System, method and apparatus for detecting an evoked response signal

1. A method for detection of an evoked response signal in noise, the method comprising:

generating a plurality of stimuli;

receiving a noisy signal related to an evoked response to the plurality of stimuli;

dividing the noisy signal into a plurality of responses to the plurality of stimuli;

calculating weights for the plurality of responses;

identifying sets of responses, and for each set:

combining the responses as a revised response;

calculating a new weight for the revised response; and

removing the responses from the plurality of responses;

constructing a final response by weight averaging the revised responses and responses not identified in a set, the final response representing the evoked response or indicative of whether or not an evoked response has been detected; and

outputting the final response.

2. The method of claim 1 , further comprising:

decomposing each response into a plurality of sub-responses to create a plurality of sets of sub-responses, wherein each sub-response contains a particular subset of information from its corresponding response;

performing the steps of calculating, identifying, combining, calculating, removing and constructing for each of the plurality of sets of sub-responses; and

combining the final response of each set of sub-responses.

3. The method of claim 2 , wherein each set of sub-responses corresponds to a particular frequency band.

4. The method of claim 3 , wherein the particular frequency band is determined using wavelet decomposition.

5. The method of claim 1 , wherein the step of dividing the noisy signal into a plurality of responses to the plurality of stimuli is based on the plurality of responses being synchronized with the plurality of stimuli.

6. The method of claim 1 , wherein identifying sets of correlated responses comprises:

estimating a statistic matrix for the plurality of responses;

determining sets of correlated responses whose combination will result in noise reduction; and

shrinking the statistic matrix at least by removing or combining sets of correlated responses whose combination will result in noise reduction.

7. The method of claim 6 , wherein shrinking the statistic matrix comprises:

creating a list of negatively correlated combinations of responses;

for each negatively correlated combination in the list:

determining if one of the responses in the combination is in a shrinkage list, and if so, removing the combination from the list of negatively correlated combinations, otherwise, add the responses of the combination to the shrinkage list;

when the list of negatively correlated combinations is empty, creating a mask for the statistic matrix;

applying the mask to the statistic matrix to provide a shrunk statistic matrix; and

returning the shrunk statistic matrix.

8. The method of claim 7 , wherein one of:

the statistic is covariance and the statistic matrix is a covariance matrix of two dimensions; and

the statistic is root mean squared and the statistic matrix is a root mean squared matrix of two or more dimensions.

9. The method of claim 7 , wherein the mask is created by:

creating a diagonal matrix of the same dimension as the statistic matrix; and

setting all non-diagonal elements of the statistic matrix corresponding to responses not in the shrinkage list to 0.

10. The method of claim 6 , wherein the new weights are calculated based on an inverse of the shrunk statistic matrix.

11. The method of claim 6 , wherein the sets of correlated responses are pair combinations of responses.

12. A non-transitory computer readable medium having stored thereon computer program code for execution by one or more processors to perform the method of claim 1 .

13. An apparatus for detection of an evoked response signal in noise, the apparatus comprising:

an input device configured to receive data related to a plurality of stimuli and a noisy signal related to the evoked response signal to the plurality of stimuli; and

a processor configured to:

receive the noisy signal from the input device and divide the noisy signal into a plurality of responses to the plurality of stimuli;

calculate weights for the plurality of responses;

identify set of responses, and for each set:

combine the responses as a revised response;

calculate a new weight for the revised response; and

remove the responses from the plurality of responses;

construct a final response by weight averaging the revised responses and responses not identified in a set, the final response representing the evoked response or indicative of whether or not an evoked response has been detected; and

outputting the final response.

14. The apparatus of claim 13 , wherein the processor is further configured to decompose each response into a plurality of sub-responses to create a plurality of sets of sub-responses, wherein each sub-response contains a particular subset of information from its corresponding response;

performing the steps of calculating, identifying, combining, calculating, removing and constructing for each of the plurality of sets of sub-responses; and

combining the final response of each set of sub-responses.

15. The apparatus of claim 13 , wherein identifying sets of correlated responses comprises:

estimating a statistic matrix for the plurality of responses;

determining sets of correlated responses whose combination will result in noise reduction; and

shrinking the statistic matrix at least by removing or combining sets of correlated responses whose combination will result in noise reduction.

16. The apparatus for detection of claim 15 , wherein when shrinking the statistic matrix the processor is further configured to:

create a list of negatively correlated combinations of responses;

for each negatively correlated combination in the list:

determine if one of the responses in the combination is in a shrinkage list, and if so, remove the combination from the list of negatively correlated combinations, otherwise, add the responses of the combination to the shrinkage list;

when the list of negatively correlated pairs combinations is empty, create a mask for the statistic matrix;

apply the mask to the statistic matrix to provide a shrunk statistic matrix; and

return the shrunk statistic matrix.

17. The apparatus of claim 16 , wherein the processor is configured to create the mask by:

creating a diagonal matrix of the same dimension as the statistic matrix; and

setting all non-diagonal elements of the statistic matrix corresponding to responses not in the shrinkage list to 0.

18. The apparatus of claim 15 , wherein the new weights are calculated based on an inverse of the shrunk statistic matrix.

19. The apparatus of claim 15 , wherein the sets of correlated responses are pair combinations of responses.

20. A system for detection of an evoked response signal in noise, the system comprising:

a stimulus generator configured to generate a plurality of stimuli;

a plurality of sensors configured to receive a noisy signal including an evoked response signal to the plurality of stimuli;

an input device configured to receive data related to the plurality of stimuli and the noisy signal;

a processor configured to:

receive the noisy signal from the input device and divide the noisy signal into a plurality of responses to the plurality of stimuli;

calculate weights for the plurality of responses;

identify set of responses, and for each set:

combine the responses as a revised response;

calculate a new weight for the revised response; and

remove the responses from the plurality of responses;

construct a final response by weight averaging the revised responses and responses not identified in a set, the final response representing the evoked response or indicative of whether or not an evoked response has been detected; and

an output device to output the final response received from the processor.

21. The system of claim 20 , wherein the processor is further configured to decompose each response into a plurality of sub-responses to create a plurality of sets of sub-responses, wherein each sub-response contains a particular subset of information from its corresponding response;

performing the steps of calculating, identifying, combining, calculating, removing and constructing for each of the plurality of sets of sub-responses; and

combining the final response of each set of sub-responses.

22. The system of claim 20 , wherein identifying sets of correlated responses comprises:

estimating a statistic matrix for the plurality of responses;

determining sets of correlated responses whose combination will result in noise reduction; and

shrinking the statistic matrix at least by removing or combining sets of correlated responses whose combination will result in noise reduction.

23. The system of claim 22 , wherein when shrinking the statistic matrix the processor is further configured to:

create a list of negatively correlated combinations of responses;

for each negatively correlated combination in the list:

determine if one of the responses in the combination is in a shrinkage list, and if so, remove the combination from the list of negatively correlated combinations, otherwise, add the responses of the combination to the shrinkage list;

when the list of negatively correlated combinations is empty, create a mask for the statistic matrix;

apply the mask to the statistic matrix to provide a shrunk statistic matrix; and

return the shrunk statistic matrix.

24. The system of claim 23 , wherein the processor is configured to create the mask by:

creating a diagonal matrix of the same dimension as the statistic matrix; and

setting all non-diagonal elements of the statistic matrix corresponding to responses not in the shrinkage list to 0.

25. The system of claim 22 , wherein the new weights are calculated based on an inverse of the shrunk statistic matrix.

26. The system of claim 22 , wherein the sets of correlated responses are pair combinations of responses.