Proefschrift_vd_Beek

F. B. van der Beek et al.

470

Current levels Before normalization, the eCAP amplitude obtained at different current levels varied consistently by a factor of up to two between the low and the high current level in almost all patients. Informal inspection of the data in Figure 3 suggests that while there was a large spread in the widths obtained in different subjects, the majority of individual subjects produced curves of similar width with different stimulus levels. There were, however, some notable exceptions, some subjects showing large changes in curve width with changing stimulus level. 28 subjects in order to examine the effect of stimulating electrode contact position on this SOE measure. This model confirmed the differences in curve width along the array for the selectivity measure as shown in Table 2. The curves of EM-A were the widest, followed by the EA-B, EB-A, and EM-B. Taking EA-B as a reference, EM-A was 1.0 contact spacing wider (SE 0.38, p 0.01), EM-B was 2.0 contacts narrower (SE 0.38, p 0.01), and EB-A was 1.2 con- tact narrower (SE 0.41, p 0.01). These ffec s are illu trated in Figure 2, which shows a typical example in an individual patient. Levels of significance for the differences found are shown in Table 3. Figure 3. Boxplots of SOE measurements. The top row shows the scanning measurements at the different electrode positions EA-B (apical), EM-A (middle in apical direction), EM-B (middle in basal direction), and EB-A (basal) with the widths measured at three current levels, low, medium, and high. The lower figure shows the corresponding selectivity measures. The boxes represent the interquartile range (IQR, i.e. 25%–75%), with the median indicated by a horizontal line. The ends of the whiskers represent the lowest data point still within 1.5 IQR of the lower quartile, and the highest data point still within 1.5 IQR of the upper quartile. Outliers (indicated with a circle) are cases with values between 1.5 and 3 times the IQR, and extremes (indicated with an asterix) are cases deviating more than 3 times the IQR. and for two subjects only selectivity measurements), confirmed differences in the width of curves obtained using the two meth- ods. Scanning produced significantly wider curves than selec- tivity (mean 7.4 contact spacing (SE 0.26) vs. 4.7 spacing (SE 0.27), p 0.01). An example of scanning and selectivity curves in a typical subject is shown in Figure 4, A. The influence of the method was largest for basal and apical contacts and less prominent for the middle contact in the apical direct on (Table 2). No factor associated with the outliers as seen in Figure 3 could be identified. Figure 3. Boxplots of SOE measurements. The top row shows the sca ning measurement at the different electrode positions EA-B (apical), EM-A (middle in apical direction), EM-B (middle in basal direction), and EB-A (basal) with the widths measured at three current levels, low, medium, and high. The lower figure shows the corresponding selectivity measures. The boxes represent the interquartile range (IQR, i.e. 25%–75%), with the median indicated by a horizontal line. The ends of the whiskers represent the lowest data point still within 1.5 IQR of the lower quartile, and the highest data point still within 1.5 IQR of the upper quartile. Outliers (indicated with a circle) are cases with values between 1.5 and 3 times the IQR, and extremes (indicated with an asterix) are cases deviating more than 3 times the IQR.

Comparison along the array A second linear mixed model was constructed that included only the selectivity measurements (i.e. excluding the scanning data) of

Current levels Before normalization, the eCAP amplitude obtained at different current levels varied consistently by a factor of up to two between

80 | Chapter 4