96 / 162
94 | Chapter 5
[Allum et al., 2002; Gordon et al., 2004; Caner et al., 2007]. Currently, in most clinics, eCAPs replaced
eABR measurements for practical reasons. Despite all efforts, however, automated fitting based on objective
measures has not replaced the traditional behavioral method in daily practice.
Since automated prediction of levels cannot be obtained and objective measures can only provide guidance
for fitting, behavioral information is used. To speed up the fitting procedure, the amount of behavioral
information is routinely reduced. For instance, the commonly used monopolar stimulation mode shows
less across-site variation than bipolar stimulation, providing relatively flat profiles along the array, making
interpolation feasible [Pfingst et al., 2004]. For fitting, M-levels can be obtained on some electrodes, and
the levels of the intermediate electrodes are based on interpolation [Plant et al., 2005] or on the aspect
of the live-voice stimuli [Smoorenburg, 2007]. Although generally yielding a significantly lower speech
perception, flat M-profiles appear to be useful, especially in children or other recipients who are not able to
provide reliable behavioral feedback [Boyd, 2010].
Also for the T-levels, behavioral levels can be applied and interpolation used for time saving. Alternatively,
the T-levels are sometimes set at 10% of M-levels (in fact, it is the default in the SoundWave fitting suite for
the CII/ HiRes 90K implant) or even at 0
μ
A. This minimization of T-levels does not create a decrement in
speech understanding [Spahr and Dorman, 2005; Boyd, 2006], although T-levels can be of importance in
more challenging listening circumstances as in soft speech [Holden et al., 2011].
Govaerts et al. [2010] recently proposed an automated fitting procedure, based on clinical level data,
further adjusting those levels using psychoacoustic test results. In this approach, fitting is not solely based
on comfort, as is common in clinical practice, but rather is outcome driven. Although this would be
interesting, the authors did not yet publish the statistical data concerning their population levels, nor the
correlation between psychoacoustic
test result
s (e.g., pure-tone and speech audiometry, loudness scaling)
and fitting levels. The idea of an outcomedriven fitting is consistent with the fitting procedure used in our
clinic, where, during fitting, emphasis is given to the higher frequencies by introducing a slightly upsloping
M-level profile towards the basal electrodes [Briaire, 2008]. This approach was based on experience with
hearing aids, where increases in high-frequency information led to improved speech understanding in noise
[Versfeld et al., 1999].
Despite the enormous research effort applied to obtain simple fitting procedures and the large amount of
time spent by audiologists in programming numerous cochlear implant recipients, no large data sets of
recipient levels are published with the intent to offer normative data. However, Wesarg et al. [2010] and
Smoorenburg [2007] analyzed large data sets of Tand M-levels of Nucleus implant recipients to investigate
parameters that determine those levels. Tand M-levels are shown to vary considerably, but the dynamic
range (DR) was, on average, 50 current levels (SD 20) in Nucleus 22 (bipolar stimulation) [Bento et al.,
2005] and Nucleus 24 cochlear implant users (monopolar stimulation) [Wesarg et al., 2010]. This means
that the thresholds were about 9 dB lower than the M-levels, i.e., the T-levels were on average at 35% of the