Although cochlear implants can improve hearing, most deaf people still can't use a telephone and can converse only with the aid of lip-reading. However, several MIT researchers are developing software and electronics to improve the performance of the implants.
Cochlear implants use electrodes surgically implanted into the cochlea (an organ of the inner ear) and a sound-processing "box" to restore some hearing to people who have lost it. The devices have been commercially available since the late 1970s and are used by about 6,000 Americans today, according to Dr. Donald Eddington, a principal research scientist with the Research Laboratory for Electronics (RLE) who helped develop one of the first multi-electrode implants at the University of Utah in the 1970s.
"Our goal is twofold: to understand the basic mechanisms that underlie the hearing provided by these implants, and to use that information as a rational basis for developing better systems," he said.
The project to improve sound processors is a joint program between MIT, Harvard Medical School, Draper Laboratory and the Massachusetts Eye and Ear Infirmary. Chief researchers are Dr. Eddington; Dr. William Rabinowitz, a principal research scientist in RLE's Sensory Communication Group; and RLE affiliates Joseph Tierney and Marc Zissman, who are also members of Lincoln Laboratory's Speech Systems Technology Group. The work is part of a larger program that involves other MIT scientists, including members of RLE's Speech Communication Group.
Implant recipients wear a small microphone on a hook over the ear that picks up sounds and transmits them along a cable to the processor, a Walkman-size box worn on the body. The processor translates the sounds into multichannel signals and sends them back up the cable to a special connector system behind the ear. The signals are conveyed to the electrodes implanted in the snail-shaped cochlea, where they elicit electric spikes on the auditory nerve fibers that go to the brain.
Although these nerve fibers number in the tens of thousands, implants are limited to only a few electrodes (usually six to 22, depending on the model). As the number of electrodes increases, the space between them decreases and the degree to which their electric fields overlap is greater, resulting in more interference between them. New signal processing schemes are being studied that attempt to minimize this interference by interleaving the stimuli at the active electrodes so only a single electrode is activated at any one time.
Commercial implants use "boxes" that are very limited in the sound processing schemes they are able to implement. The MIT researchers have developed a laboratory system that allows them to implement a wide range of new sound processing systems in a very flexible and efficient manner.
At the heart of this system is PISCES-the Programmable Interactive System for Cochlear Implant Electrode Stimulation that was designed by Drs. Tierney and Zissman. Subjects try out the
new processing schemes implemented on PISCES by unhooking their own processors and connecting to the lab's, which presently consists of a mainframe-sized bank of equipment that includes a computer workstation and wires running through a wall to a soundproof testing room. The real-time hardware and software of this laboratory system build on the expertise developed at Lincoln Laboratory for military applications of digital speech technology.
In a joint effort between researchers at MIT, the University of Geneva and the Research Triangle Institute, a new prototype processor has been built which is about the same size as the commercially available devices. This new wearable unit will enable the researchers to download promising processing systems from the large laboratory system for use by subjects in their everyday lives. The scientists expect the processing systems that are producing significant improvements in the lab (where subjects have almost no time to adapt to the new sounds) will show even more benefit once subjects experience them for 10 to 15 hours each day.
The new prpcessing systems increase the clarity of what subjects hear, although it's almost impossible for the researchers to know subjectively just how the sound quality has changed. "We don't have a very good language to describe what we hear," Dr. Eddington noted. However, the improvement is unmistakable. The hearing of some improves to the point where they no longer need to read lips as they carry on a conversation. "That's the goal that we're aiming for with the new processor-to move ever closer to the gold standard of communicating without lip-reading," he said.
Michael Pierschalla is one of two subjects using the new portable prototype that incorporates the latest laboratory sound processing schemes. Since he began using the new box last July, his hearing is "very much clearer;" he now understands about 95 percent of words he hears, compared to 75 percent before, he said. "Missing [that] 20 percent makes a substantial difference. Before, I could talk on the telephone, but it was a struggle. Now I have no problem with that kind of stuff."
Until he suddenly lost his hearing at the age of 20, music had been an important part of his life. Getting a cochlear implant 10 years ago gave him back some hearing, but not enough to recapture his appreciation of music. In the past few months, however, he began "an orgy of tape-buying," he said. Listening to music with the aid of the prototype processing box, "I'm hearing so much more detail than I ever thought I'd hear again. It's been one of the most delightful things about it."
"Music sounds richer-I can hear more instruments, although I still can't hear the words very well," said John Anderson, the other subject using the new prototype. Another thing he's now able to hear is the sound of his wife's voice upstairs when she talks on the phone. "People tell me I'm picking up their voices quicker. And in group situations, it's easier for me to follow conversations," he said. Mr. Anderson has had a cochlear implant since late 1984 and was hard of hearing all his life until going completely deaf in 1982.
Although the new processing system represents an improvement over what is now available, it does have limitations. For example, it may not provide the amplitude resolution that normal hearing does. The loudest tolerable sounds are normally 10,000 times greater than the softest discernible sounds, while with implants, the range is only a factor of 10. Some deaf people are helped only slightly by implants, and people born with hearing but who lost it later (because of disease, trauma or age) usually receive more benefit than those born deaf.
The work is being supported by the National Institutes of Health, NYNEX and internal funding from Lincoln Laboratory and Draper Laboratory.
A version of this article appeared in MIT Tech Talk on May 10, 1995.