FITZGERALD, Matthew, Cochlear Implants, 25 April 2019
<p>FITZGERALD, Matthew, Cochlear Implants, 25 April 2019
this is at 12:30PM today at Prof. Daniel Palanker's series of seminars.</p>
<p> </p>
<p>lateral wall is farther from spiral ganglion but preserves it;</p>
<p> </p>
<p>medial wall is closer but may damage the spiral ganglion.</p>
<p> </p>
<p>short array (only proximal to help just with high freq sound)</p>
<p> </p>
<p>12 to 22 electrical contacts...</p>
<p> </p>
<p>usu have 15,000 hair cells... but gets funneled down to 12 to 22 trodes.</p>
<p> </p>
<p>spread of excitation in the fluid... and cuz of distance to nerve (Daniel emphasizes this)</p>
<p> </p>
<p>with medial wall... how far away are you...</p>
<p> </p>
<p>dendrites are first to die off... prob stim spiral ganglion bodies themselves...</p>
<p>he says distance may be 2 mm!!! it is mm</p>
<p> </p>
<p>student also asks about change in distance from lateral to medial... perhaps a mm or so.</p>
<p> </p>
<p>trodes do not line up with frequencies....</p>
<p> </p>
<p>sometimes two trodes will sound exactly the same to thepatient... can turn them on/off to experiment.</p>
<p> </p>
<p>doing an FFT. decomposes into low freq versus high freq.</p>
<p> </p>
<p>every company filters about 300 hz (strips out all the fine structure) ie each band is about 300 hz.</p>
<p> </p>
<p>try to reshape the envelope of the FFT.</p>
<p> </p>
<p>implants.... (go from 15k fine struc down to 22 trodes)</p>
<p> </p>
<p>how much trouble do you have with background noise.</p>
<p> </p>
<p>need fine structure and need to solve the diffusion problem.pick</p>
<p> </p>
<p>22 trodes.... only stim perhaps 8 that are the most active... and don't fire the others (dynamically)</p>
<p> </p>
<p>pick the subset with the most energy....</p>
<p> </p>
<p>fine temporal structure is the holy grail</p>
<p> </p>
<p>the freq info is converted to electrical pulses that follow the envelope of the sound.</p>
<p> </p>
<p>(some expts with laser to nerve to avoid diffusion)</p>
<p> </p>
<p>but notice that evein in energy gets sent to correc tpart of cochlea, the tempral fine struc is discarded.</p>
<p> </p>
<p>things get muddy... when current is spread out (he shows broad pyramids instead of sharp peaks)</p>
<p>that'd due to spread of excitation...</p>
<p> </p>
<p>formants: oooo versus shhhh versus eeee... energy at difft frequencies.</p>
<p> </p>
<p>brain dies off when input is cut....</p>
<p> </p>
<p>another problem...</p>
<p> </p>
<p>frequency mismatch....</p>
<p> </p>
<p>eg close in at outer part of snail shell is</p>
<p> </p>
<p>3800, nex tis 2600, next 2000 to 2600 next 1450 to 2000; next 1015 to 1450 next</p>
<p>730 to 1015 hz next 500 to 730 next 250 to 500 hx\z.</p>
<p> </p>
<p>but those are differen thant the charac freq expecte dby the neurons.</p>
<p> </p>
<p>he starts with Lincoln speech: (may sound like Darth Vader or sorse)</p>
<p> </p>
<p>8 channels with spectal degad and freq mismatch...</p>
<p> </p>
<p>so, limits....</p>
<p> </p>
<p># of trodes,</p>
<p> </p>
<p>spread of the current;</p>
<p> </p>
<p>no temporal fine struc (poor pitch perception)</p>
<p> </p>
<p>limited dynamic range </p>
<p> </p>
<p>variable neural survival (some regions are just gone!)</p>
<p> </p>
<p>he encourages hearing aids...</p>
<p> </p>
<p>incomplete insertion depth (upward shift of freq)</p>
<p> </p>
<p>and lack of binaul synchrony (get poor spatial localization)</p>
<p> </p>
<p>Adaptation in users of cochlear implants....</p>
<p> </p>
<p>(in OR... monitor audit N firing... are we damaging the nerve, eg)</p>
<p> </p>
<p>how do we assess adaptation?</p>
<p> </p>
<p>how shows sentence ID with cocheal rimplants a t 1 month... avg 70% and then goe sup</p>
<p>to about 87% over 6 months to 2 years.</p>
<p> </p>
<p>CI = cochlear implant users.</p>
<p> </p>
<p>1) is freq mismatch a problem?</p>
<p> </p>
<p>Micky Mouse mini mouse versus Darth Vader...</p>
<p> </p>
<p>want to do real-time frequency maping by Svirsky et al... inlcuding Matt.</p>
<p> </p>
<p>got kids book and did filtering into say 22 channels.</p>
<p> </p>
<p> </p>
<p>(played " 4 score and 7 years ago... our fathers brought forth")</p>
<p> </p>
<p>tws the night before XMAS and all thru the house; not a creature was stirring...</p>
<p> </p>
<p>change mapping in real time.</p>
<p> </p>
<p>Gridstream....</p>
<p> </p>
<p>The player lost a shoe....</p>
<p> </p>
<p>standard is from about 300 to 8000 Hz... but some people like higher and some lower.</p>
<p> </p>
<p>75 yo woman with Nucleus device in 2007.</p>
<p> </p>
<p>people like to select their own freq table. (makes diff in 50%)</p>
<p> </p>
<p>iPhone based FAT selection = freq table.</p>
<p> </p>
<p>grid squares on iPhone... you select with app.</p>
<p> </p>
<p>Q does FAT change over time?</p>
<p> </p>
<p>A yes... some change others never.</p>
<p> </p>
<p> </p>
<p> </p>
<p> </p>
<p> </p>