DETECTION OF EPILEPTOGENIC BRAINS WITH NON-LINEAR ANALYSIS OF ELECTROMAGNETIC SIGNALS

• US 20150216436A1
• Filed: 09/06/2013
• Published: 08/06/2015
• Est. Priority Date: 09/07/2012
• Status: Active Application

• Current Assignee: Children's Medical Center Corporation
• Original Assignee: Children's Medical Center Corporation
• Inventors: Bosl, William J., Loddenkemper, Tobias
• Granted Patent: US 10,278,608 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• 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 diagnosing a patient as having epilepsy, the method comprising:

receiving electroencephalography (EEG) data recorded from the patient;

applying, using at least one computer processor, a multiscale algorithm to the received EEG data to produce scaled EEG data;

determining at least one nonlinear feature value for the received EEG data and/or the scaled EEG data; and

diagnosing the patient as having epilepsy based, at least in part, on the at least nonlinear feature value.

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 the scaled EEG data as a coarse-grained time series.

5. The method of claim 1, wherein determining at least one nonlinear feature value comprises determining an entropy value for the received EEG data and/or the scaled EEG data.

6. The method of claim 1, wherein determining at least one nonlinear feature value comprises determining at least one recurrence quantitative analysis value.

7. The method of claim 6, wherein the at least one recurrence quantitative analysis value comprises recurrence rate, determinism, laminarity, or any combination thereof.

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

9. The method of claim 1, further comprising:

recording a time series of EEG data from the patient, wherein the received EEG data is a segment of the recorded time series of EEG data.

10. The method of claim 9, wherein the segment of the recorded time series of EEG data is at least ten seconds in duration.

11. The method of claim 1, further comprising:

normalizing the at least one nonlinear feature value prior to using the at least one nonlinear feature value for diagnosis.

12. The method of claim 1, wherein diagnosing the patient based, at least in part, on the at least one nonlinear feature value comprises using a statistical learning algorithm to determine whether the patient is likely to have epilepsy.

13. The method of claim 12, wherein the statistical learning algorithm uses a support vector machine classifier.

14. The method of claim 1, wherein the received EEG data does not include epileptiform activity.

15. The method of claim 1, wherein the at least one nonlinear feature value comprises a set of nonlinear feature values, wherein the set of nonlinear feature values includes at least one nonlinear feature value determined for at least two scales of the scaled EEG data.

16. The method of claim 15, wherein diagnosing the patient based, at least in part, on the at least one nonlinear feature value comprises diagnosing the patient based, at least in part, on the set of nonlinear feature values.

17. The method of claim 1, wherein the received EEG data comprises EEG data from a plurality of channels, and wherein determining the at least one nonlinear feature value comprises determining the at least one nonlinear feature value for each of the plurality of EEG channels of the received EEG data.

18. The method of claim 1, wherein the received EEG data comprises EEG data from a plurality of channels, and wherein determining the at least one nonlinear feature value comprises determining the at least one nonlinear feature value for a subset of the plurality of EEG channels of the received EEG data.

19. The method of claim 18, wherein the subset of the plurality of EEG channels includes a single EEG channel.

20. The method of claim 18, wherein the subset of the plurality of EEG channels includes less than all of the plurality of EEG channels.

21. 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 at least one nonlinear feature value.

22. A computer-readable medium, encoded with a plurality of instructions that, when executed by a computer, performs a method of diagnosing a patient as having epilepsy, the method comprising:

applying a multiscale algorithm to electroencephalograpy (EEG) data recorded from the patient to produce scaled EEG data;

determining at least one nonlinear feature value for the EEG data recorded from the patient and/or the scaled EEG data; and

diagnosing the patient as having epilepsy based, at least in part, on the at least nonlinear feature value.

23. A computer, comprising:

an input interface configured to receive electroencephalography (EEG) data recorded from a patient;

at least one processor programmed to;

apply a multiscale algorithm to the received EEG data to produce scaled EEG data;

determine at least one nonlinear feature value for the received EEG data and/or the scaled EEG data; and

determine whether the patient has epilepsy based, at least in part, on the at least nonlinear feature value; and

an output interface configured to output an indication of the determination whether the patient has epilepsy.

24. A method of identifying a set of nonlinear features useful for evaluating an epileptogenicity level of a patient, the method comprising:

receiving first EEG data from patients with epilepsy and second EEG data from patients without epilepsy;

generating, using at least one processor, first multiscaled EEG data from the first EEG data and second multiscaled EEG data from the second EEG data;

determining first nonlinear feature values for the first multiscaled EEG data and second nonlinear feature values for the second multiscaled EEG data; and

determining the set of nonlinear features based, at least in part on a comparison of the first nonlinear feature values and the second nonlinear feature values.

25. The method of claim 24, wherein the set of nonlinear features includes a single nonlinear feature.

26. The method of claim 25, wherein values for the single nonlinear feature are determined for multiple scales of the corresponding first and second multiscaled EEG data.

27. The method of claim 24, wherein the set of nonlinear features comprises at least two nonlinear features, and wherein each of the at least two nonlinear features are determined for multiple scales of the corresponding first and second multiscaled EEG data.

28. The method of claim 24, wherein determining the set of nonlinear features based, at least in part on a comparison of the first nonlinear feature values and the second nonlinear feature values comprises performing a cross-validation computation by applying a statistical learning algorithm to the first nonlinear feature values and the second nonlinear feature values.

29. The method of claim 28, wherein the statistical learning algorithm uses a support vector machine classifier.

30. The method of claim 28, wherein the cross-validation computation is performed for combinations of two, three, or four nonlinear feature values, on each scale of the first and second multiscale EEG data.

31. The method of claim 24, wherein determining the set of nonlinear features based, at least in part on a comparison of the first nonlinear feature values and the second nonlinear feature values comprises determining which combination of nonlinear feature values demonstrate the highest sensitivity in distinguishing the first EEG data and the second EEG data.

32. The method of claim 24, wherein determining first nonlinear feature values for the first multiscaled EEG data and second nonlinear feature values for the second multiscaled EEG data comprises determining the first nonlinear feature values and the second nonlinear feature values corresponding to at least one recurrence quantitative analysis value.

33. The method of claim 24, further comprising:

recording the first EEG data from the patients with epilepsy and the second EEG data from the patients without epilepsy, wherein the first EEG data and the second EEG data does not include epileptiform activity.

34. The method of claim 24, wherein the first EEG data and the second EEG data, each include EEG data recorded from a plurality of EEG channels, and wherein the method further comprises:

determining a subset of the plurality of EEG channels exhibiting significant differences in at least one nonlinear feature value between the first EEG data and the second EEG data.

35. The method of claim 34, further comprising:

generating at least one scalp map identifying the subset of the plurality of EEG channels.