Search

EP-4000523-B1 - METHOD OF FORMING MODIFYING DATA RELATED TO DATA SEQUENCE OF DATA FRAME INCLUDING ELECTROENCEPHALOGRAM DATA, PROCESSING METHOD OF ELECTROENCEPHALOGRAM DATA AND ELECTROENCEPHALOGRAM APPARATUS

EP4000523B1EP 4000523 B1EP4000523 B1EP 4000523B1EP-4000523-B1

Inventors

  • VÄYRYNEN, Eero
  • ALA-KURIKKA, Jussi
  • SALMINEN, Miikka
  • TOBÓN, Marcela
  • NIEMELÄ, Erika

Dates

Publication Date
20260506
Application Date
20201111

Claims (16)

  1. A computer-implemented method of forming modifying data related to a data sequence for a data frame (100) including electroencephalogram data, characterized by forming (800) the modifying data by performing a selection of one sequence from at least one surrogate data sequence (662), where the selected one becomes a neutral data sequence (500), the neutral data sequence (500) replacing a missing or corrupted data sequence of the electroencephalogram data, by forming (802) the neutral data sequence (500) by optimizing an error between a first data and a second data, wherein the first data comprises a reference result formed by applying a result algorithm to reference data, which corresponds to electroencephalogram data that is non-corrupted and without missing portions, and the second data comprises at least one result each formed by applying the result algorithm to the electroencephalogram data with the one sequence from the at least one surrogate data sequence (662) in the location of the corrupted or missing data sequence (102), the result algorithm providing characterizing information on the data frame (100) including the electroencephalogram data, wherein forming the neutral data sequence (500) comprising performing the selection to form the neutral data sequence (500) by means of an optimization comparison (606) based on the error between the first data and the second data in order to limit disturbance caused when the neutral data sequence (500) applied to the data frame (100), for further processing that utilizes the neutral data sequence (500) and the data frame.
  2. The computer-implemented method of claim 1, characterized by forming (802) the neutral data sequence (500) by optimizing, in the optimization comparison (606), an error between the first and second data, the second data being formed by applying the result algorithm to the electroencephalogram data and the at least one surrogate data sequence (662), and the first data being formed by applying the result algorithm to the reference data, which corresponds to the electroencephalogram data.
  3. The computer-implemeted of claim 1, characterized by combining (804) the data frame (100) including the electroencephalogram data and the neutral data sequence (500).
  4. The computer-implemented method of claim 1, characterized by performing the optimization of the error by substituting a corrupted or missing data sequence (102) of the electroencephalogram data with the at least one surrogate data sequence (662) for forming the first data by applying the result algorithm to the electroencephalogram data with the at least one surrogate data sequence (662); and replacing the corrupted or missing data sequence (102) of the electroencephalogram data with a neutral data sequence (500) formed by the optimization comparison (606).
  5. The computer-implemented method of claim 1, characterized by receiving the electroencephalogram data from a plurality of electroencephalogram channels (N); processing the plurality of electroencephalogram channels as a vector such that a data frame (100) of a single channel of the plurality of the electroencephalogram channels is processed as a single element of the vector; and performing the optimization comparison (606) to at least one element of the vector.
  6. The computer-implemented method of claim 1, characterized by forming a plurality of the surrogate data sequences (662) for a plurality of the data frames (100), and selecting, for each of the data frames (100), only one of the neutral data sequence (500) based on the optimization comparison.
  7. The computer-implemented method of claim 1, characterized by applying (1000) the result algorithm to the electroencephalogram data with the neutral data sequence for providing at least one measured parameter, the neutral data sequence (500) for the electroencephalogram data; and outputting (1002) the at least one measured parameter.
  8. The computer-implemented method of claim 1, characterized by forming the neutral data sequence (500) for the data frame (100) including the electroencephalogram data in real time while a measurement of the electroencephalogram data or transfer thereof is on-going.
  9. The computer-implemented method of claim 1, characterized by performing the selection of one surrogate algorithm (652) from at least one surrogate algorithm (652), each of which is for generating at least one surrogate data sequence (662), which includes a neutral data sequence (500), the selection being based on an optimization comparison (606) between a first data and a second data in order to limit disturbance caused by the neutral data sequence (500) applied to the data frame (100).
  10. The computer-implemented method of claim 9, characterized by forming the surrogate data sequence (662) using the surrogate algorithm (652), which utilizes the electroencephalogram data that is uncorrupted, the surrogate algorithm (652) being dependent on the result algorithm.
  11. The computer-implemented method of claim 9, characterized by forming (900), based on the result algorithm, at least one first parameter, which defines the surrogate algorithm (652); forming (902) the at least one surrogate data sequence (662) using the surrogate algorithm (652) determined by the at least one first parameter for the optimization comparison (606), the surrogate algorithm (652) being random, pseudorandom or deterministic; and combining (904) the data frame (100) and a neutral data sequence (500) formed by the optimization (662).
  12. The computer-implemented method of claim 10 or 11, characterized by determining the at least one first parameter of the surrogate algorithm (652) based on at least one frame parameter of the data frame (100), the at least one frame parameter defining a property of the data frame (100).
  13. The computer-implemented method of claim 10, 11 or 12, characterized by determining the at least one first parameter of the surrogate algorithm (652) based on at least one second parameter of the result algorithm, the at least one second parameter defining the result algorithm.
  14. The computer-implemented method of claim 9, characterized by forming, for the optimization comparison, at least one reference result by applying the result algorithm to at least one model electroencephalogram data, a single reference result corresponding to a single model electroencephalogram data of the at least one model electroencephalogram data, and/or at least one reference algorithm that provides at least one reference result similar to those formed by applying the result algorithm to at least one model electroencephalogram data.
  15. An electroencephalogram apparatus for forming modifying data for a data sequence of a data frame (100) including electroencephalogram data, characterized in that the electroencephalogram apparatus comprises one or more processors (702); and one or more memories (704) including computer program code; the one or more memories (704) and the computer program code configured to, with the one or more processors (702), cause the electroencephalogram apparatus at least to: form the modifying data by performing a selection of one sequence from at least one surrogate data sequence (662), where the selected one is configured to become a neutral data sequence (500), the neutral data sequence (500) replacing a missing or corrupted data sequence of the electroencephalogram data, and form the neutral data sequence (500) by optimization, within the optimization comparison (606), of an error between a first data and a second data, wherein the first data comprises a reference result is formed by application of a result algorithm to reference data corresponding to electroencephalogram data that is non-corrupted and without missing portions, and the second data comprises at least one result formed by application of the result algorithm to the electroencephalogram data with the one sequence from the at least one surrogate data sequence (662) in the location of the corrupted or missing data sequence (102), the result algorithm providing characterizing information on the data frame (100) including the electroencephalogram data, wherein the formation of the neutral data sequence (500) comprises perform the selection to form the neutral data sequence (500) based on an optimization comparison (606) between a first data and a second data in order to limit disturbance caused in case the neutral data sequence (500) is applied to the data frame (100).
  16. The apparatus of claim 15, characterized in that the one or more memories (704) and the computer program code configured to, with the one or more processors (702), cause the electroencephalogram apparatus to: perform a selection of one surrogate algorithm (652) from at least one surrogate algorithm (652), each of which is for generating at least one surrogate data sequence (662), which includes the neutral data sequence (500), based on an optimization comparison (606) between the first data and the second data in order to limit disturbance caused when the neutral data sequence (500) is applied to the data frame (100).

Description

Field The invention relates to a method of forming modifying data related to a data sequence of a data frame including electroencephalogram data, and an electroencephalogram apparatus. Background Electroencephalography (EEG) signals can be analysed by performing a qEEG (quantitative EEG) analysis using algorithms of computer programs. However, such a computer based analysis is sensitive to situations where some raw EEG data is missing because of a data sequence lost due to communication errors, for example. Alternatively or additionally, the raw EEG data may be contaminated by artefacts. Such undesired phenomena may render the EEG analysis incalculable or result in a large analysis error. The resending of lost or contaminated data is not always practical or possible. The reason may be that the analysis needs to be done in real-time, for example. Sometimes the data frame requires an additional data sequence in order to be analysed. Patent document US7672717 presents a method and a system for the denoising of large-amplitude artifacts in electrograms using time-frequency transforms. Patent document US6042548 presents a virtual neurological monitor and method. Urigüen, J.A., Garcia-Zapirain, B., EEG artifact removal-state-of the-art and guidelines, Journal of Neural Engineering, Vol. 12, No. 3, 2015, Article 031001. DOI: 10.1088/1741-2560/12/3/031001 presents a review on the artifact removal algorithms used to remove the main sources of interference encountered in the electroencephalogram (EEG), specifically ocular, muscular and cardiac artifacts. The prior art has attempted to solve the problem by replacing a lost or contaminated data sequence of an EEG data frame by a copy of a properly received data sequence of an EEG data frame which is contamination free or by adding a proper EEG data frame to the data frame. However, such a substitution is also known to result in a large analysis error. Hence, the present signal processing is inadequate and an improvement would be welcome. Bahador, Nooshin and Jokelainen, Jarno and Mustola, Seppo and Kortelainen, Jukka: "Reconstruction of missing channel in electroencephalogram using spatiotemporal correlation-based averaging",vol. 18, no. 5 5 October 2021 (2021-10-05),DOI: 10.1088/1741-2552/ac23e2. Internet:URL:https:// doi.org/10.1088/1741-2552/ac23e2 teaches correlation-based averaging of neighbouring channels, producing one deterministic reconstruction of missing or corrupted EEG data according to prior art. Brief description The present invention seeks to provide an improvement in the formation of a data sequence and processing method of the data frame(s). The invention is defined by the independent claims. Embodiments are defined in the dependent claims. List of drawings Example embodiments of the present invention are described below, by way of example only, with reference to the accompanying drawings, in which Figure 1 illustrates an example of a data frame of a signal with a unusable data sequence, where data may be missing or corrupted;Figure 2 illustrates an example of a data frame with a copy of an available data sequence of the data frame is pasted in a location of a corrupted or missing data sequence;Figure 3 illustrates an example of a data frame with a shuffled copy of an available data sequence of the data frame is pasted in a location of a corrupted or missing data sequence;Figure 4 illustrates an example of a data frame with a copy of another available data sequence of the data sequence is pasted in a location of a corrupted or missing data sequence;Figure 5A illustrates an example of a neutral data sequence of a good fit pasted in a location of a corrupted or missing data sequence;Figure 5B illustrates an example of a data frame pasted in front of the data sequence of the data frame;Figure 6A illustrates an example how process is performed;Figure 6B illustrates an example of selection of a surrogate algorithm for forming a neutral data sequence;Figure 6C illustrates an example of selection of a neutral data sequence from a plurality of candidates;Figure 7 illustrates an example of electroencephalographic measurement with a data processing unit comprising at least one processor and at least one memory;Figure 8 illustrates of an example of a flow chart of a method to form modifying data for a data frame including electroencephalogram data;Figure 9 illustrates an example of a flow chart of a method utilizing at least one first parameter, which defines the surrogate algorithm of candidates of neutral data sequences;Figure 10 illustrates an example of of a flow chart of a method of a processing method of electroencephalogram data. Description of embodiments The following embodiments are only examples. Although the specification may refer to "an" embodiment in several locations, this does not necessarily mean that each such a reference is to the same embodiment(s), or that the feature only applies to a single embodiment. Single features of different embodim