Cochlear implants provide functional hearing to people who are profoundly deaf or hearing impaired by replacing the function of missing sensory receptor cells with an array of stimulating electrodes. We have developed a model for predicting the optimal number of electrodes, as well as the optimal locations and usage probabilities of electrodes, from an information theoretic perspective. However, the information theoretic method does not quantify the performance of electrode place discrimination. Here, we apply a neural network classifier to simulations of the cochlear implant model to predict electrode classification error rates. We also investigate the locations along the electrode array where errors are most likely to occur. We conclude based on our model that i) the classification error rate increases with increasing number of electrodes and the classification errors occur predominantly between adjacent electrodes, ii) by inserting the electrode array deeper into the cochlea, more electrode locations can be distinguished and the electrodes for which most errors occur are strongly influenced by the geometry between electrode-neural interface.
Ms Xiao (Demi) Gao is currently a PhD student at the Institute for Telecommunications Research, University of South Australia under the supervision of Assoc. Prof. Mark D. McDonnell and Prof. David B. Grayden (University of Melbourne). Her PhD research is focused on theoretically estimating the performance of cochlear implants and modeling the cochlear implant system.
Demi received a Bachelor in Computer Science in Central China Normal University in 2006, and a Master in Biology in Huazhong Agricultural University in 2009. She worked at the Institute of Hydrobiology, Chinese Academy of Science as a research assistant from 2008 to 2009. Then she worked as a software engineer in South Crown Industry and Commerce Co Ltd of Zhuhai, China from 2009 to 2012.
Demi was the 2013 winner of the University of South Australia 3 Minutes Thesis competition, and progressed to the final 8 at the 2013 Trans-Tasman final.
|Thursday, 9th April, 2015
LG28 Napier Building (Lower Ground)