Page 58 - Spring2019

The Remarkable Cochlear Implant
distance combine to produce broad spreads of the excitation fields from each electrode along the length of the cochlea (length constant of about 10 mm or greater compared with the ~35-mm length of the human cochlea). Also, the fields from each electrode overlap strongly with the fields from other electrodes. The aspect of processing is to present the pulses across channels and their associated electrodes in a sequence rather than simultaneously. The nonsimultaneous or “interleaved” stimulation eliminates direct summation of electric fields from the different electrodes that otherwise would sharply degrade the perceptual independence of the channels and electrodes. CIS gets its name from the continu- ous (and fixed rate) sampling of the mapped envelope signals by interleaved pulses across the channels.
The overall approach is to utilize the perceptual space fully and to present the information in ways that will preserve the independence of the channels and minimize perceptual dis- tortions as much as possible. Of course, in retrospect, this approach also allowed the brain to work its magic. Once we designers “got out of the way” in presenting a relatively clear and unfettered signal rather than doing anything more or more complicated, the brain could take over and do the rest.
Some of the first results from comparisons of CIS with the best strategy in clinical use at the time are presented in Figure 2. Results from four tests are shown and range in difficulty from easy to extremely difficult for speech presented in otherwise quiet conditions. Each subject had had at least one year of dai- ly experience with their clinical device and processing strategy, the IneraidTM CI and the “compressed analog” (CA) strategy, re- spectively, but no more than several hours of experience with CIS before the tests. (The CA strategy presented compressed analog signals simultaneously to each of four intracochlear electrodes and is described further in Wilson, 2015.) The green lines in Figure 2 show the results for a first set of sub- jects selected for high performance with the CA strategy (data from Wilson et al., 1991), which was fully representative of the best performances that had been obtained with CIs as of the time of testing. The blue lines in Figure 2 show the results for a second set of subjects who were selected for their more typi- cal levels of performance (data from Wilson et al., 1992). The scores for all tests and subjects demonstrated an immediate and highly significant improvement with CIS compared with the alternative strategy.
Not surprisingly, the subjects were thrilled along with us by this outcome. One of the subjects said, for example, “Now you’ve got it!” and another slapped the table in front of him
and said, “Hot damn, I want to take this one home with me!” All three major manufacturers of CIs (which had more than 99% of the market share) implemented CIS in new versions of their products in record times for medical devices after the re- sults from the first set of subjects were published (Wilson et al., 1991), and CIS became available for widespread clinical use within just a few years thereafter. Thus, the subjects got their wish and the CI users who followed them benefitted as well.
Many other strategies were developed after CIS, but most were based on it (Fayad et al., 2008; Zeng and Canlon, 2015; Zeng, 2017). CIS is still used today and remains as the prin- cipal “gold standard” against which newer and potentially beneficial strategies are compared. Much more information about CIS and the strategies that followed it is presented in recent reviews (Wilson and Dorman, 2008, 2012; Zeng et al., 2008). Additionally, most of the prior strategies are described in Tyler et al. (1989) and Wilson (2004, 2015).
Performance of Unilateral
Cochlear Implants
The performance for speech reception in otherwise quiet conditions is seen in Figure 3, which shows results from two large studies conducted approximately 15 years apart. In Figure 3, the blue circles and lines show the results from a study conducted by Helms et al. (1997) in the mid-1990s and the green circles and lines show the results from tests with patients who were implanted from 2011 to mid-2014 (data courtesy of René Gifford at the Vanderbilt University Medi- cal Center [VUMC]). For both studies, the subjects were postlingually (after the acquisition of language in childhood with normal or nearly normal hearing) deafened adults, and the tests included recognition of sentences and monosyllabic words. The words were comparable in difficulty between the studies, but the low-context Arizona Biomedical (AzBio) sentences used in the VUMC study were more difficult than the high-context Hochmair-Schultz-Moser (HSM) sentences used in the Helms et al. (1997) study. Measures were made at the indicated times after the initial fitting of the device, and the means and standard error of the means (SEMs) of the scores are shown in Figure 3. Details about the subjects and tests are presented in Wilson et al. (2016).
The results demonstrate (1) high levels of speech reception for high-context sentences; (2) lower levels for low-context sentences; (3) improvements in the scores for all tests with increasing time out to 3-12 months depending on the test; (4) a complete overlapping of scores at every common test
56 | Acoustics Today | Spring 2019