Normal hearing relies on being able to interpret pitch perfectly. This isn’t just important in allowing us to hear all the tones of a beautiful piece of music, it’s also key to basic understanding of speech. This is especially true for tonal languages like Chinese, Vietnamese, Punjabi, Zulu, and Navajo.
At least one and a half billion people worldwide speak tonal languages, which have a musical quality to them that is critical to meaning. Even in English — which can usually be understood even when spoken in monotone — inflections can mean the difference between a question and a statement, or provide key information about emphasis.
So when a person receives a cochlear implant to help restore their hearing, tuning the device so they can hear the right tones is incredibly important. Anything less is like listening to an instrument that is always out of tune.
To help accomplish this, researchers at Western University have developed a mathematical tool to help tune cochlear implants to each patient’s unique anatomy. Their study was published in IEEE Transactions on Biomedical Engineering.
The cochlea is the last part of the inner ear that contains the sensory cells that actually pick up sound: the ear drum vibrates in response to sounds in the air, amplifying and transferring them through tiny bones called ossicles, which then relay those vibrations to the cochlea by tapping on its oval window.
The cochlea is made of a thin membrane shaped like a snail shell, coiled into a spiral and encased in bone. It’s a fluid-filled organ lined with sensory hair cells, each armed with bundles of hair-like projections that sway in response to the motion of the fluid. Depending on a hair cell’s position inside the spiral, the stiffness of the membrane it sits on top of varies, making different parts of the cochlea more sensitive to certain tones. The cells then convert those mechanical waves into electrical impulses that travel to the brain.
But depending on the person, the exact size and shape of the cochlea varies, and that also means that the positions of the nerves that pick up each tone of sound aren’t universal. Trying to use just one map for all cochlear implant recipients can send the wrong tones to the brain.
The implant itself is a thin wire-like device surgically threaded into the spiral of the cochlea, and along its length are electrodes that can deliver electrical signals directly to the cochlear nerves. It works by converting sounds picked up by a microphone that rests outside the ear, bypassing any damaged parts of the inner ear to send signals to the brain. The challenge lies in knowing which electrodes to turn on and when.
“Until now, cochlear implant programming was performed in a sort of one-size-fits-all approach,” said first author and PhD candidate Luke Helpard in a press release.
“We’re using imaging data to be able to customize the pitch map for each patient based on individual patient anatomy and post-operative electrode locations within the cochlea.”
The personalized pitch maps help match up the right tones to the right electrodes. The team started by doing ultra high-resolution 3D scans of ten cadaveric human cochlea, collaborating with Canadian Light Source in Saskatoon. This provided an unprecedented look at cochlear anatomy without any distortions, as they are tiny as a pea and encased in the densest bone in the body. With these frequency maps in hand, the information can now be used to adjust cochlear implants based on patient CT scans.
The researchers hope that in the future, gathering even larger datasets could help train deep learning tools to quickly adapt to every patient’s unique anatomy. This will help restore a rich and accurate perception of sound.