AU764891B2 - Improved sound processor for cochlear implants - Google Patents

Description

WO 01/19135 PCT/AU00/01038 -1- IMPROVED SOUND PROCESSOR FOR COCHLEAR IMPLANTS Field of the Invention This invention relates to improvements in sound processors for cochlear implants, and more particularly to a Differential Rate Sound Processor (DRSP) Background of the Invention The multi-channel cochlear implant was first implanted in 1978. Early signal processing designs extracted the second formant (F2) and pitch (FO) to control electrode stimulation. The frequency of F2 controlled the location of electrode stimulation, and FO controlled the rate of stimulation. This was later improved by also extracting the first formant (F1) and adding a second stimulated electrode for each pitch period. The MULTI-PEAK (MPEAK) stimulation strategy added stimulation of a number of fixed electrodes to better represent high-frequency information. The next stages of development were the SMSP and SPEAK strategies. These were a departure from the others at they used a fixed stimulation rate and stimulated electrodes that corresponded to maxima in the sound spectra. Another fixed-rate strategy, CIS, was developed overseas. This strategy stimulated all of a small number of electrodes to represent the sound spectra. All of the above processing strategies involve fixed-rate sound processing.

The named inventors have determined that some speech features are better perceived using low-rates of simulation, while some are better perceived using high rates of stimulation. Higher rates of stimulation present more information about phonetic manner of articulation, but spectral information tends to be smeared at such higher rates.

Summary of the Invention and Object It is an object of the present invention to provide an improved sound processor for use with cochlear implants in which the problems associated with fixed rate stimulation are ameliorated.

The invention provides in one form a sound processor for a cochlear implant having apical and basal electrodes for stimulating the auditory nerve, including means for receiving sounds, means for processing the sounds and converting them to electrical stimulation signals for application to the electrodes of the cochlear implant for stimulation of the auditory nerve, said sound processing means including means for generating electrical signals to be applied to the basal electrodes at a first same predetermined rate of stimulation and for generating electrical signals to be applied to the apical electrodes at a second same predetermined rate of stimulation.

By causing stimulation of the basal electrodes at a higher rate of stimulation than the apical electrodes, the manner of articulation features of speech will be more optimally presented to the cochlear implant user, leading to improved speech understanding performance. High rates of stimulation at the basal electrodes will present good information about temporal events and frication. The low rates of stimulation of the apical electrodes will present good spectral information in this regard, where most place of articulation features reside.

In a preferred embodiment, the more apical electrodes will be chosen as those that contain the voice bar and lower formants of speech. In this frequency region, spectral detail is important and the apical electrodes will be stimulating 20 using a stimulation rate of between about 250 cycles per second and about 800 cycles per second, depending on the user. By adopting stimulation rates falling within the above range, better information about place of articulation of speech, which is largely represented by the formant structure, is obtained by the user.

The more basal electrodes represent higher frequency components of the 25 incoming sound, and higher rates of stimulation of these electrodes will be used to better represent noise and more precisely present information about temporal events such as rapid changes in amplitude. The latter is important for perception of manner of articulation and voicing. These electrodes will be stimulated at a higher rate than the apical electrodes, with stimulation rates at or above about 800 cycles per second, and preferably up to about 1600 cycles per second, being selected Melbourne\004314196 Printed 18 June 2003 (16:57) -3depending on the user.

In the case of an implant having 20 electrodes available for stimulation, the apical electrodes are electrodes 0 to 12, and the basal electrodes are electrodes 13 to 19. The apical electrodes represent sound frequencies from 0 to about 2700Hz, while the basal electrodes represent frequencies from about 2700Hz to about 7900Hz. The stated apical electrode frequencies are sufficient to contain the first three formants of most speech.

In a particularly preferred form of the invention, the apical electrodes are stimulated at about 250 cycles per second while the basal electrodes are stimulated at about 1500 cycles per second. To ensure that stimulation levels are suitable for these different rates, the threshold levels and comfort levels of the patient are carefully set. The electrodes to be stimulated are chosen by selecting the eight largest spectral energies within filterbanks derived from the Fast Fourier Transform (FFT) or the Discrete Wavelet Transform (DWT) which is performed by the processor.

In another form, the invention provides a sound processor for a cochlear implant having electrodes for stimulating the auditory nerve, including means for receiving sounds, means for processing the sounds and converting to electrical stimulation signals for application to the electrodes of the cochlear implant whereby the auditory nerve is electrically stimulated, said sound processing means having o* means for analysing the received sounds to determine when a higher rate of stimulation of the electrical stimulation signals is required and for selecting the higher stimulation rate from one or more predetermined stimulation rates.

0 Melbourne\004314196 Printed 18 June 2003 (16:57) WO 01/19135 PCT/AU00/01038 -4- By varying the rate of stimulation of the cochlear implant electrodes depending on the incoming speech signal, key speech features will be more optimally presented to the cochlear implant user thereby leading to improved speech understanding performance.

In a preferred form of this aspect of the invention, the sound processing means will be programmed to continually adjust the rate of stimulation of the electrical stimulation signals depending on the parameters of the incoming speech signal. To this end, the incoming speech signal will be processed to detect events that are better represented using a higher rate of stimulation. Such events include plosive onset bursts, frication and other rapid spectral changes.

The rate of stimulation across all electrodes will be increased for the average duration of these events. The standard rate will be between 250 cycles/s and 800 cycles/s depending on the user. The higher rate will be above about 800 cycles/s, and preferably up to about 1600 cycles/s, also depending on the user.

In order that the invention may be more readily understood, one presently preferred embodiment of the invention will now be described.

Description of Preferred Embodiment The invention is preferably designed for use with the CI-24M Cochlear Implant as manufactured by Cochlear Ltd, and as described in US Patent No.

4532930, the contents of which are incorporated herein by cross-reference, and later patents by Cochlear Ltd to be found in the patent literature.

Although the CI-24M Implant will be used in most cases, the invention could be applied to any implant that uses pulsatile stimulation. The stimulation strategy is based on the Spectral Maxima Sound Processor (SMSP), which is described in United States Patent 5597380 and Australian Patent 657959, the contents of which are incorporated herein by cross reference although other strategies may be used with similar results. For example, the SPEAK strategy as discussed in US Patent No. 5,597,380, the contents of which are incorporated by cross reference.

WO 01/19135 PCT/AU0O/01038 The electrode selection strategy from the SMSP is varied to ensure that electrodes are stimulated at the desired predetermined frequencies for each cycle of stimulation. The preferred signal processing device will be the SPEAR processor, which is currently under development at The Bionic Ear Institute, and which is described in the following paper: Zakis, J.A. and McDermott, H.J. (1999). "A new digital sound processor for hearing research," Proceedings of the Inaugural Conference of the Victorian Conference of the Victorian Chapter of the IEEE Engineering in Medicine and Biology Society, February 22-23, pp. 54-57. The contents of this paper are lo similarly incorporated herein by cross reference.

The processor is a generic processor based on the Motorola DSP56300 family, such as the DSP56302, or the DSP56309, although any digital signal processor, including those produced by Cochlear Ltd and their competitors, could be used to run the differential rate sound processor program of the present invention, provided they have adequate processing speed.

In the implementation of the first form of the invention, the differential rate stimulation processor software embodying the invention is downloaded to the SPEAR processor and stored on EPROM. Patient map details, including frequency bands, threshold levels and comfort levels, are also stored on the device. Monopolar stimulation mode is used to reduce current levels and for longer battery life.

For the case where 20 electrodes are available for stimulation, the apical electrodes are electrodes 0 to 12, and the basal electrodes are electrodes 13 to 19. The apical electrodes then represent frequencies from 0 to 2700Hz; the basal electrodes represent frequencies from 2700Hz to 7900Hz. The stated apical electrode frequencies are sufficient to contain the first three formants of most speakers' speech.

The apical electrodes are stimulated at about 250 cycles/s and the basal electrodes at about 1500 cycles/s. The patient's T and C levels are carefully set WO 01/19135 PCT/AUOO/01038 -6to ensure that stimulation levels are suitable for the two different rates and adjustments made if necessary. The electrodes to be stimulated are chosen by selecting the eight largest spectral energies within filterbanks derived from the Fast Fourier Transform (FFT) or the Discrete Wavelet Transform (DWT).

The values quoted above are examples. Patient-to-patient variability is large and some need higher stimulation rates on the apical electrodes and/or lower stimulation rates on the basal electrodes. These are determined for each individual by evaluating a number of rate combinations in every day usage.

Also, some patients do not have as many electrodes available and so the choice of electrodes is altered to suit their situation. However, the spectral ranges of the apical and basal electrodes remain much the same.

By using the Differential Rate Sound Processor (DRSP) program of the invention, features of speech will be more optimally presented to the cochlear implant user leading to improved speech understanding performance.

In the implementation of the second aspect of the invention, the software necessary to provide a variable rate of stimulation depending on the incoming speech signal is downloaded to the SPEAR processor and stored on an EPROM.

Patient map details, including frequency bands, threshold levels and comfort levels, are also stored on the device. Monopolar stimulation mode is used to reduce current levels and for longer battery life.

The standard rate of stimulation is about 250 cycles/s and the higher rate is about 1500 cycles/s. The patient's T and C levels are carefully set to ensure that stimulation levels are suitable for the two different rates. The electrodes to be stimulated are chosen by selecting the eight largest spectral energies within filterbanks derived from the Fast Fourier Transform (FFT) or the Discrete Wavelet Transform (DWT).

The changes in spectral energies and the amount of frequency energy are monitored over time. When there is a significantly large change between frames separated by the period of the lower stimulation rate then the higher stimulation WO 01/19135 PCT/AU00/01038 -7rate is used for 50 ms. This procedure locates plosive bursts and other rapid spectral changes. The higher stimulation rate is also used when the ratio of energy below about 300Hz to that above about 2000Hz is less than about This locates phonemes with significant frication.

The values quoted above are examples. Patient-to-patient variability is large and some need a higher stimulation rate for the standard rate and/or a lower stimulation rate for the higher rate. These are determined for each individual by evaluating a number of rate combinations in every day usage.

Thresholds for changes in energy and ratio of energies are also adjustable for each individual.

Claims (12)

1. A sound processor for a cochlear implant having apical and basal electrodes for stimulating the auditory nerve, including means for receiving sounds, means for processing the sounds and converting them to electrical stimulation signals for application to the electrodes of the cochlear implant for stimulation of the auditory nerve, said sound processing means including means for generating electrical signals to be applied to the basal electrodes at a first same predetermined rate of stimulation and for generating electrical signals to be applied to the apical electrodes at a second same predetermined rate of stimulation.

2. The sound processor of claim 1, wherein the more apical electrodes are selected for stimulator signals that represents the voice bar and lower formants of the sounds.

3. The sound processor of claim 2, where the more apical electrodes apply stimulation signals having a stimulation rate of between about 250 cycles per second and about 800 cycles per second depending on the user, to provide precise spectral and place of articulation information. 20

4. The sound processor of any one of claims 1 to 3, wherein the more basal electrodes apply stimulation signals having a stimulation rate of at or above about 800 cycles per second depending on the user, to provide precise information about temporal events and frication.

5. The sound processor of any one of claims 1 to 4, including a twenty electrode implant, the apical electrodes are electrodes 0 to 12 and the basal electrodes are electrodes 13 to 19, the apical electrodes representing sound frequencies from 0 to about 2700Hz, while the basal electrodes represent frequencies from about 2700Hz to about 7900Hz, the apical electrode frequencies, are sufficient to contain the first three formants of most speakers speech. are sufficient to contain the first three formants of most speakers speech. Melbourne\004314196 Printed 18 June 2003 (16:57) -9-

6. The sound processor of any one of claims 1 to 5, wherein the apical electrodes are stimulated at about 250 cycles per second while the basal electrodes are stimulated at about 1500 cycles per second.

7. The sound processor of claim 6, wherein the threshold levels and comfort levels of the patient are carefully set, the electrodes to be stimulated are chosen by selecting the eight largest spectral energies within filterbanks derived from the Fast Fourier Transform (FFT) or the Discrete Wavelet Transform (DWT) which is performed by the processor.

8. A sound processor for a cochlear implant having electrodes for stimulating the auditory nerve, including means for receiving sounds, means for processing the sounds and converting to electrical stimulation signals for application to the electrodes of the cochlear implant whereby the auditory nerve is electrically stimulated, said sound processing means having means for analysing the received sounds to determine when a higher rate of stimulation of the electrical stimulation signals is required and for selecting the higher stimulation rate from one or more i predetermined stimulation rates.

9. The sound processor of claim 8, wherein the sound processing means is programmed to continually adjust the rate of stimulation of the electrical oO.o stimulation signals depending on the parameters of the incoming speech signal, to this end, the incoming speech signal will be processed to detect events that are better represented using a higher rate of stimulation.

10. The sound processor of claim 8 or 9, as defined in any one of claims 1 to 7.

11. The sound processor of any preceding claim, wherein the implant is one which uses pulsatile stimulation. Melbourne\004314196 Printed 18 June 2003 (16:57)

12. The sound processor of any preceding claims wherein a SPEAR processing device is programmed using the Differential Rate Sound Processor (DRSP) program defined in any preceding claim to optimally present the features of speech to the implant. DATED this 18 June 2003 FREEHILLS CARTER SMITH BEADLE Patent Attorneys for the Applicant THE BIONIC EAR INSTITUTE Melbourne\004314196 Printed 18 June 2003 (16:57)