1. Modifying the Auditory Nerve InputModifying the Auditory Nerve Input
to a Computational Model of theto a Computational Model of the
Dorsal Cochlear NucleusDorsal Cochlear Nucleus
Adam GiangAdam Giang
Spring 2008Spring 2008
Boston University College of Engineering
2. OverviewOverview
ObjectiveObjective
Auditory System and ModelingAuditory System and Modeling
Techniques (Background Information)Techniques (Background Information)
Experimental ProcedureExperimental Procedure
ResultsResults
SummarySummary
3. ObjectiveObjective
Study the effects of auditory trauma on theStudy the effects of auditory trauma on the
brain’s interpretation of soundbrain’s interpretation of sound
Simulate impaired auditory nerve (AN) andSimulate impaired auditory nerve (AN) and
dorsal cochlear nucleus (DCN) usingdorsal cochlear nucleus (DCN) using
computational modelscomputational models
Evaluate models’ effectiveness in aEvaluate models’ effectiveness in a
physiological contextphysiological context
4. Modeling the Auditory SystemModeling the Auditory System
Sound
Pressure
Signal s(t)
AN
Model
DCN
Model
Response
Maps
5. BruceBruce AuditoryAuditory NerveNerve
ModelModel
Project uses Auditory Nerve model as described by ZilanyProject uses Auditory Nerve model as described by Zilany
and Bruce (2006)and Bruce (2006)
Input: Instantaneous Pressure WaveformInput: Instantaneous Pressure Waveform
Output: Spike Train (Action Potentials)Output: Spike Train (Action Potentials)
Allows for inner (IHC) and outer (OHC) cochlear hair cellAllows for inner (IHC) and outer (OHC) cochlear hair cell
impairmentimpairment
6. DCN ModelDCN Model
DCN model asDCN model as
described by Hancockdescribed by Hancock
and Voigt (1999)and Voigt (1999)
5 cell groups of 8005 cell groups of 800
isofrequency slicesisofrequency slices
spaced 0.005 octavesspaced 0.005 octaves
apart and centered atapart and centered at
5 kHz5 kHz
P cell behaviorP cell behavior
dependent ondependent on
connection parametersconnection parameters
7. NeuronNeuron ModelModel
Neuron model based on MacGregor neuromime (1987)Neuron model based on MacGregor neuromime (1987)
Each neuron modeled as a parallel circuit withEach neuron modeled as a parallel circuit with
membrane capacitance, leakage conductance, amembrane capacitance, leakage conductance, a
potassium channel branch, and excitatory/inhibitorypotassium channel branch, and excitatory/inhibitory
connection branchesconnection branches
8. MethodsMethods
AN Model Input: Wide Range of tonal stimulusAN Model Input: Wide Range of tonal stimulus
with intensity varying from 0 to 90 dB SPL in 6with intensity varying from 0 to 90 dB SPL in 6
dB SPL increments and frequency varying in 1.5dB SPL increments and frequency varying in 1.5
Octave band above and below 5 kHzOctave band above and below 5 kHz
Spike Trains Generated by AN model inputtedSpike Trains Generated by AN model inputted
into DCN modelinto DCN model
Response Maps generated for a wide range ofResponse Maps generated for a wide range of
connection Parametersconnection Parameters
Sound
Pressure
Signal s(t)
AN
Model
DCN
Model
Response
Maps
14. ConclusionsConclusions
Impaired AN models exhibit elevated thresholds,Impaired AN models exhibit elevated thresholds,
broadened tuning, and shifted rate-level curves,broadened tuning, and shifted rate-level curves,
consistent with physiological dataconsistent with physiological data
DCN response maps show a decrease inDCN response maps show a decrease in
regions of both excitation and inhibition, coupledregions of both excitation and inhibition, coupled
with an increase in spontaneous areas (tinnitus)with an increase in spontaneous areas (tinnitus)
Connection parameters for DCN model whichConnection parameters for DCN model which
best modelbest model physiological responses still to bephysiological responses still to be
determineddetermined
Future use of models to reveal insight intoFuture use of models to reveal insight into
hearing loss and auditory system itselfhearing loss and auditory system itself
Sound travels through ear canal to cochlea, IHC and OHC connect via AN to DCN, located on brain stem and is site of complex auditory processing. OHC pre-amplifier and IHC transducer, converting sound pressure signal to electrical signal. The way we model this system is…
3 filters, c1 low level, c2 high level, control path filter regulates gain and bandwidth, note impairment parameters
Rate level curves plot average firing rate versus sound level, increase with sound level until saturation. Shift in curves for ohc, ihc more dramatic shift, which dominates when both are impaired.
Tuning curves plot threshold dB versus frequency (in this case, sound pressure level that causes firing rate of 30/s). Impairing OHC causes elevated thresholds and broadened tuning. Impairing IHC had similar effects but without broadened tuning.
Response maps plot excitatory and inhibitory respones to stimulus on a sound pressure level vs. frequency plane. Impairing OHC caused a drop in low level responses. We also observed a favoring of excitatory over inhibitory response, agreeing with the physiological observations of Ma and Young
Impairing IHC had similar results of decreased low level responses and a favoring of excitation over inhibition. Impairing IHC also caused a drop in mid level responses
Impairing all hair cells leads to mainly spontaneous activity, and the chart looks like IHC impairment alone (since its effect encompasses that of the OHC).