Surgery – Diagnostic testing – Ear or testing by auditory stimulus
Reexamination Certificate
2000-06-06
2002-06-18
Shaver, Kevin (Department: 3736)
Surgery
Diagnostic testing
Ear or testing by auditory stimulus
Reexamination Certificate
active
06406438
ABSTRACT:
This invention relates to a method of and apparatus for obtaining evoked otoacoustic emission response data by deconvolving stimulus and response signals in the form of maximum length sequences (MLS). References herein to MLS are intended to include similar sequences or variants of similar sequences.
Much previous audiological research has been forced, through technological and intellectual limitations, to make the assumption that the cochlea is time-invariant and linear. There is now much evidence that this is not so. There is also an increasing awareness that non-linearities are the key to important aspects of normal auditory function and that some of the non-linear components may reflect auditory dysfunction more sensitively than the linear components previously studied.
The applicants' published specification WO 94/25925 discloses deconvolution of MLS response data to detect otoacoustic emissions. Such a known technique is applicable to situations in which, for both research and clinical practice, it is an advantage to obtain the emission very quickly. If one tries to speed up the process by increasing click presentation rates, the responses would overlap both each other and the stimulus clicks for rates greater than about 50 clicks/s. It would be impossible to recover the normal evoked emission from these overlapped recordings. However, if a particular sequence of clicks and silences, known as a maximum length sequence, is presented, then the overlapped responses may be deconvolved to give the original response that would have been obtained from conventional averaging.
Hence, previous work has shown that it is possible to obtain an otoacoustic emission response by deconvolving an MLS stimulus. As the stimulus rate is increased, non-linear temporal interaction components (NLTICs) are produced because of the non-linear nature of the auditory system, and the invention results from work carried out to identify and record NLTICs.
According to one aspect of the invention there is provided a method of obtaining evoked otoacoustic emission (EOAE) response data by deconvolving stimulus and response signals in the form of a maximum length sequence (MLS), comprising detecting non-linear temporal interaction components (NLTICs) in the response data.
Preferably, the response data is recorded in real time.
For both evoked otoacoustic emissions (EOAEs) and evoked potentials (EPs), three types of non-linearity can be distinguished:
(a) the non-linear growth of response amplitude with increasing stimulus level;
(b) the frequency domain non-linear distortion products that occur when two tones are used as the stimulus and which may be seen as combinations of sum and difference tones; and
(c) non-linear temporal interaction, which creates an increase in the non-linear activity as the time between stimuli decreases.
Types (a) and (b) have been described in the prior art. In simple systems without memory, (b) can be derived from (a). However, the cochlea is a non-linear system with memory and incorporates interactions between different frequency elements. Such a system can have non-linearities of types (a), (b) and (c), none of which can be well predicted from the others. For example, the type (c) non-linearities arise essentially out of the interactions between responses to stimuli juxtaposed in time, irrespective of the non-linear input/output function of the EOAE.
A linear system can be characterised by its impulse response and, for any input to the system, the output may be calculated by convolving the input with the system's impulse response. However, more information is needed to define a non-linear system. If the input to a non-linear system is an MLS, then the output may be expressed as a Volterra series in which the elements are known as “kernels”. The first order kernel represents the convolution of the impulse response with the input; ie, it defines the output in exactly the same way as it is defined in a linear system and it is sometimes referred to as the “linear component” of the series. The second order kernel is 3-dimensional (amplitude×time for click
1
×time for click
2
) and represents the convolution of all possible non-linear interactions created by pairs of stimuli. The third order kernel is 4-dimensional (amplitude×time for click
1
×time for click
2
×time for click
3
) and represents the convolution of all possible non-linear interactions caused by triples of stimuli.
Because the MLS is a discrete, digital input, the kernels are defined only at intervals which are a multiple of the minimum inter-stimulus interval of the MLS. Each of these segments of the kernel is known as a slice. When the MLS is deconvolved, the slices from the various kernels are scattered along the entire deconvolved MLS waveform. The slices have to be extracted and combined to estimate the kernels. In this way the temporal non-linearities of the system can be characterised by a Volterra series and the individual elements described by the Volterra kernel slices. Previously, this type of approach had been carried out only for auditory brainstem responses (ABRs), where it appeared that two or three major components were sufficient to adequately represent the non-linear behaviour of the system.
Thus, in accordance with a further feature of the invention, the non-linear temporal interaction components may be recorded as slices through higher order Volterra kernels, with each Volterra kernel representing a term in a Volterra series which models the stimulus/response system. The Volterra series gives information on the linear components and also the second and higher order non-linear components, and it is the second and higher order components with which the invention is principally concerned.
The location of the NLTIC slices is very hard to predict. For any particular MLS they may occur anywhere within that MLS, even overlapping each other and the linear component at the start of the MLS.
In order accurately to record the linear and the NLTICs for each order of MLS, all possible MLSs had to be generated and the location of the NLTICs computed. The “optimum” MLS is the one that has the minimum overlap between these NLTICs with each other and with the linear component at the start. Following this computation the optimum MLSs were defined for each order of MLS.
Special software was written to implement these MLSs and to record the otoacoustic emissions. It was not known whether otoacoustic emissions would show these NLTICs but, in the event, they did.
The identification of these components and the knowledge of any overlap is possible only because of the extensive computation carried out on all possible combinations of MLS.
The computation of the position of each slice in the recovered MLS may be computed so that each slice can be properly identified.
According to another aspect of the invention there is provided apparatus for obtaining evoked otoacoustic emission (EOAE) response data by deconvolving stimulus and response signals in the form of a maximum length sequence (MLS), comprising means for detecting non-linear temporal interaction components (NLTICs) in the response data.
Mathematically an MLS is a pseudo-random sequence of −1s and +1s. Its auto-correlation function is 1 for 0 lag and otherwise is −1/L, which is also its average value.
The particular variant of MLS that has been developed for use in evoked response recording is obtained by replacing the −1s with +1s and the +1s with 0s. The +1 represents a click stimulus and the 0 represents silence. In practice, evoked response recordings require the rejection of noisy epochs so that they do not add to the average and worsen the signal-to-noise ratio. Now, with MLS stimulation for OAEs, there are stimuli occurring at intervals as small as 200 &mgr;s. Thus it is impossible to apply the normal rejection template to detect noisy epochs. It was therefore considered important to develop a technique in which the deconvolution of the MLS was carried out “on-the-fly” as the ML
Lee Mann Smith McWilliams Sweeney & Ohlson
Medical Research Counsel
Shaver Kevin
Wingood Pamela P.
LandOfFree
Method and apparatus for obtaining evoked otoacoustic emissions does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for obtaining evoked otoacoustic emissions, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for obtaining evoked otoacoustic emissions will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2943451