Detection of epileptogenic brains with non-linear analysis of electromagnetic signals

• US 10,278,608 B2
• Filed: 09/06/2013
• Issued: 05/07/2019
• Est. Priority Date: 09/07/2012
• Status: Active Grant

• Current Assignee: Children's Medical Center Corporation
• Original Assignee: Children's Medical Center Corporation
• Inventors: Bosl, William J., Loddenkemper, Tobias
• Primary Examiner(s): Jang, Christian
• Application Number:
  US14/426,424
• Publication Number:
  US 20150216436A1
• Time in Patent Office: 2,069 Days
• CPC Class Codes: A61B 5/316 Modalities, i.e. specific diagnostic methods
• CPC Class Codes: A61B 5/369 Electroencephalography [EEG] devices for psychotechnics A61B5/16
• CPC Class Codes: A61B 5/4094 Diagnosing or monitoring seizure diseases, e.g. epilepsy
• CPC Class Codes: A61B 5/7253 characterised by using transforms
• CPC Class Codes: A61B 5/726 using Wavelet transforms
• CPC Class Codes: A61B 5/7264 Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems neural networks per se G06N3/00; expert systems per se G06N5/00
• CPC Class Codes: A61B 5/7275 Determining trends in physiological measurement data; Predicting development of a medical condition based on physiological measurements, e.g. determining a risk factor
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: G16H 50/30 for calculating health indices; for individual health risk assessment
• CPC Class Codes: G16Z 99/00 Subject matter not provided for in other main groups of this subclass

Abstract

Methods and apparatus for identifying and using at least one nonlinear feature determined from multiscale electroencephalography (EEG) data to evaluate an epileptogenicity level of a patient is described. A multiscale algorithm is applied to EEG data recorded from the patient to produce scaled EEG data. At least one nonlinear feature value for the received EEG data and/or the scaled EEG data is determined and the at least one nonlinear feature value is used, at least in part, to evaluate the epileptogenicity level of the patient.

Claims

1. A method of treating a patient having epilepsy based on an quantitative assessment of an epileptogenicity level of the patient'"'"'s brain, the method comprising:

storing, in a database, a multi-dimensional dataset including non-linear feature set values for a first plurality of patients diagnosed with epilepsy and a second plurality of patients without an epilepsy diagnosis;

receiving electroencephalography (EEG) data recorded from the brain of the patient, wherein the received EEG data includes a time series of EEG data that does not include and is not associated with epileptiform activity;

generating, for the time series of EEG data that does not include and is not associated with epileptiform activity, a plurality of additional time series of EEG data by applying, using at least one computer processor, a multiscale algorithm to the time series of EEG data;

calculating, for the time series of EEG data and each of the generated plurality of additional time series of EEG data, values for each of plurality of non-linear features in a set of non-linear features, wherein the set of non-linear features includes sample entropy and a plurality of recurrence quantitative analysis features;

determining, using at least some of the calculated values for the plurality of non-linear features in the set of non-linear features, the multi-dimensional dataset stored in the database, and a statistical learning algorithm, the epileptogenicity level of the patient'"'"'s brain; and

treating the patient as a result of the determined epileptogenicity level.

2. The method of claim 1, wherein applying the multiscale algorithm comprises applying a wavelet transform to the received EEG data.

3. The method of claim 2, wherein the wavelet transform is a Haar wavelet transform with scales defined by factors of two.

4. The method of claim 1, wherein applying the multiscale algorithm comprises applying successive averaging of the received EEG data to create a coarse-grained time series.

5. The method of claim 1, wherein the plurality of recurrence quantitative analysis features includes recurrence rate, determinism, laminarity, or any combination thereof.

6. The method of claim 1, wherein the received EEG data does not include activity related to hyperventilation.

7. The method of claim 1, further comprising:

recording the time series of EEG data from the brain of the patient.

8. The method of claim 1, wherein the time series of EEG data is at least ten seconds in duration.

9. The method of claim 1, further comprising:

normalizing the calculated values for the plurality of non-linear features prior to using the calculated values to determine the epileptogenicity level of the patient'"'"'s brain.

10. The method of claim 1, wherein the statistical learning algorithm uses a support vector machine classifier.

11. The method of claim 1, wherein the received EEG data comprises EEG data recorded from a plurality of channels, and wherein calculating values for each of a plurality of non-linear features in a set of non-linear features comprises calculating values for each of the plurality of non-linear features for each of the plurality of EEG channels of the received EEG data.

12. The method of claim 1, wherein the received EEG data comprises EEG data recorded from a plurality of channels, and wherein calculating values for each of a plurality of non-linear features in a set of non-linear features comprises calculating values for each of the plurality of non-linear features for a subset of the plurality of EEG channels of the received EEG data.

13. The method of claim 12, wherein the subset of the plurality of EEG channels includes a single EEG channel.

14. The method of claim 12, wherein the subset of the plurality of EEG channels includes less than all of the plurality of EEG channels.

15. The method of claim 1, further comprising:

generating at least one scalp map, wherein the at least one scalp map is generated based, at least in part, on the calculated values for each of the plurality of non-linear features.

16. A non-transitory computer-readable medium, encoded with a plurality of instructions that, when executed by a computer, performs a method of selecting a treatment for a patient, the method comprising:

storing, in a database, a multi-dimensional dataset including non-linear feature set values for a first plurality of patients diagnosed with epilepsy and a second plurality of patients without an epilepsy diagnosis;

generating, for a time series of electroencephalograpy (EEG) data recorded from the brain of the patient that does not include and is not associated with epileptiform activity, a plurality of additional time series of EEG data by applying a multiscale algorithm to the time series of EEG data;

calculating, for the time series of EEG data and each of the generated plurality of additional time series of EEG data, values for each of plurality of non-linear features in a set of non-linear features, wherein the set of non-linear features includes sample entropy and a plurality of recurrence quantitative analysis features;

determining, using at least some of the calculated values for the plurality of non-linear features in the set of non-linear features, the multi-dimensional dataset stored in the database, and a statistical learning algorithm, the epileptogenicity level of the patient'"'"'s brain; and

selecting a treatment for the patient as a result of the determined epileptogenicity level.

17. A computer system, comprising:

a database having stored thereon, a multi-dimensional dataset including non-linear feature set values for a first plurality of patients diagnosed with epilepsy and a second plurality of patients without an epilepsy diagnosis;

an input interface configured to receive electroencephalography (EEG) data recorded from a brain of a patient, wherein the received EEG data includes a time series of EEG data that does not include and is not associated with epileptiform activity;

at least one computer processor programmed to;

generate, for the time series of EEG data that does not include and is not associated with epileptiform activity, a plurality of additional time series of EEG data by applying a multiscale algorithm to the-time series of EEG data;

calculate, for the time series of EEG data and each of the generated plurality of additional time series of EEG data, values for each of plurality of non-linear features in a set of non-linear features, wherein the set of non-linear features includes sample entropy and a plurality of recurrence quantitative analysis features;

determine, using at least some of the calculated values for the plurality of non-linear features in the set of non-linear features, the multi-dimensional dataset stored in the database, and a statistical learning algorithm, the epileptogenicity level of the patient'"'"'s brain; and

select a treatment for the patient as a result of the determined epileptogenicity level; and

an output interface configured to output an indication of the selected treatment.