A comparison of two loudness balancing tasks in cochlear implant subjects using bipolar stimulation.

In this study, the accuracy of independent measurement of the loudness of different electrodes in a cochlear implant (the "reference" method) was compared with the accuracy of measurements that depend on the results of previous measurements (the "adjacent" method) by evaluating the similarity between and the slopes of the loudness balance curves, and the variability in the measured loudness balance values.The two methods of loudness balancing differed only in the reference electrode used. In the adjacent method, the loudness of the test electrode was sequentially adjusted to match the loudness of an adjacent reference electrode, whereas in the reference method, the loudness of all test electrodes was adjusted to match that of a common reference electrode. Five subjects implanted with the Nucleus 22 device completed both methods of loudness balancing for all of their functioning electrodes. Each test/reference electrode pair was loudness balanced six times to assess the variability of the two methods.The loudness balance curves for the two methods were statistically correlated (p < 0.001) for all subjects. The slopes of the regression lines for the loudness balance curves were statistically different from zero (p < 0.05) for roughly half of the subjects for each method. A sign test indicated statistically different means for the basal set and apical set of measurements for only one subject for both methods. The variance in the measured values across electrodes for the reference method was significantly greater for three of the five subjects (p < 0.01).It was hypothesized that because of its dependence on previously measured values, the adjacent method could be susceptible to "drift," i.e., a shift in the overall loudness to which the electrodes are balanced. However, none of the statistical measures employed to test for drift indicated that the adjacent method was more susceptible to drift than the reference method, nor were the responses to the adjacent method more variable. Thus, based on these results, dependent measurements do not seem to be less accurate than independent measurements. The relatively higher variance for the reference method in some subjects may be due to the difficulty of comparing the loudness of stimuli that are far apart in pitch.

Duke Scholars

Author Leslie M. Collins Electrical and Computer Engineering