Cochlear implant surgery

Making speech clearer for cochlear implant recipients like myself

25 July 2018
A biomedical engineering researcher’s personal journey
Greg Watkins, a PhD student in biomedical engineering at the University of Sydney, is profoundly deaf. Now with his own cochlear implant, Greg is developing new metrics to improve how recipients hear speech. This is his inspiring story.

No-one could explain to me why, in my 50s, I was profoundly deaf. My hearing had started to deteriorate in my early 40s and while my mother had bi-lateral cochlear implants (CIs), I was told the condition was not necessarily hereditary.

Even with hearing aids, speech became increasingly difficult for me to understand, but even more so when there was background noise. I could comprehend less than 50% of speech with my right ear and almost nothing with my left ear, which was frustrating not only for me but for family, friends and work colleagues.

As my hearing deteriorated, I wondered if my 30+ years’ technical knowledge as an Electrical Engineer in the telecommunications industry and my personal experience of hearing disability might provide new insights into the challenges facing CI recipients.

Motivated by a desire to make a real difference, I started a PhD research degree in 2014 and then later transferred to the University of Sydney under the supervision of Professor Gregg Suaning – a global leader in implantable bionics – and Dr Brett Swanson, a research scientist at Cochlear.

Receiving my own CI in my left ear earlier this year has been a life-changing experience. My hearing has significantly improved, and the concentration required to understand speech has dramatically reduced.

People with CIs often learn to understand 90% or more of speech in ideal conditions. With background noise or multiple people speaking, speech perception is much worse. I’m experiencing these challenges first-hand as I learn to hear again with my left ear. Restaurants, family gatherings and business meetings have all been challenging to navigate.


What is a cochlear implant?

A cochlear implant is composed of two main components: the implant and the sound processor. The implant is a small capsule of electronics that is inserted under the skin behind the ear and sends stimulation signals to electrodes implanted in the cochlea. The cochlea is the organ that translates sound vibrations into nerve stimulation. A cochlear implant directly stimulates the nerves. The sound processor fits behind the ear, like a hearing aid, and communicates with the implant. It has the task of converting everyday sounds to stimulation patterns that let people understand speech again.

Sound processor improvements, that might make speech clearer, are typically tested by playing recorded sentences to experienced implant recipients under a range of test conditions. The recipients repeat the sentence and a score, determined by the accuracy of the repetition, is assigned.

Alternatively, a mathematical model (or metric) could be used to predict speech intelligibility. With this approach, sound processing improvement ideas could be evaluated by computer simulation and then the most promising ideas tested with recipients. The metric could possibly be used to optimally configure an implant for an individual. The problem is that, while many such metrics have been proposed over the years, few of them have been specifically designed for CIs.

My research is investigating a metric called “Output Signal to Noise Ratio” (OSNR) and how well this predicts speech intelligibility for CI recipients. In a typical listening situation, people hear speech and background noise such as chatter or traffic noise. We call the ratio of speech to noise the Signal-to-Noise Ratio (SNR). When a CI recipient’s sound processor converts sound to nerve stimulation waveforms, the SNR that is heard by the recipient is changed. The idea of the metric is that the recipient’s level of speech understanding will be determined by the SNR that they actually hear – the OSNR.

To date, I’ve shown that OSNR is an accurate predictor in many situations, including more complex scenarios where other metrics fail. The next step is to take existing hearing test data for one test condition, and to use OSNR to predict intelligibility under a different test conditions. If this is feasible, it will open the door for the development of CI algorithms and configurations that are optimised for an individual’s hearing abilities.

As an engineer I apply my technical skills to find solutions to real-life problems. My passion for biomedical engineering lies in applying sophisticated theoretical concepts to develop innovative healthcare solutions that change peoples’ lives.

I was asked recently if being able to hear with my CI had made a difference to my life. Before I could answer, my wife jumped in and said, “it has been amazing”. If my research can change the lives of people with hearing disability by improving how they understand speech, this will be even more amazing.

Related articles