995):

[0091] 1. The signal under the response of interest is invariant in terms of time delay and shape, and the signal is composed of, for example... Figure 1A The event lock-in period of ERP is described in the text.

[0092] 2. The signal of interest is independent of the ongoing background activity and therefore is not expected to change as a result of the experiment.

[0093] In contrast, induced activity is time-locked rather than phase-locked to stimulus initiation. In other words, induced changes in EEG amplitude have a variable trial-to-trial initiation time course after stimulus presentation. Conversely, evoked changes in EEG amplitude will have a sustained trial-to-trial initiation time course after stimulus presentation. Induced activity is calculated as a function of time following stimulus presentation, based on the trial-averaged percentage change in short-time windowed spectral power across one or more frequency bands.

[0094] These changes in oscillation power over time depend on the relative magnitude of their changes and are often referred to as event-related desynchronization (if the band oscillation power increases relative to the pre-stimulation baseline) (ERD) or event-related synchronization (if the band oscillation power decreases relative to the pre-stimulation baseline) (ERS). Such changes are often grouped together and referred to as event-related spectral perturbations (ERSP; e.g., “Uncovering Event-Related Brain Dynamics” by Makeig et al., Trends in Cognitive Science, 1995; 8:204-210).

[0095] Both induced and evoked activities have been used to better understand how the brain processes stimuli and have served as the basis for attempting to monitor brain function in healthy individuals (e.g., consciousness / alertness), during medical interventions (e.g., the effects of anesthesia), and in individuals with illnesses (e.g., dementia, schizophrenia, depression).

[0096] Both of these standard processing methods can be used for longitudinal assessment of brain function, but they cannot detect individual patterns or characteristics of brain activity. Therefore, based on current technology, there are two options available for quantifying EEG activity in response to stimuli:

[0097] Option 1: Used to obtain the time ensemble average of induced activities

[0098] Based on the following function, perform some form of time ensemble averaging on the extracted data / measurement fragments that coincide with the stimulus onset (i.e., τ = 0) according to the additive noise pattern to obtain ERP or some other event-related response depending on the chosen modality:

[0099] s i (τ)=r i (τ)+e i (τ) (1)

[0100] =r(τ)+e

[0101] Where i indicates the presentation of the i-th stimulus, s ir(τ) is the recorded response to the i-th stimulus presentation as a function of the stimulus delay τ, r(τ) is the hypothetical trial-invariant stimulus response under a latent, phase-locked, τ-delay (i.e., ERP), and e is the variance of the zero-mean, independent, and identically distributed (iid) response hypothesized to be caused by irrelevant background, EEG, or other activity. Additive noise (by definition independent of test and stimulus delays).

[0102] Based on this, s i The expected value of (τ) E[s i [(τ)] is r(τ) (i.e., ERP), and the variance of the experimental mean is This makes the signal-to-noise ratio (SNR) of the ERP extracted from the time set so that it is That is, the SNR of the extracted ERP increases with the square root of the average number of trials.

[0103] Alternatively, in addition to or as an alternative to Option 2 described below, the analysis described above may be performed in a typical EEG stimulation treatment paradigm.

[0104] Option 2: Used for time-frequency analysis to obtain induced activity

[0105] Using one of many available methods (e.g., windowed short-time FFT, continuous or discrete wavelet transform), a time-frequency analysis is performed on each extracted segment before ensemble averaging to obtain ERS / ERD / ERSP activity for a given band (although other baselines such as z can be used, typically normalized to the pre-stimulation baseline by a percentage change in the event-related spectrum (standard score)). Typical EEG bands for such induced activity can be calculated including δ (typically 0 Hz to 4 Hz), θ (typically 4 Hz to 8 Hz), α (typically 8 Hz to 13 Hz), β (typically 13 Hz to 30 Hz), and γ (typically >30 Hz) or any other (typically narrow) classically defined human EEG bands.

[0106] It should be understood that, by design, the standard approach to ERP is to destroy variability information (often referred to as "neurophysiological noise") by obtaining relevant averages from successive trials. Figure 1B The diagram schematically depicts an exemplary stage of the method, as well as an exemplary ERD (in this case, an α-band ERSP) obtained according to the method. Figure 2ATypical components of System 10 are described, which can employ the methods and approaches disclosed herein, using an EEG measurement modality as a reference. Those skilled in the art will understand that similar arrangements (modified as required relative to each sensing modality) can be used for other sensing modalities in the study. Many steps of our method will be familiar to those skilled in the art of recording brain activity and are not discussed further herein.

[0107] 1. Stimulus selection

[0108] Without departing from the scope of this disclosure, any potential visual, auditory, tactile / somatosensory, or gustatory / olfactory stimulus or category of stimulus may be selected and presented to the subject (e.g., pictures of specific nouns, spoken words, visual checkerboard stimuli, brief exposure to odors, etc.).

[0109] From a practical perspective, based on previous studies involving ERP or ERD / ERS / ERSP analysis or based on neuropsychological requirements, stimuli or stimulus sets are selected with the expectation that these stimuli or stimulus sets will probe specific aspects of cognitive functions (e.g., memory, attention, semantic cognitive processing, or similar functions).

[0110] One example could be a visual memory task, which involves presenting a series of visual images at semi-regular short intervals, with one or more repetitive stimuli present, in order to study or probe short-term memory.

[0111] As illustrated in Example 1 below, an exemplary experiment could ask subjects to participate in a silent reading task in which a single word is briefly presented, followed by a black (“blank”) screen; and subjects are asked to silently read the words themselves. Responses can then be recorded under high or low memory load (by asking subjects to memorize new sequences of six random integers presented every 8 to 12 words (high load) or six identical integers (low load)) to investigate the electrophysiological relevance of word responses to systematic changes in background cognitive activity.

[0112] Another exemplary experimental setup is described in Example 2 below. It should be understood that the details of the experimental setup are not important to this disclosure, which is concerned with interpreting the results obtained.

[0113] Typically, as with the collection of ERP (“evoked”) and ERD / ERS / ERSP (“induced”) EEG responses, stimuli are presented sequentially at intervals deemed sufficiently long to capture the temporal course of a single trial “evoked” / “induced” response. Similar considerations will apply to other monitoring modalities.

[0114] The duration of stimulation is generally kept as short as possible, but must meet the following conditions: 1) ensure that it will elicit an appropriate response (which would be an EEG relative to Example 1) and 2) whether to reduce the “conscious” stimulation treatment.

[0115] For visual stimuli, they are typically displayed between 50 ms and 500 ms, and presented every approximately 1.5 s to 2.5 s. However, for those skilled in the art who record brain activity in response to stimulus presentation for EEG monitoring, there are many variations in the presentation sequence, stimulus contingencies, and timing familiar to them. After stimulus presentation, depending on the task requirements, participants may be asked to make some form of auditory or motor response, such as pressing a specific button.

[0116] In many cases, it will be necessary to normalize stimulus presentation to one or more perceptual thresholds in order to account for unavoidable changes in visual, auditory, or somatosensory functions during health conditions (e.g., age-related hearing loss / presbyopia) or illnesses.

[0117] 2. Recording of the reaction

[0118] Electrical activity in the brain, electrical (EEG, MEG, ECoG) or other (BOLD fMRI, NIRS) data can be recorded before and during stimulus presentation to measure the brain's response to stimuli. Because the brain exhibits considerable topographic / spatial variability in its responses, recording will need to occur at multiple and widely distributed brain locations.

[0119] In the case of recording the brain's electrical activity, this is typically achieved by recording EEG using a standardized system (appropriately referenced and based on) according to the placement of scalp electrodes (wet or dry, active or passive) (e.g., a 10-20 system accommodating 21 scalp locations, an extended 10-20 system (also known as the "10%" or 10-10 system) that can accommodate up to 74 electrodes, or a 10-5 or "5%" system that specifies up to 345 electrode locations).

[0120] For EEG, the fidelity and quality of the recorded signal are typically determined by the digitization depth (12 to 24 bits), sampling frequency (approximately 80 Hz to 5000 Hz), and the surrounding electrical environment (which current existing technologies mitigate by utilizing active electrode configurations).

[0121] For subsequent analysis, the timing of the stimulus start must be recorded simultaneously with the recorded signal using an appropriate single / multi-bit / word trigger.

[0122] This is typically achieved by the stimulus presentation computer sending a trigger (by means of being properly configured as a serial or parallel port) to the computer running the brain activity measurement acquisition system at the start of the stimulus.

[0123] 3. Pretreatment of recorded reactions

[0124] After single / multichannel recordings of responses to a large number of sequentially presented stimuli (in cases where EEG and MEG are typically >30 to 40), and the responses to each presentation are typically referred to as trials, the continuous time series for each spatial location (e.g., each EEG recording electrode) needs to be segmented with respect to the start time of each stimulus.

[0125] This process is often referred to as "time-slicing" the data.

[0126] Before the “blinking” occurs, various filtering techniques can be applied to the data / measurements to remove incidental noise (in the case of the most significant 50 / 60Hz master artifact in EEG; or in the case of BOLD fMRI cardiac impaction and subject movement) as well as intrinsic physiological artifacts (most notably blink-induced electrical activity).

[0127] Following such filtering, "timing" involves extracting data segments that coincide with the stimulus start time. These data / measurement segments extend from a time before the stimulus start (e.g., 100ms to 500ms for EEG / MEG) to a time after the stimulus start (e.g., approximately 1000ms to 2000ms for EEG / MEG) and are temporally consistent, such that the stimulus start time (τ) is defined as τ = 0. That is, for EEG / MEG, data / measurement segments (or periods) are typically defined as extending over intervals τ = [-500 to -100, approximately 1000 to 2000] ms. It should be understood that other predetermined periods or time intervals may be utilized.

[0128] The time after stimulus presentation is defined as positive delay (relative to stimulus presentation), i.e., τ>0, while the time before stimulus presentation is defined as negative delay (relative to stimulus presentation), i.e., τ<0.

[0129] After such data / measurement segments, individual periods can be examined or otherwise processed to ensure that no endogenous or exogenous artifacts are present.

[0130] As referenced above Figure 1A (Adapted from Luck et al., Trends in Cognitive Science, 2000; 4(1):432) and Figure 1B (Adapted from Park et al., Journal of Motor Behaviour, 2018; 50(4):457)) The current prior art would be to calculate some form of average response (representing brain activity generated by a stimulus), thus eliminating all trial-to-trial variability. However, in our disclosure and as referenced below, Figure 3The methods discussed are significantly different from those used to calculate induced and induced activities.

[0131] refer to Figure 2B A schematic representation of an exemplary embodiment of the analysis system 15 is depicted.

[0132] One or more processors 17 are configured to perform the steps described below to analyze brain activity measurements according to instructions stored in memory 19.

[0133] Specifically, the acquisition module 21 is configured to acquire brain activity measurements of the subject at electrode / sensor / channel / brain location over multiple predetermined time periods, wherein the predetermined time periods are selected to include the duration of brain activity before and after presenting the subject with multiple repetitive external stimuli.

[0134] The assessment module 23 is configured to assess the variability of the brain activity measurements obtained over multiple predetermined time periods.

[0135] The determination module 25 is configured to determine changes in the brain's response state based on variability over multiple predetermined time periods.

[0136] like Figure 3 The description in the document records the responses of subjects to the same stimulus presented multiple times at different time intervals or periods. See also... Figure 3 This is where it differs significantly from traditional methods used to compute evoked and induced activities, in which trial-to-trial variability is preserved, enabling the development of important methods for characterizing brain function in individual individuals and corresponding groups of subjects.

[0137] Advantageously, this can be achieved by empirically constructing / estimating the EEG amplitude of the j-th electrode / sensor (or at a specific brain location for any other brain-related signal) X j The probability density function (probability distribution) of (τ). This is accomplished as a function of the stimulus delay τ. Here, X... j (τ) is a random variable corresponding to the EEG amplitude (or any other brain-related signal discussed earlier) at a fixed time delay τ relative to the stimulus presentation (τ=0).

[0138] Several methods can be used to empirically determine the probability density function, including histogram estimation by dividing into bars and kernel density estimation.

[0139] Based on this, it depends on the properties of the possibly known distribution. The ERP specified for EEG or MEG at the y-th electrode / sensor can be specified as r. j (τ)=E[Xj (τ)] or alternatively specified as r j (τ)=median[X j (τ)] or r j (τ)=mode[X j (τ)。

[0140] It is worth noting that in this method, it is not necessary to assume the existence of independent and identically distributed additive noise processes, because there is no variance Var[X] of the recorded signal. j [(τ)] is a restriction independent of the stimulus presentation delay τ.

[0141] As this disclosure clarifies, the removal of this restriction allows experimenters to define stimulus-induced EEG, responses (or any other measure of brain activity based on the measurement modality used) according to various information theory measures, including but not limited to: differential entropy, transfer entropy, relative entropy / Kulback-Leibler divergence, negative entropy, and multiscale entropy.

[0142] Therefore, this disclosure assumes that, when calculated over repeated stimulus presentations, any other measure of variance or dispersion must actually change rather than remain constant after stimulus presentation. This expectation that variance will not change is the basis for common methods in the prior art of time-ensemble averaging of readings from successive trials.

[0143] From a biological perspective, when an organism responds to a given stimulus in its environment, it should be associated with a decrease in the list of possible associated behavioral responses. If the list of possible associated behavioral responses does not decrease, the stimulus can be considered meaningless to the organism.

[0144] When understood from an electrophysiological perspective, within the context of higher human cognition, this disclosure clarifies that meaningful presentation of a time-locked stimulus should be associated with a reduction in the uncertainty of the subsequently evoked neurophysiological response. Therefore, for a known meaningful stimulus, the absence of any reduction in the subsequent variability of the electrophysiological response could only mean that, in that particular case, the stimulus has no received meaning, and / or the measured evoked activity has no causal (informational) correlation with any subsequent cognitive response.

[0145] This contrasts with standard additive noise models in the prior art, which assume that stimulus-induced activity cannot be associated with any reduction in uncertainty, since the trial variance remains constant relative to time delay after stimulus presentation. That is, the actual induced response is invariant in the trial. This is behaviorally meaningless, and this disclosure has clearly stated that this assumption is biologically untenable.

[0146] refer to Figure 3Furthermore, understanding the inappropriateness of the underlying assumptions of the prior art allows for the understanding of other aspects of this disclosure.

[0147] After segmenting continuous EEG recordings, various methods can be used to construct, for a given electrode / sensor / channel / brain location j, a probability density function estimating the stimulus initiation time of brain activity for a given time delay τ. For example, in an exemplary non-limiting example, such methods include 1) empirical distribution function estimation, 2) scale histogram estimation, and 3) kernel density estimation; all methods are familiar to and well described by those skilled in the art with empirical estimates of probability density functions.

[0148] For any system that adapts to its environment in some way (such as the human brain), variance (but more specifically differential entropy (a measure of the fit of a continuous random variable to a mean singular value) h) j (τ)

[0149]

[0150] The value must be instantaneously reduced from its pre-stimulus value, such as within a short period of time (sufficiently short enough for the "meaning" of the stimulus to remain unchanged as expected) based on the presentation of multiple stimuli.

[0151] In fact, as specified in Example 1, the analysis of time-locked EEG activity in response to a passive reading task (where the timing of stimulus presentation is defined as the start of a single word presented in succession at 500 ms intervals separated by 1200 ms intervals) explicitly indicates that h j The numerical estimate of (τ) decreases instantaneously after the presentation of heterogeneous stimuli on the terrain. Furthermore, as specified in Example 2, such a change can be used to distinguish between the lexical features (spelling features) and semantic features of word stimuli presented in such a visual presentation.

[0152] Example 1 - Detecting changes in stimulus activation:

[0153] Example 1 demonstrates the information measure h in the exemplary experiment. j The probability density function defined in the calculation of (τ)(differential entropy) is as follows. The use of.

[0154] A. Participants and Tasks

[0155]

[0156]

[0157] The average number of trials under high and low memory conditions was 722.77 (SD 44.17), or approximately 361 per 50:50 presentation chance condition.

[0158] B. EEG Recording and Preprocessing

[0159]

[0160] Tests with significant artifact residues were rejected after independent component analysis based on visual inspection. Common mean (CA) reference data / measurements and estimated scalp current density (SCD) were analyzed using zero-mean periods and baseline-uncorrected periods. No other forms of baseline correction were performed. The SCD was estimated using a spherical spline method, as implemented by ft_scalpcurrentdensity in FieldTrip.

[0161] C. Calculation of information theory measures including differential entropy

[0162] For the j-th electrode / position, an amplitude distribution is formed on each trial with a fixed stimulus delay τ to empirically estimate the probability density. In order to calculate the differential entropy. For this continuous probability density, the differential entropy h j (τ) is defined as

[0163]

[0164] However, because the number of periods is finite, it will be necessary to estimate based on the amplitude data / measurements divided into bars. Therefore, assuming the random amplitude X at time τ j (τ) is divided into bars of width Δ, and we will use the differential entropy h j (τ) is calculated as

[0165]

[0166] in, It is a discrete probability distribution The Shannon entropy. However, it is well known that this simple interpolation estimate of entropy is a biased estimate because it underestimates the true entropy. Several ways exist to correct for this bias, including bootstrap bias correction. Because this bootstrap correction also allows for the calculation of a 95% confidence interval, we choose this method for our bias correction. Our bootstrap bias-corrected differential entropy. Calculated as

[0167]

[0168] Where B is the number of bootstrap samples, and This is the differential entropy of the b-th bootstrap sample calculated using the interpolation estimation of equation (5) above. By choosing B = 1000, we can obtain the differential entropy of the b-th bootstrap sample through... Sort the data and select the lower 2.5% and higher 97.5% percentile boundaries to obtain our 95% confidence interval. Plot the results. Topographic mapping relative to the pre-stimulus baseline. Specifically, the baseline change of differential entropy for -0.3 < τ < 1.4 s is plotted.

[0169]

[0170] in Differential entropy is calculated based on the pre-stimulus amplitude distribution formed during the time interval from -0.3 s to 0 s. Calculating such a baseline change, compared to the discrete (Shannon entropy) counterpart, also mitigates some undesirable properties of differential entropy (boundedness and negativity). Typically, such a change will depend on the nature or classification of the stimulus, where such bias is calculated based on the nature or classification of the stimulus. (τ|stimulus). Topographic maps of the ensemble mean ERP relative to the pre-stimulus baseline were also plotted. To illustrate the specific characteristics of the estimated differential entropy, we compared it to a measure of variability already in use, known as directional variance (dva), which was employed by Schurger et al. (cortical activity is more stable when sensory stimuli are consciously perceived (PNAS 2015; 112(16):E2083-E2092)). This measure of variability was calculated using all recording electrodes / sensors / channels / brain locations. directional (circular or angular) variance (dva) was calculated solely from data / measurements obtained from scalp current density, allowing for better demonstration of the orthogonality of the channel data / measurements. This directional variance was defined as...

[0171] dva(τ)=1-R(τ) (7)

[0172]

[0173] Among them, v i (τ)=[x1(τ),…,x M (τ)] i R(τ) is the vector of M channels of the lock-time EEG amplitude under the stimulus delay τ at the i-th period. R(τ) is often referred to as directional coherence. Specifically, dva is estimated by averaging across all electrodes. Compare them.

[0174] In addition, the theoretical information content of a single channel and the Kullback-Leibler divergence (relative entropy) D were calculated. KL [X j(τ)>0||X j [(τ≤0)] and negative entropy.

[0175] Negative entropy (a non-Gaussian measure) is defined as

[0176]

[0177] in, It is the amplitude probability distribution defined above. It has the same The differential entropy of a Gaussian distribution with the same variance, and It is already defined The deviation correction differential entropy. Calculate the negative entropy to determine h. j To what extent is the change in (τ) caused by variance variation or non-Gaussianity?

[0178] D. Data Analysis Results

[0179] Figures 4A to 10B Initial results for evaluating differential entropy and other experimental measures are shown. Figures 4A to 4D and Figures 5A to 5D The representative subjects are shown according to Equation 6 Topographic map of the baseline variation of differential entropy. Figures 4A to 4D We used the baseline uncorrected period as the basis for all measures. We have shown a single frame of an animation plot of a 0.39s delay after stimulation. In contrast, Figures 5A to 5B Using the mean-reduced period, it can be seen that the differential entropy in channels PO5, PO6, and POz of the two widely used electrode derivatives (common mean and Laplace reference) decreases rapidly from baseline after stimulus presentation, reaching a minimum at a delay of approximately 0.39 s. This minimum corresponds topographically to the maximum decrease in differential entropy in the parietal electrodes, a result clearly consistent with findings from neurological studies investigating the topological changes in brain activity during reading.

[0180] Figure 4A and Figure 4C Colored topographic maps were constructed, illustrating the spatial pattern of the change in differential entropy (in nats) from the pre-stimulus baseline (referred to here as “variational entropy” / ΔH) at 390 ms after stimulus presentation (as indicated by the vertical red line in the attached subplots (i), (ii), and (iii) from left to right). These attached subplots show the differential entropy, Kullback-Leibler divergence (ΔH) of the three selected channels / electrodes (PO6, PO5, and POz), and the time of minimum post-stimulus differential entropy (ΔH) relative to the pre-stimulus baseline. KL [X j (τ)>0||X j(τ≤0)]). Figures 4A to 4D The data / measurements in this document are uncorrected baseline period data / measurements.

[0181] Figure 4B and Figure 4D Colored topographic maps were depicted, showing the event-related potentials averaged for the corresponding standard time set and the corresponding subplots, which showed the time processes of negative entropy (as defined in Equation 8). The event-related potential amplitudes of the three selected channels / electrodes (PO6, PO5, and POz).

[0182] Similarly, Figures 5A to 5D The accompanying subplots present estimates of topographic data / measurements and differential entropy (labeled H here), as well as other theoretical quantities of information as a function of stimulus delay; this is the mean-free period.

[0183] review Figures 4A to 4D and Figures 5A to 5D And the accompanying subplots, important features to note are 1) the instantaneous decrease in differential entropy after stimulus presentation and 2) the spatial heterogeneity of the magnitude of differential entropy changes. These results are based on nearly 300 stimulus presentations.

[0184] Figure 6A and Figure 6B It shows Figures 4A to 4D and Figures 5A to 5D The bias-corrected differential entropy plots of three selected channels for the same subjects, including the channel with the largest entropy change after stimulus presentation (O1).

[0185] Furthermore, the post-stimulus change in differential entropy is more clearly viewed as being caused by… Figure 6A The solid lines in (i) channel OZ, (ii) channel O1, and (iii) channel O2 represent the mean bias-corrected differential entropy. Bootstrap 95% confidence intervals (shaded) for the differential entropy are also plotted. These changes in post-stimulation differential entropy (and 95% confidence intervals) for a single typical subject's silent / passive reading task are shown for electrode channels Oz, O1, and O2.

[0186] Important points to note in these figures are 1) the instantaneous reduction of differential entropy after stimulation and 2) the magnitude of the temporal change of differential entropy across electrodes, in which the timing of the minimum channel differential entropy varies considerably. Figure 6B The distribution of time delays across all channels is shown, where the differential entropy is at its minimum.

[0187] Figures 7A to 7H and Figures 8A to 8HThe average differential entropy across all electrodes (hereafter referred to as the “spatial integral” differential entropy) and the dva of all subjects are shown for both zero-mean and baseline-uncorrected data. For both baselines, the “spatial integral” differential entropy decreased significantly within the first 0.2 to 0.3 seconds after stimulation. For zero-mean data, this response became more uniform. The temporal course of this response showed great similarity for both CA-referenced and SCD-calculated data. Figure 9A and Figure 9B The average “spatial integral” differential entropy and dva of all participants are shown, along with their bootstrap computation confidence intervals.

[0188] Figure 9A The average integral differential entropy of all participants and the confidence intervals calculated by bootstrapping are depicted (shaded).

[0189] Figure 9B The dva values ​​of all participants and the confidence intervals calculated by bootstrapping are depicted (shaded).

[0190] Figure 10A The correlation between the differential entropy, corrected for bias across all channels and time intervals (-0.3, 1.4) s for subjects in the previous plot, and the amplitude of the ensemble mean ERP is shown. It should be noted that no significant correlation was observed, and this is a typical result when examining other subjects. In other words, changes in the estimated differential entropy are uncorrelated and therefore unrelated to changes in the ensemble mean ERP amplitude.

[0191] Figure 10B The correlation between the (spatial average) differential entropy of the integral over the time interval (-0.3, 1.4) s and dva is shown. A weak positive correlation can be observed between the integral h^ and dva.

[0192] E. Summary of Results and Conclusions

[0193] In summary, Example 1 has provided an explicit explanation, as disclosed and defined above, of the probability density function. The response changes to stimuli presented by information theory measures including differential entropy, “spatial average” differential entropy, Kullback-Leibler divergence, and negative entropy. Furthermore, such quantifiable changes are considered independent of the changes obtained from the corresponding time sets of the ERP amplitude.

[0194] Example 2 - Distinguishing changes in stimulus activation:

[0195] In a further example, it is demonstrated that according to Computational information measures can effectively quantify trial-by-trial variability and provide specific interpretations of evoked responses. Importantly, they can also distinguish electrophysiological responses to different categories of stimuli (lexically variable versus semantically variable visual word stimuli).

[0196] A. Participants and Tasks

[0197]

[0198] B. EEG Recording and Preprocessing

[0199]

[0200] C. Calculation of differential entropy and negative entropy in information theory.

[0201] In the experiment, an amplitude distribution is formed at each time point corresponding to the stimulation delay τ and the j-th electrode / sensor / channel / brain location, and this amplitude distribution is used to calculate the differential entropy, as in Example 1. For amplitude X... j Continuous probability density function of (τ) Differential entropy h j (τ) is defined as

[0202]

[0203] However, because the number of periods is finite, estimation is based on amplitude data / measurements divided into bars. Therefore, a histogram of the amplitudes for all experiments with a fixed time delay was constructed. Figure 3 We choose the width of the histogram bars such that a fixed number of bars will cover the range of a single trial amplitude for a given channel. In this way, the trial-by-trial variability of each word condition is quantified. Now, suppose the amplitude X at time τ... j (τ) is divided into bars of width Δ, and the differential entropy h can be expressed as... j (τ) is calculated as

[0204]

[0205] in, It is a discrete probability distribution The Shannon entropy. It is well known that this simple estimate of differential entropy is biased because it underestimates the true entropy. However, a bootstrap bias correction can be used (as in Example 1 – Equation 5), which also allows for the calculation of a 95% confidence interval and a bias-corrected negative entropy.

[0206] Also calculate the negative entropy to determine h j To what extent is the change in (τ) caused by variance variation or non-Gaussianity? As in Example 1, the negative entropy is calculated as...

[0207]

[0208] in, It is the amplitude probability distribution defined above. It has the same The differential entropy of a Gaussian distribution with the same variance, and It is already defined The deviation correction differential entropy.

[0209] Following the standard ensemble averaging method described previously, the mean event-related potential (ERP) for each electrode under each condition was calculated separately for all trials. The ERP was then normalized to a Z-score to achieve treatment consistent with other calculated measures (notably, the unnormalized ERP amplitude was also calculated, which yielded substantially the same results in all subsequent statistical analyses outlined below).

[0210] D. Statistical Analysis: Time Window and Electrode Cluster

[0211] Following ERP calculations, ERP component literature was used to guide selection, and the data was visually examined to run statistical analyses within the time windows of interest. ERP chart ( Figure 11 and Figure 12 The study revealed a clear early lexical component with a negative peak at 100 ms (N100) and a positive peak at approximately 210 ms. Time windows were set to capture the full breadth of N100 relative to early lexical access in the review literature. The examination also revealed a wide negative trough between approximately 250 ms and 650 ms, where a negative drop in interest occurred at 450 ms. ERP literature indicated that a range of word-related semantic and memory processes were associated with the negative peak at 400 ms, therefore the analysis window was set between 400 ms and 500 ms. Finally, a late positive component (LPC) window of 600 ms to 800 ms was used to capture any post-processing of word stimuli. Therefore, three time windows were defined on this basis (statistical comparisons will be made across these three windows): i) τ = [100, 200] ms, ii) τ = [400, 500] ms, and iii) τ = [600, 800] ms.

[0212] To simplify statistical analysis, a subset of electrodes was defined. Based on established specifications for cognitive ERP analysis, the 64-channel electrode array was divided into electrode clusters to investigate regions where high and low samples differed significantly in ERP amplitude, differential entropy, and negative entropy for each operation. Three clusters were formed along the anterior, central, and posterior parts of the sagittal axis. Three additional clusters were formed on the left, middle, and right sides of the transverse axis. Figure 11 ).

[0213] Based on time windows and electrode clusters, analyses were run using the R statistical package employing the afex library, as defined above. ANOVA was performed with repeated measures and type III sums of squares, and Greenhouse-Geisser correction was used in cases of violation of the sphericity assumption. Data / measurement analyses were performed for each time window of interest using 3 sagittal (anterior / central / posterior) × 3 lateral (left / middle / right) × 2 word types (high / low). For significant principal effects, comparisons were run to determine which particular cluster showed significant word differences.

[0214] E. Data Analysis Results

[0215] Two participants had to be completely removed from the analysis because more than 50% of their data were unusable. After preprocessing, an average of 79% of each participant's trials were included (see Table 1), and the average number of trials per condition was also included.

[0216] Table 1: Number of trials for each word condition in Example 2.

[0217] High vocabulary 89.00(6.33) low vocabulary 89.42(7.15) High semantic 88.58(6.62) low semantic 90.08(5.31)

[0218] Event-related potential (ERP) amplitude

[0219] For both lexical neighborhood and semantic complexity feature operations, clear ERP components of N100, P200, N400, and late positive component (LPC) are evident in the front electrode. Figure 12 and Figure 13 In the back electrode, these components are still evident; however, due to the use of a common average reference, the voltages of these components are inverted because the voltages must sum to zero, so that any negative is balanced by an equivalent positive.

[0220]

[0221]

[0222]

[0223] Differential entropy

[0224] After presenting high or low categories in lexical word operations and semantic word operations, the trial-by-trial variability calculated as differential entropy using a fixed bar width is reduced across all electrode clusters. Figure 14 and Figure 15 In all experiments, the overall trend showed that the differential entropy decreased rapidly within the window of τ = [100, 200] ms, reached its lowest point between 400 ms and 500 ms, and then increased.

[0225]

[0226]

[0227] Negative entropy

[0228] Figure 15 and Figure 16 The negative entropy of word operations as a computation of non-Gaussianity is plotted in the figures. These two figures show that the negative entropy decreases with time delay from the start of word stimulus presentation. It reaches its minimum within a time window of approximately 400 to 500 ms (matching the time window of the differential entropy).

[0229]

[0230]

[0231] No significant differences or interactions were found between words operating on semantic complexity features within any predefined time window (100ms to 200ms, 400ms to 500ms, 600ms to 800ms).

[0232] F. Summary of Results and Conclusions

[0233] The only significant difference in ERP lexical neighborhood word type occurred within the 400ms to 500ms window of the right electrode. Words with high lexical neighborhoods caused a larger N400 bias than words with low lexical neighborhoods. Those familiar with cognitive ERP processing and interpretation will understand that such an N400 effect is generally interpreted as being caused by high lexical neighborhood word type conditions that facilitate semantic processing.

[0234] Similarly, the only significant difference in the impact of word type on ERP semantic complexity was found in the 600ms to 800ms window. Based on the conclusion that positive components occurring within 400ms to 800ms are associated with the memory of specific information, this can be linked to higher semantic features and more detailed meanings. It can be concluded that words with more semantic features are expected to show larger amplitudes in the positive components, as can be seen here. We have not considered the possibility that later positive components might be affected by semantic richness in the visual word recognition task.

[0235] In contrast, word stimuli had a much more significant impact on the change in differential entropy. All subjects showed a significant reduction in differential entropy when assessed across all word types, and this reduction occurred immediately after stimulus presentation. Significant reductions in both lexical and semantic operations were observed between 100 and 150 ms (particularly in the post-electrode phase), reaching their lowest point between 400 and 500 ms.

[0236] Significant differences in word type within lexical neighborhoods are reflected in the differences in calculated differential entropy, supporting the argument that changes in differential entropy correspond to meaningful changes in brain activity and reflect important aspects of neural information processing. Lower differential entropy values ​​were found after the presentation of words with low lexical neighborhoods in time windows of 400 ms–500 ms and 600 ms–800 ms, suggesting that, as expected, spelling unique words is more meaningful. The only significant interaction between word type and electrode location revealed a significant difference in word type between the anterior and posterior electrodes in the 600 ms–800 ms window.

[0237] Similarly, and as expected, significant differences were observed in the changes in differential entropy after word presentation between words with high and low semantic complexity. Specifically, in time windows of 100ms–200ms and 400ms–500ms, words with a large number of semantic features produced lower differential entropy values ​​than words with low semantic features. This supports the central idea that stimuli with greater meaning (in this case, a single word) lead to a significant reduction in the uncertainty of recorded electrophysiological activity; this aligns with the understanding of the richness effect in contemporary literature, where a greater number of semantic features enriches the meaning of the target word, thus translating into better semantic processing.

[0238] It should be noted that when all the basic perceptual components of a word remain constant, there is a significant difference in differential entropy between words with high and low semantic richness. The expected significant semantic complexity differences between word types appear within a time window of 400ms to 500ms. This 400ms to 500ms window clearly demonstrates (…). Figure 14 Words with high semantic richness were associated with greater reductions in differential entropy. The topography of differential entropy changes identified after word presentation was largely consistent with a series of other neuroimaging studies, in which significant semantic complexity word type effects were found in both posterior and anterior electrodes.

[0239] Finally, as Figure 15 and Figure 16As shown, a global reduction in negative entropy can be observed in both lexical neighborhood and semantically complex word-type stimuli, suggesting that the reduction in "free" energy indicates an increase in dynamic stability, consistent with the stability of recurrent neural networks entering decision states. This reduction in negative entropy supports the view that the stimulus-induced computational differential entropy h... j The change in (τ) is likely caused by non-Gaussianity rather than by alternative mechanisms that affect the variability of individual trials (such as reduced EEG power or increased phase coherence of individual trials).

[0240] In the examples described above, using information measures such as differential entropy to quantify the trial-by-trial neural variability of stimulus-driven evoked potentials provides a broader representation of neural responses. Based on the results of Examples 1 and 2, such increases / decreases in differential entropy can be reasonably interpreted as corresponding to increases / decreases in the cognitive “meaning” of the stimulus. This was explicitly observed in the early semantic processing of word-type stimuli, where words with high semantic complexity showed a greater decrease in differential entropy compared to words with low semantic complexity. In the context of typical cognitive paradigms (such as Oddball sequences), differential entropy can be usefully used to characterize the trial-by-trial variability of a range of neurological problems and mental illnesses, including ADHD, dyslexia, and psychosis.

[0241] The spatiotemporal pattern of differential entropy change relative to a given stimulus or stimulus category, and from Any other theoretical quantity of information obtained is predicted to be individual-specific and therefore expected to represent some form of "cognitive fingerprint." Based on this, deviations from such a single "normative" pattern assessed longitudinally will be used for the objective diagnosis of pathological functional neurological problems and mental states.

[0242] according to Other information theory measures that can be computed can also be used for such diagnostic purposes. These measures will include, but are not limited to, pairwise symmetric mutual information I / [X]. j (τ); X j’ [τ'], negative entropy, asymmetric transitive entropy, conditional entropy, relative entropy / Kulback-Leibler divergence, joint entropy, and multiscale entropy. As is well known, such information-theoretical measures can specify... An important aspect, particularly its systemic interdependence in time and space. Such a measure will be central to systematically identifying task and individual differences, and these results are crucial from a diagnostic perspective.

[0243] Understandably, when considering brain activity monitoring, It can be configured to estimate continuously over time based on multiple sessions of a specified subject. Each session can include measurements of brain activity against multiple repetitive stimuli. Sessions can be spaced at intervals selected from multiple hours, days, months, or years. Similarly, h can be estimated longitudinally based on multiple sessions of a specified subject. j (τ), where each session includes brain activity measurements for multiple repetitive stimuli, and these sessions are separated by intervals chosen from multiple hours, multiple days, multiple months, or multiple years.

[0244] Empirical estimates of these quantities can be achieved using various existing methods (e.g., integral kernel density estimation, Gaussian dependence, etc.).

[0245] Advantageously, this disclosure can define an “information fingerprint” for each stimulus / stimulus category, at a given moment, and for each individual at a given spatiotemporal scale (depending on the brain-related signaling modalities used to observe brain activity).

[0246] In the context of learning and ongoing activity, any changes in this "information fingerprint" will provide important information for the ongoing assessment of neurological and mental function in both healthy and ill individuals. Therefore, the methods and systems disclosed herein can provide a meaningful objective instrument for longitudinal assessment of brain information processing. Furthermore, it can be applied to the assessment of brain function in individual subjects and thus can be used for a wide range of neurodiagnostic monitoring purposes, including monitoring disease progression, monitoring sedation levels (e.g., under anesthesia to prevent intraoperative recall or awakening), monitoring alertness / arousal / attention, and providing methods for establishing brain-computer interfaces (BCIs).

[0247] It should be understood that, by comparison, the measures provided by current prior art (primarily relying on ensemble-averaged event-related potentials and event-related desynchronization / synchronization) are too small to provide meaningful, objective longitudinal assessments of brain information processing. This disclosure is not limited to the calculation of first and second moments of band-limited brain activity, as opposed to prior art extraction of stimulus-evoked (event-related potentials) and induced (event-related desynchronization / synchronization) EEG activity.

[0248] It should be understood that the above embodiments are described by way of example only. Many variations are possible without departing from the scope of the invention as defined in the appended claims. For clarity, in some instances, the technology may be presented as comprising various functional blocks, which include steps or routines in a method implemented in software or a combination of hardware and software.

[0249] The methods described in the examples above can be implemented using computer-executable instructions stored in or otherwise made available from a computer-readable medium. These instructions may include, for example, instructions and data that cause or otherwise configure a general-purpose computer, special-purpose computer, or special-purpose processing device to perform a specific function or group of functions. Some of the computer resources used may be accessible via a network. The computer-executable instructions may be, for example, binary files, intermediate format instructions such as assembly language, firmware, or source code. Examples of computer-readable media that may be used to store instructions, information used, and / or information created during the process of the methods according to the described examples include disks or optical discs, flash memory, universal serial bus (USB) devices with non-volatile memory, networked storage devices, etc.

[0250] Devices implementing the methods disclosed herein may include hardware, firmware, and / or software, and may take any of a variety of form factors. Typical examples of these form factors include laptop computers, smartphones, minicomputers, personal digital assistants, etc. The functionality described herein may also be implemented in peripheral devices or add-on cards. As a further example, such functionality may also be implemented on a circuit board of different chips, or on different processes that may be executed in a single device.

[0251] Instructions, media for transmitting such instructions, computing resources for executing such instructions, and other structures for supporting such computing resources are means for providing the functionality described in these disclosures.

[0252] Although various examples and other information are used to interpret aspects within the scope of the appended claims, no limitation on the claims should be implied based on specific features or arrangements in such examples, as those skilled in the art will be able to derive a wide variety of implementations from these examples. Furthermore, and although a subject matter may have been described in language specific to structural features and / or method steps, it should be understood that the subject matter defined in the appended claims is not necessarily limited to these described features or actions. For example, such functionality may be distributed differently or performed in components other than those identified herein. Rather, the described features and steps are disclosed as examples of components of systems and methods within the scope of the appended claims.

Claims

1. A method for constructing a representation of the brain response state changes of a mammalian subject to multiple repetitive external stimuli, comprising: Acquire multiple brain activity measurements of the subject over multiple predetermined time periods, wherein each of the measurements is a brain activity measurement of the subject before and after the presentation of an external stimulus; The processor was used to assess the variability of acquired brain activity measurements to multiple repetitive external stimuli; and Reports on changes in brain response states are generated based on the variability of brain activity measurements over the multiple predetermined time periods. The variability of said brain activity is assessed using a probability density function based on the following brain activity measurements: The brain activity measurements are related to the random variable X, which is recorded at multiple physical brain locations at time τ = 0, and the stimulus presentation at time τ = 0. j (τ) Correspondingly, each physical brain location is indexed by an integer j.

2. The method as described in claim 1, wherein, The probability density function is estimated from the acquired brain activity measurements using a method selected from the group including empirical cumulative distribution function, proportional histogram estimation, and nuclear density estimation.

3. The method as described in claim 1, wherein, Estimating temporally consecutively based on multiple sessions of a designated subject. Each session includes brain activity measurements for multiple repetitive stimuli, and the sessions are separated by intervals selected from multiple hours, multiple days, multiple months, or multiple years.

4. The method of claim 3, wherein, For the designated subjects The continuous estimates provide an indication of changes in brain function between sessions.

5. The method of claim 1, wherein, The brain activity mentioned is the stimulus-evoked activity (ERP) estimated over time. j (τ), which can be estimated differently as one of the following functions: (i)ERP j (τ) = E[X j (τ)], where E[•] is the expectation operator; or (ii)ERP j (τ) = median[X j (t)]; or (iii)ERP j (τ) = mode[X j (t)]。 6. The method of claim 1, wherein, Brain activity measurements based on bandwidth limitations are estimated empirically. .

7. The method of claim 1, wherein, For the designated subjects Continuous assessment provides an indication of brain function during cumulative sessions.

8. The method as described in any one of claims 1 to 7, wherein, The brain activity measurements were obtained by selecting from a modality of electrocorticography (ECoG), electroencephalography (EEG), magnetoencephalography (MEG), oxygen-dependent BOLD functional magnetic resonance imaging (fMRI), and near-infrared spectroscopy (NIRS).

9. The method of claim 1, wherein, The stimulus is selected from a group including auditory, visual, olfactory, or tactile stimuli.

10. The method of claim 1, wherein, The measurements were obtained from multiple brain locations over multiple predetermined time periods.

11. The method of claim 1, wherein, A brain response state is determined by differential entropy according to the following equation: Among them, h j (τ) is the differential entropy of the physical brain location specified by the index j at time τ after the stimulus is presented.

12. The method of claim 11, further comprising: The differential entropy is estimated empirically based on a finite number of samples; as well as Define the change in differential entropy relative to the baseline reference value.

13. The method of claim 11 or claim 12, wherein, The baseline reference values ​​include h when τ < 0 before stimulus presentation. j The value of (τ).

14. The method of claim 11 or claim 12, wherein, The differential entropy is estimated by using one of the techniques selected from the group including histogram-based bias correction estimation, kernel density estimation, and k-nearest neighbor estimation.

15. The method of claim 11 or claim 12, wherein, Based on longitudinal estimation of h from multiple sessions of a specified subject j (τ), wherein each session includes brain activity measurements for multiple repetitive stimuli, and the sessions are separated by intervals selected from multiple hours, multiple days, multiple months, or multiple years.

16. The method of claim 15, wherein, Drawing h indexed by j relative to the physical brain location j Topographic map of longitudinal estimation of (τ).

17. The method of claim 1, wherein, The variability of brain activity is used to obtain one or more quantitative information theoretical measures representing the state of brain response.

18. The method of claim 17, wherein, The one or more quantitative information theoretical measures representing brain function are selected from a group including negative entropy, "spatial average" differential entropy, Kullback-Leibler divergence, and multiscale entropy.

19. The method of claim 18, wherein, The changes in the brain's response state are independent of the changes obtained from the corresponding time set of ERP amplitudes.

20. The method of claim 17, wherein, One or more quantitative information theoretical measures representing brain function are estimated longitudinally from multiple sessions of a designated subject, wherein each session includes brain activity measurements for multiple repetitive stimuli, and the sessions are separated by intervals selected from multiple hours, multiple days, multiple months, or multiple years.

21. The method of claim 17, wherein, The quantitative information theory measure representing brain function, estimated longitudinally from multiple sessions of a specified subject, provides an indication of the specified subject's brain function over a cumulative interval.

22. The method of claim 21, wherein, A topographic map of longitudinal estimates of quantitative information theoretical measures is drawn relative to the physical brain location indexed by j.

23. A system for representing changes in the brain response state of a mammalian subject to multiple repetitive external stimuli, comprising: An acquisition module, comprising a processor, is configured to acquire multiple brain activity measurements of a subject over multiple predetermined time periods, wherein each of the measurements is a brain activity measurement of the subject before and after the presentation of an external stimulus. An evaluation module, comprising a processor, configured to receive the brain activity measurements and evaluate the variability of the brain activity measurements over the plurality of predetermined time periods; and A determination module, comprising a processor, is configured to determine changes in brain response states based on the variability over the plurality of predetermined time periods and to generate a report based on the variability, wherein the variability is evaluated by a processor in an evaluation module configured to utilize a probability density function of the following brain activity: The brain activity is related to the presentation of stimuli at time τ = 0, and the random variable X recorded at multiple physical brain locations at time τ. j (τ) Correspondingly, each physical brain location is indexed by an integer j.

24. The system of claim 23, wherein, The probability density function is estimated from the acquired brain activity measurements using a method selected from the group including empirical cumulative distribution function, proportional histogram estimation, and nuclear density estimation.

25. The system of claim 23, wherein, Estimating temporally consecutively based on multiple sessions of a designated subject. Each session includes brain activity measurements for multiple repetitive stimuli, and the sessions are separated by intervals selected from multiple hours, multiple days, multiple months, or multiple years.

26. The system of claim 25, wherein, For the designated subjects The continuous estimates provide an indication of changes in brain function between sessions.

27. The system of claim 23, wherein, A brain response state is the differential entropy calculated by the processor of the determining module according to the following equation: Among them, h j (τ) is the differential entropy of the physical brain location specified by the index j at time τ after the stimulus is presented.

28. A computer-readable medium comprising program instructions that, when executed by one or more processors, implement the method according to any one of claims 1 to 22.

Images