78 | Chapter 4
are not completely matching the analysis with mixed linear models, which take missing data points into
account.
For analysis of subsets of data, three additional linear mixed models were constructed with only data of the
fi ve subjects tested in the subset. Besides the comparison of scanning versus selec-tivity measures, these
subsets included the analysis of selectivity measures alone as well as the measurements with the recording
contact at apical or basal position, and with the masker or probe fi xed. Additionally, a subset with data on
artefact rejection method in scanning was analysed. Furthermore, a mixed linear model was constructed in
which the patients ’ speech perception was compared with their selectivity widths measured with the apical
recording position. Using a mixed linear model made it possible to take the current level and location
along the array into account in the analysis with speech data.
RESULTS
Comparison of scanning vs. selectivity
Figure 3 shows boxplots for scanning (upper panels) and selectivity (lower panels) measures at the apical
(EA-B), middle in apical and basal directions (EM-A and EM-B respectively), and basal (EB-A) electrodes,
measured at three different current levels (low, medium, and high).
The majority of curves (95% for selectivity and 75% for scanning) met the 0.6 criterion. However, for the
curves of the middle electrode contact in the apical direction this criterion was often not met (42% for
Parameter Manipulations in SOE (eCAP)
469
at the three current levels. A horizontal line indicates 60% of the
amplitude. The horizontal solid arrows then indicate the width of the
curve (in basal or apical direction) as defined above for the highest
current level.
Statistics
The design of this study is basically a within-patient analysis with
three fact rs, which mean that at the p tient lev l the measurements
are correlated. So-called linear mixed models take this correlation
into account, by considering the responses from a subject to be the
sum of so-called fixed effects, affecting the population ean, and
random effects, associated with a sampling procedure (e.g. subject
effects). The random effects often introduce correlations between
cases and should be taken into account to elucidate the fixed, popu-
lation affecting, effects. The SE (standard error of the mean) gen-
erated by the model is used in significance analysis. Using linear
mixed models enables investigation of the effects of each parameter
separately as well as the interactio between different parameters.
Furthermore, linear mixed models can effectively use all data, even
when one or more data points are missing (Fitzmaurice et al, 2004).
In the present study SPSS 16.0 was used to construct mixed lin-
ear models to address the influence of the measuring technique, the
electrode position, and the current level separately. For significance
levels in this study t-tests are used, both in the context of descriptive
statistics as well as with linear mixed models.
To create a comprehensive overview data are plotted in boxplots.
However, it should be noted that boxplots are not completely match-
ing the analysis with mixed linear models, which take missing data
points into account.
For analysis of subsets of data, three additional linear mixed
models were constructed with only data of the five subjects tested
in the subset. Besides the comparison of scanning versus selec-
tivity measures, these subsets in luded the analysis of selectivity
measures alone as well as the measurements with the recording
contact at apical or basal position, and with the masker or probe
fixed. Additionally, a subset with data on artefact rejection method
in scanning was analysed. Furthermore, a mixed linear model was
constructed in which the patients’ speech perception was compared
with their selectivity widths measured with the apical recording
position. Using a mixed linear model made it possible to take
the current level and location along the array into account in the
analysis with speech data.
Results
Comparison of scanning vs. selectivity
Figure 3 shows boxplots for scanning (upper panels) and selectivity
(lower panels) measures at the apical (EA-B), middle in apical and
basal directions (EM-A and EM-B respectively), and basal (EB-A)
electrodes, measured at three different current levels (low, medium,
and high).
The majority of curves (95% for selectivity and 75% for scanning)
met the 0.6 criterion. However, for the curves of the middle electrode
c ntact in the apical irecti this criterion was often not met (42%
for scanning and 14% for selectivity). The curves that did not meet
the 0.6 criterion were set as the limit of the array in the apical or
basal direction (as per Abbas et al, 2004).
Descriptive statistics, showing the means and differences of
scanning and selectivity along th array, are summarized i Table 2.
The calculated linear mixed model, containing data of 31 dif-
ferent subjects (for 26 subjects, measures of both scanning and
selectivity were obtained, for three subjects only scanning data,
Figure 2.
Selectivity curves obtained in one subject, in response to stimulation at the apical electrode contact (EA, left), middle contact
(EM, middle) and basal contact (EB, right). The figures show the normalized response amplitudes recorded from locations along the array,
using low (dotted line), medium (dashed line), and high (solid line) current levels, together with the 0.6 criterion line. The width of the curve
is then defined by the number of electrodes above the 0.6 line (indicated by arrows) for apical electrode in basal direction (EA-B), middle
electrode in apical and basal direction (EM-A, EM-B), and basal electrode in apical direction (EB-A). The location of electrode of interest
(dotted line over full figure height) and position where curve meets the 0.6 criterion (dotted line over half figure height) are indicated.
Table 2.
Descriptive statistics including mean widths in terms of
electrode contacts for scanning (top rows) and selectivity (lower
rows) for the different electrode contact locations (with standard
deviations). Significance levels are shown in the bottom row. Data
shown are descriptive statistics and are incorporated in the linear
mixed model for further analysis of separate parameters, which are
described in the Results section.
EA-B
EM-A
EM-B
EB-A
Scanning
Mean
8.2
6.8
5.9
9.0
Std. Dev.
4.0
1.1
2.4
3.6
Selectivity
Mean
5.3
6.3
3.3
4.2
Std. Dev.
3.5
2.5
1.8
2.8
Significance
p
0.01
p
0.1
p
0.01
p
0.01
Figure 2.
Selectivity curves obtained in one subject, in response to stimulation at the apical electrode contact (EA, left), middle contact
(EM, middle) and basal contact (EB, right). The figures show the normalized response amplitudes recorded from locations along the array,
using low (dotted line), medium (dashed line), and high (solid line) current levels, together with the 0.6 criterion line. The width of the
curve is then defined by the number of electrodes above the 0.6 line (indicated by arrows) for apical electrode in basal direction (EA-B),
middle electrode in apical and basal direction (EM-A, EM-B), and basal electrode in apical direction (EB-A). The location of electrode
of interest (dotted line over full figure height) and position where curve meets the 0.6 criterion (dotted line over half figure height) are
indicated.