• Researchers at the Bionics Institute are developing a way to measure tinnitus by recording brain activity.
  • Currently, it is difficult for doctors to diagnose and monitor tinnitus because they can only rely on self-reported symptoms.
  • The aim of our research is to develop a definitive test of tinnitus, to aid diagnosis and development of potential treatments.

What is tinnitus?

Tinnitus is the experience of ringing, buzzing, whirring or other noises in one or both ears. The sounds heard by people with tinnitus can’t be heard by others and originate in the brain.

About 1 in 5 Australians have tinnitus that severely affects their quality of life. It has many causes and can lead to anxiety, depression and sleep issues.

Currently there is no cure for tinnitus and treatment tends to treat distress symptoms or provide ways to mask the noises, but they are not always effective.

The issue with tinnitus diagnosis

Tinnitus is described differently by everyone who experiences it, and reliance on self-reported symptoms makes diagnosis and monitoring of this condition difficult.

Our tinnitus research team at the Bionics Institute have developed an objective measure of tinnitus using a non-invasive optical imaging device.

Finding a new way to measure tinnitus

The optical imaging device uses a cap (pictured below) to shine near-infrared light over the head and measure changes in blood oxygen levels in the brain.

The light reflected back provides detailed information on brain activity. This information is recorded on a computer and analysed by researchers, with the aim of setting a baseline for tracking changes in the brain triggered by tinnitus.

The device and analysis methods, described in more detail below, have been tested in a small study on people with tinnitus and people with no symptoms.

The results show differences in brain activity between people with and without tinnitus as well as individuals experiencing tinnitus at different severity levels.

This and further information gathered from further research trial participants will be used to develop a definitive test of the presence and severity of tinnitus to aid diagnosis and develop potential treatments.

Media stories

For more information, listen to our recent media stories:

BBC Digital Planet
Tinnitus Talk Podcast

Get involved with Bionics Institute tinnitus research

Researchers need to perform tests on more people, both with and without tinnitus. We are looking for volunteers to visit our facilities in East Melbourne and have a test using the optical imaging device.

If you would like to take part in the research, you can find out more about the clinical research study for people with tinnitus.

If you are over 50 and have no history of tinnitus, you can find out more about the clinical research study for people with standard hearing.

We are also carrying out a clinical research study with people who have a cochlear implant and experience tinnitus.

The research team

BI researchers: Dr Mehrnaz Shoushtarian (PI)Associate Professor James Fallon, Michelle Bravo, Shreyasi Datta

More information for researchers

We are working on developing an objective measure of tinnitus using a non-invasive brain imaging technique called functional near-infrared spectroscopy (fNIRS) together with machine learning analysis techniques. Our initial study showed that fNIRS is a viable technique for measuring tinnitus-related brain activity.

We recorded fNIRS signals both at rest and in response to different stimuli and extracted features from the recordings under different conditions. We then applied machine learning methods, including feature extraction and classification, to the fNIRS features.  

Our algorithms were able to classify patients with tinnitus and controls as well as tinnitus at different severity levels with promising accuracy (see publication below). We are now collecting further data and improving our algorithms to verify our initial findings.


Shoushtarian M, et al. (2020) Objective measurement of tinnitus using functional near-infrared spectroscopy and machine learning. PLOS ONE 15(11): e0241695. https://doi.org/10.1371/journal.pone.0241695