Project Summary/Abstract
Cochlear implants are highly successful neural prostheses that enhance or restore hearing to severely
hearing-impaired adults and children. Sound from the environment is converted into electrical pulses and
conveyed to the auditory nerve by an array of electrodes. The cochlear implant provides important spectral and
temporal auditory information to the listener. However, speech perception performance varies considerably
among cochlear implant listeners, particularly in noisy environments and for complex stimuli. A significant
contributing factor to this performance variability is the quality of the electrode-neuron interface, which is defined
by the position of individual electrodes, the relative health and integrity of the neurons and bone and tissue
growth related to trauma from electrode array insertion.
Currently, there is not a clinical tool to query the electrode-neuron interface in cochlear implant listeners,
nor is there an accurate computational model based on human cochlear histopathology to estimate the electrode-
neuron interface. One step towards improving our understanding is to update the classical approach to
quantification of the density and integrity of the auditory neurons in donated human temporal bones from
recipients with cochlear implants. We will harness the power of machine learning to generate 3-D reconstructions
of the neurons relative to the electrodes in the cochlear implant array. To develop clinical measures to predict
variations in the electrode-neuron interface, we will leverage an interdisciplinary team of experts, machine
learning, computational modeling, an unparalleled collection of temporal bones, and a wealth of clinical data.
The long-term goal of the proposed experiments is to improve cochlear implant programming and speech
perception outcomes by improving our understanding of the electrode-neuron interface.
Three aims are proposed: 1) To produce quantitative 3-D maps of cochlear histopathology at the
electrode-neuron interface in temporal bones from cochlear implant recipients and relate them to audiometric
measures and outcomes; 2) To develop and validate a computational model of the electrode-neuron interface
using histopathology, research and clinical measures; and 3) To improve speech perception for cochlear implant
listeners by individualizing programming based on electrode-neuron interface estimates.
The results of the proposed studies are expected to lead to a shift in how we think about cochlear
histopathology and clinical care of individuals with cochlear implants. We will characterize the largest set of
implanted temporal bones with the highest degree of granularity to date. We will provide insight and tools for
assessing the relationship between histopathology, research, and clinical measures. Finally, we will develop
translatable methods for individualized programming for cochlear implant listeners based on a robust
understanding and a validated model of the electrode-neuron interface in humans.
