Patient-specific parameter selection for neurological event detection

• US 7,177,674 B2
• Filed: 05/13/2002
• Issued: 02/13/2007
• Est. Priority Date: 10/12/2001
• Status: Expired due to Term

• Current Assignee: Neuropace Inc.
• Original Assignee: Neuropace Inc.
• Inventors: Tcheng, Thomas K., Pless, Benjamin D., Esteller, Rosana, Echauz, Javier
• Primary Examiner(s): Layno; Carl
• Application Number:
  US10/146,663
• Publication Number:
  US 20030073917A1
• Time in Patent Office: 1,737 Days
• Field of Search: 607 45- 46, 607/60, 600/544
• US Class Current: 600/544
• CPC Class Codes: A61B 5/369 Electroencephalography [EEG] devices for psychotechnics A61B5/16
• CPC Class Codes: A61B 5/372 Analysis of electroencephalograms
• CPC Class Codes: A61B 5/4094 Diagnosing or monitoring seizure diseases, e.g. epilepsy
• CPC Class Codes: A61B 5/7217 of noise originating from a therapeutic or surgical apparatus, e.g. from a pacemaker
• CPC Class Codes: G16H 20/30 relating to physical therapies or activities, e.g. physiotherapy, acupressure or exercising
• CPC Class Codes: G16H 40/60 for the operation of medical equipment or devices

Abstract

An epileptiform activity patient-specific template creation system permits a user to efficiently develop an optimized set of patient-specific parameters for epileptiform activity detection algorithms. The epileptiform activity patient template creation system is primarily directed for use with an implantable neurostimulator system having EEG storage capability, in conjunction with a computer software program operating within a computer workstation having a processor, disk storage and input/output facilities for storing, processing and displaying patient EEG signals. The implantable neurostimulator is operative to store records of EEG data when neurological events are detected, when it receives external commands to record, or at preset or arbitrary times. The computer workstation operates on stored and uploaded records of EEG data to derive the patient-specific templates via a single local minimum variant of a multidimensional greedy line search process and a feature overlay process.

Claims

1. A method for creating a patient-specific template from at least one record of electrographic signal data stored by an implantable device, the method comprising the steps of:

uploading the at least one record of electrographic signal data from the implantable device to an external apparatus;

processing the at least one record to generate at least one parameter; and

forming the patient-specific template from at the at least one parameter by performing a single local minimum variant of a greedy line search.

2. The method for developing a patient-specific template of claim 1, wherein performing a single local minimum variant of a greedy line search comprises the steps of:

selecting a variable parameter from a working parameter set;

adjusting a value of the variable parameter in a first direction and checking a result value until the result value increases or becomes incomputable, or an end of range is reached;

determining whether a local minimum has been identifield, and if not, adjusting a value of the variable parameter in a second direction and checking a result value until the result value increases or becomes incomputable, or an end of range is reached; and

identifying a desired value of the variable parameter at which the result value is at a minimum.

3. The method for developing a patient-specific template of claim 2, further comprising the step of selecting the at least one test record as a subset of the at least one record of electrographic signal data.

4. The method for developing a patient-specific template of claim 2, further comprising the step of incorporating the desired value of the variable parameter into the working parameter set.

5. The method for developing a patient-specific template of claim 2, further comprising the steps of:

selecting an initial parameter set; and

deriving the working parameter set from the initial parameter set.

6. The method for developing a patient-specific template of claim 5, wherein selecting an initial parameter set comprises the steps of:

employing at least one heuristic algorithm to generate at least one initial parameter; and

forming an initial parameter set from the at least one initial parameter.

7. The method or developing a patient-specific template of claim 2, further comprising the step of selecting a termination criterion.

8. The method for developing a patient-specific template of claim 7, further comprising the step of comparing the working parameter set to a stored parameter set.

9. The mothod for developing a patient-specific template of claim 8, wherein the step of comparing the working parameter set to a stored parameter set comprises the steps of:

identifying a plurality of performance factors on the at least one test record;

combining the plurality of performance factors into a working performance metric; and

comparing the working performance metric to a stored performance metric corresponding to the stored parameter set.

10. The method for developing a patient-specific template of claim 9, further comprising the step of replacing the stored parameter set with the working parameter set if the working performance metric is better than the stored performance metric.

11. The method for developing a patient-specific template of claim 8, further comprising the step of determining whether the termination criterion is met.

12. The method for developing a patient-specific template of claim 7, further comprising the step of iteratively testing parameter sets to accomplish a multidimensional search by repeating the steps of selecting a parameter, adjusting a value of the variable parameter in a first direction, determining whether a local minimum has been identified, and if not, adjusting a value of the variable parameter in a second direction, and identifying a desired parameter value, for each variable parameter in the working parameter set.

13. The method for developing a patient-specific template of claim 12, wherein the step of iteratively testing parameter sets is performed repeatedly until the termination criterion is met.

14. The method for developing a patient-specific template of claim 13, wherein the termination criterion comprises a convergence bound.

15. The method for developing a patient-specific template of claim 13, wherein the termination criterion comprises a maximum number of iterations.

16. The method for developing a patient-specific template of claim 2, further comprising the step of converting the patient-specific template into a format recognizable by the implantable device.

17. A method for creating a patient-specific template from at least one record of electrographic signal data stored by an implantable device, the method comprising the steps of:

uploading the at least one record of electrographic signal data from the implantable device to an external apparatus;

processing the at least one record to generate at least one parameter; and

forming the patient-specific template from the at least one parameter by calculating a threshold from a feature overlay.

18. The method for developing a patient-specific template of claim 17, wherein calculating a threshold from a feature overlay comprises the steps of:

selecting a feature to be tested on the at least one record;

calculating an overlay feature based on a value of the feature and a trend for the feature on the at least one record; and

identifying a threshold value for the feature based on the overlay feature.

19. The method for developing a patient-specific template of claim 18, further comprising the steps of checking the at least one record for noise or an artifact, and discarding the at least one record if noise or an artifact is present.

20. The method for developing a patient-specific template of claim 18, further comprising the steps of checking a length of the at least one record, and discarding the at least one record the length is too short.

21. The method for developing a patient-specific template of claim 18, further comprising the steps of checking a location of an annotation within the at least one record, and discarding the at least one record if the annotation is too near a beginning or an end of the at least one record.

22. The method for developing a patient-specific template of claim 18, further comprising the step of calculating at least one statistic from the overlay feature.

23. The method for developing a patient-specific template of claim 22, wherein the at least one statistic comprises a maximum value.

24. The method for developing a patient-specific template of claim 23, further comprising the steps of:

determining whether the at least one record is a baseline record type or an event-containing record type; and

updating a statistical value for a category associated with the record type.

25. The method for developing a patient-specific template of claim 24, wherein the statistical value comprises overlay feature maxima.

26. The method for developing a patient-specific template of claim 25, wherein the at least one record is an event-containing record type, and wherein the overlay feature maxima are obtained from a region surrounding an annotation of the at least one record.

27. The method for developing a patient-specific template of claim 26, wherein the region comprises a time band of approximately ±

12.8 seconds around an annotation of unequivocal seizure onset.

28. The method for developing a patient-specific template of claim 25, wherein the step of identifying a threshold value for the feature comprises the steps of:

identifying a maximum value of maxima associated with baseline record types;

identifying a minimum value of maxima associated with event-containing record types;

setting a threshold at a desired interval between the minimum value and the maximum value.

29. The method for developing a patient-specific template of claim 28, wherein the desired interval is halfway between the minimum value and the maximum value.

30. The method for developing a patient-specific template of claim 28, wherein the desired interval is 5% between the minimum value and the maximum value.

31. The method for developing a patient-specific template of claim 28, wherein the desired interval is 10% between the minimum value and the maximum value.

32. The method for developing a patient-specific template of claim 29, further comprising the step of rounding the threshold up to a discrete value.

33. The method for developing a patient-specific template of claim 25, wherein the step of identifying a threshold value for the feature comprises the steps of:

identifying a maximum value of maxima associated with baseline record types;

setting a threshold a desired interval higher than the maximum value.

34. The method for developing a patient-specific template of claim 33, wherein the desired interval is 5%.

35. The method for developing a patient-specific template of claim 34, further comprising the step of rounding the threshold up to a discrete value.

36. The method for developing a patient-specific template of claim 18, wherein the overlay feature comprises a difference between the value of the feature and the trend for the feature, and wherein the threshold value comprises a fixed offset threshold.

37. The method for developing a patient-specific template of claim 18, wherein the overlay feature comprises a quotient between the value of the feature and the trend for the feature, and wherein the threshold value comprises a percentage offset threshold.

38. The method for developing a patient-specific template of claim 17, further comprising the step of selecting the at least one test record as a subset of the at least one record of electrographic signal data.

39. The method for developing a patient-specific template of claim 17, further comprising the step of incorporating the desired value of the variable parameter into the working parameter set.

40. The method for developing a patient-specific template of claim 17, further comprising the steps of:

selecting an initial parameter set; and

deriving the working parameter set from the initial parameter set.

41. The method for developing a patient-specific template of claim 40, wherein selecting an initial parameter set comprises the steps of:

employing at least one heuristic algorithm to generate at least one initial parameter; and

forming an initial parameter set from the at least one initial parameter.

42. The method for developing a patient-specific template of claim 17, further comprising the step of converting the patient-specific template into a format recognizable by the implantable device.

43. A method for creating a patient-specific template from at least one record of electrographic signal data stored by an implantable device, the method comprising the steps of:

uploading the at least one record of electrographic signal data from the implantable device to an external apparatus;

determining whether the at least one record is a statistical outlier, and if so, discarding the at least one record;

processing the at least one record to generate at least one parameter; and

forming the patient-specific template from the at least one parameter.

44. The method for developing a patient-specific template of claim 43, wherein the step of determining whether the at least one record is a statistical outlier comprises the steps of:

calculating at least one statistic from the record, wherein the at least one statistic comprises a maximum value;

determining whether the at least one record is a baseline record type or an event-containing record type;

updating a plurality of statistical values for a category associated with the record type, wherein the plurality of statistical values comprises a mean, a standard deviation, and a maximum of maxima;

identifying the at least one record as an outlier based on a comparison of the maximum value and at least one of the plurality of statistical values.

45. The method for developing a patient-specific template of claim 44, wherein the identifying step comprises the steps of:

if the at least one record is a baseline record type, identifying the at least one record as an outlier if its maximum value is greater or equal to than two standard deviations above the mean for baseline record types; and

if the at least one record is an event-containing record type, identifying the at least one record as an outlier if its maximum is greater than or equal to one standard deviation below the mean or event-containing record types.

46. The method for developing a patient-specific template of claim 44, wherein the identifying step comprises the steps of:

if the at least one record is a baseline record type, identifying the at least one record as an outlier if its maximum value is greater than or equal to two standard deviations above theme for baseline record types; and

if the at least one record is an event-containing record type, identifying the at least one record as an outlier if its maximum is greater than or equal to one standard deviation below the mean for event-containing record types and its maximum is lower than the maximum for baseline record types.

47. A method for creating a patient-specific template from at least one record of electrographic signal data stored by an implantable device, the method comprising the steps of:

unloading the at least one record of electrographic signal data from the implantable device to external apparatus;

processing the at least one record to generate at least one parameter for each of a plurality of detection tools; and

combining the at least one parameter for each of the plurality of detection tools into a template;

wherein each of the plurality of detection tools has a detection persistence, and wherein the combining step comprises identifying a preferred persistence value for each of the plurality of detection tools.

48. The method for creating a patient-specific template of claim 47, wherein the step of identifying preferred persistence value comprises performing a greedy line search across the plurality of detection tools.

49. The method for creating a patient-specific template of claim 47, wherein the step of identifying preferred persistence value comprises performing a single local minimum variant of a greedy line search across the plurality of detection tools.

50. The method for creating a patient-specific template of claim 47, wherein the step of identifying preferred persistence value comprises performing an exhaustive search across the plurality of detection tools.

51. The method for creating a patient-specific template of claim 47, wherein the combining step further comprises the step of identifying a preferred subset of the plurality of detection tools.