Automatic determination of the threshold of an evoked neural response

• US 8,965,520 B2
• Filed: 06/15/2005
• Issued: 02/24/2015
• Est. Priority Date: 06/15/2004
• Status: Expired due to Fees

• Current Assignee: Cochlear Limited
• Original Assignee: Cochlear Limited
• Inventors: Botros, Andrew, Dijk, Bastiaan van, Killian, Matthijs
• Primary Examiner(s): Koharski, Christopher D
• Assistant Examiner(s): So, Elizabeth K
• Application Number:
  US10/569,054
• Publication Number:
  US 20080319508A1
• Time in Patent Office: 3,541 Days
• US Class Current: 607/57
• CPC Class Codes: A61B 5/24 Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
• CPC Class Codes: A61B 5/388 Nerve conduction study, e.g. detecting action potential of peripheral nerves
• 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: A61B 5/7282 Event detection, e.g. detecting unique waveforms indicative of a medical condition A61B5/4094, A61B5/4818, A61B5/349, A61B5/0823 take precedence
• CPC Class Codes: A61N 1/00 Electrotherapy; Circuits therefor A61N2/00 takes precedence; irradiation apparatus A61N5/00
• CPC Class Codes: A61N 1/0541 Cochlear electrodes
• CPC Class Codes: A61N 1/36039 fitting procedures
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems

Abstract

Automatically analyzing neural activity. The method comprises: applying electrical stimulation to a target neural region at an initial current level that approximates a typical threshold-Neural Response Telemetry (NRT) level; recording an NRT measurement of neural activity within the target neural region in response to the stimulation; and utilizing a machine-learned expert system to predict, based on one or more features of the neural activity, whether the NRT measurement contains a neural response.

Claims

1. A method for automatically analyzing neural activity within a target neural region comprising:

applying electrical stimulation to the target neural region at an initial current level that approximates a typical threshold-Neural Response Telemetry (NRT) level;

recording an NRT measurement of neural activity within the target neural region in response to the stimulation; and

utilizing a machine-learned expert system configured with a decision tree that includes at least two levels of nodes which consider parameters relating to the NRT measurement, respectively, so as to predict, based on one or more features of the neural activity, whether the NRT measurement includes a neural response or does not include a neural response, the step of utilizing including;

evaluating at least two successive ones of the levels of nodes in order to predict whether the NRT measurement includes a neural response or does not include a neural response, wherein at least one of the evaluated levels includes a node which considers a correlation coefficient relating the NRT measurement to a predetermined value.

2. The method of claim 1, further comprising:

selecting the initial current level to insure safety while minimizing a quantity of measurements required to determine the threshold-NRT (T-NRT) level of the target neural region.

3. The method of claim 2, wherein selecting the initial current level comprises:

in intra-operative applications, selecting as an initial current level a current level at which a neural response is not expected to be evoked.

4. The method of claim 3, wherein selecting the initial current level comprises:

selecting an initial current level substantially below the typical T-NRT.

5. The method of claim 2, wherein selecting the initial current level comprises:

in intra-operative applications, selecting an initial current level that is above or below the typical T-NRT.

6. The method of claim 1, wherein when the machine-learned expert system predicts that the NRT measurement does not contain a neural response, the method further comprises:

incrementing the current level of the electrical stimulus;

repeating the applying, recording and utilizing steps; and

repeating the incrementing and repeating steps until a neural response has been evoked.

7. The method of claim 1, further comprising:

locally establishing threshold-NRT (T-NRT) at a finer stimulus resolution.

8. The method of claim 1, wherein when the machine-learned expert system predicts that the NRT measurement does not contain a neural response, the method further comprises:

decreasing the current level of the electrical stimulus;

repeating the applying, recording and utilizing steps; and

repeating the decreasing and repeating steps until a neural response has been evoked.

9. The method of claim 1, wherein at least one of the evaluated levels in the decision tree includes at least one node which respectively considers one or more of:

a parameter relating to a ratio of the peak-to-peak amplitude of the measurement to the noise of the measurement;

a parameter relating to a correlation of the measurement to a predefined expected neural response;

a parameter relating to the correlation of the measurement to a predefined expected trace containing neural response plus stimulus artifact;

a parameter relating to the correlation of the measurement to a predefined expected trace containing stimulus artifact only;

a parameter relating to the correlation of the measurement to a previous measurement with neural stimulus of similar level; and

a parameter relating to a stimulus current level.

10. A cochlear implant system comprising:

a cochlear implant configured to communicate with an external device;

an automatic neural response measurement system, communicably coupled to the cochlear implant, configured to record a Neural Response Telemetry (NRT) measurement of neural activity within a cochlea in response to electrical stimulation of the cochlea by electrodes of the cochlear implant, wherein an initial current level of electrical stimulation applied to evoke neural activity is set to a level which approximates a typical threshold-NRT level; and

an expert system configured with a decision tree that includes at least two levels of nodes which consider parameters relating to the NRT measurement, respectively, so as to determine, based on one or more features of the evoked neural activity, whether the NRT measurement includes a neural response or does not include a neural response, the expert system being further configured with a machine learning algorithm operable to evaluate at least two successive ones of the levels of nodes in order to predict whether the NRT measurement includes a neural response or does not include neural response, at least one of the evaluated levels including a node which considers a correlation coefficient relating the NRT measurement to a predetermined value.

11. The cochlear implant system of claim 10, wherein at least one of the evaluated levels in the decision tree includes at least one node which respectively considers one or more of:

a parameter relating to the ratio of the peak-to-peak amplitude of the measurement to the noise of the measurement;

a parameter relating to the correlation of the measurement to a predefined expected neural response;

a parameter relating to the correlation of the measurement to a predefined expected trace containing neural response plus stimulus artifact;

a parameter relating to the correlation of the measurement to a predefined expected trace containing stimulus artifact only;

a parameter relating to the correlation of the measurement to a previous measurement with neural stimulus of similar level; and

a parameter relating to a stimulus current level.

12. A method for determining a minimum stimulus threshold at which a neural response is evoked, comprising:

applying a neural stimulus;

obtaining a Neural Response Telemetry (NRT) measurement of neural activity responsive to the neural stimulus;

determining whether the neural stimulus has or has not evoked a neural response by applying an expert system to the NRT measurement, wherein the expert system considers one or more features extracted from the NRT measurement including a signal to noise ratio thereof, comprising;

evaluating at least two successive levels of nodes of a decision tree in order to predict whether the neural stimulus has or has not evoked a neural response, wherein at least one of the evaluated levels includes a node which considers a correlation coefficient relating the NRT measurement to a predetermined value.

13. The method of claim 12, wherein the neural stimulus is applied to a target neural region at an initial current level that approximates a typical threshold-Neural Response Telemetry (NRT) level, the method further comprising:

selecting the initial current level to insure safety while minimizing a quantity of measurements required to determine a threshold-NRT level of the target neural region.

14. The method of claim 13, wherein selecting the initial current level comprises:

selecting as an initial current level a current level at which a neural response is not expected to be evoked.

15. The method of claim 14, wherein selecting the initial current level comprises:

selecting an initial current level substantially below the typical T-NRT.

16. The method of claim 13, wherein selecting the initial current level comprises:

selecting an initial current level that is at least one of above and below the typical T-NRT.

17. The method of claim 12, wherein when the expert system predicts that the measurement does not contain a neural response, the method further comprises:

incrementing the current level of the neural stimulus;

repeating the applying, obtaining and evaluating steps; and

repeating the incrementing and repeating steps until a neural response has been evoked.

18. The method of claim 12, wherein when the expert system predicts that the measurement does not contain a neural response, the method further comprises:

decreasing the current level of the neural stimulus;

repeating the applying, obtaining and evaluating steps; and

repeating the decreasing and repeating steps until a neural response has been evoked.

19. An expert system configured to determine whether a Neural Response Telemetry (NRT) measurement contains a neural response or does not contain a neural response, wherein the NRT measurement having been made in response to a cochlear implant (CI) stimulus, the expert system comprising a machine learning algorithm using the induction of at least one decision tree having at least two levels of nodes, wherein the nodes of the decision tree consider, respectively:

a parameter relating to the ratio of the peak-to-peak amplitude of the measurement to the noise of the measurement; and

one or more of;

a parameter relating to the correlation of the NRT measurement to a predefined expected neural response;

a parameter relating to the correlation of the NRT measurement to a predefined expected trace containing neural response plus CI stimulus artifact;

a parameter relating to the correlation of the NRT measurement to a predefined expected trace containing CI stimulus artifact only; and

a parameter relating to the correlation of the NRT measurement to a previous measurement with CI stimulus of similar level.

20. A system for analyzing neural activity within a target neural region comprising:

means for applying electrical stimulation to the target neural region at an initial current level that approximates a typical threshold-Neural Response Telemetry (NRT) level;

means for recording an NRT measurement of neural activity within the target neural region in response to the stimulation; and

means for predicting whether the NRT measurement contains a neural response or does not contain a neural response, wherein the means for predicting include an expert system configured to consider one or more features of the NRT measurement including a signal to noise ratio thereof and to evaluating at least two successive levels of nodes of a decision tree in order to predict whether the NRT measurement does or does not contain a neural response, wherein at least one of the evaluated levels includes a node which considers a correlation coefficient relating the NRT measurement to a predetermined value.

21. The system of claim 20, further comprising:

means for selecting the initial current level to insure safety while minimizing a quantity of measurements required to determine the threshold NRT.

22. The system of claim 21, wherein the means for selecting the initial current level, in intra-operative applications, selects as an initial current level a current level at which a neural response is not expected to be evoked.

23. The system of claim 22, wherein the means for selecting the initial current level selects an initial current level substantially below the typical T-NRT.

24. The system of claim 21, wherein the means for selecting the initial current level, in intra-operative applications, selects an initial current level that is above or below the typical T-NRT.