Fitting based on objective measures
Since this method was developed for Nucleus implants the terminology in this section is Nucleus® focused.
The idea that there exist objective measures that correlate well enough with psychophysical data to allow a fitting based on such objective measures is an attractive one. The most obvious candidate for this is Neural Response Telemetry (NRT™). NRT is a method whereby the Electrically Evoked Compound Action Potential (ECAP) is measured using only a Nucleus implant, external speech processor and a PC. One might suppose that the ECAP threshold (lowest stimulation level that elicits a visible ECAP) has some relation to the map T- and/or C-levels. Similar measures are possible with other manufacturer’s devices.
Unfortunately, research shows that the direct correlation between T and C levels and ECAP threshold is not good enough to warrant a map based directly on ECAP thresholds. For example, Brown et al. (2000) found only a moderate correlation between T (r=0.55) and C (r=0.57) levels and ECAP thresholds, other authors found similar numbers (see Cafarelli Dees et al., 2005 for recent data). Also with different objective measures (E-ABR, stapedius reflex, etc) only weak correlation are generally found (see, for example, Brown et al., 2000).
There are two important reasons for this weak correlation:
1) The ECAP is a purely peripheral measure and does not take into account central loudness effects (summation, masking, etc).
2) The C-levels set in a map are highly subjective and differ between subjects and even between audiologists making the measurements. A variance analysis of a large dataset of clinical data showed that apart from subject and electrode, the CLINICIAN is the only significant factor determining C-level (van Dijk, unpublished data). This means that it may not even be useful to correlate objective data to C-level data, and one must focus purely on performance results.
Recent data indicate that higher order evoked potentials may have a better correlation with psychophysical methods.
The above discussion shows that it is very difficult to determine what the best fitting method is, certainly the best fitting method is not necessarily the one that gives the best prediction of classical T and C levels, since it is not clear at all if we can take this as a golden standard, besides, T and C levels contain a high level of arbitrarily and subjectivity.
The best evaluation of a new fitting method remains the clinical performance in terms of speech understanding and client satisfaction; this means that fitting studies take time and require care.
The following items are related