Slide 1Slide 2Slide 3Slide 4Slide 5Slide 6Slide 7Slide 8Slide 9Slide 10Slide 11Slide 12Localization cues with bilateral cochlear implantsBernhard U. Seeber and Hugo Fastl (2007) Maria Andrey BerezinaHST.723 April 8th, 2009Goals and BackgroundInvestigate good localization ability that certain people with bilateral cochlear implants (CI) have.Localization in normal hearing subjects: 1) ITD is the dominant cue for low frequency tones (low-passed noises)2) ILD is the dominant cue for high frequency tones (high-passed noises)3) For wide band noises both cues have significant weight, although ITD has larger weight than ILD.4) Ongoing ITD cues in envelop of high-pass noise contributes to localization.Present study: CI processors transmit ILD and envelop ITD, hence contribution of these two cues to localization is of interest.MethodTwo subjects:BW: 50 y.o. male both processors: 300 – 5547 Hz for 12 channelsDF: 54 y.o. male left processor: 200 - 8500 Hz for 8 channels right processor: 200 - 8500 Hz for 11 channelsLocalization: light pointer method – subject adjusted a movable light spot to perceived directionExperiment 1: Localization of free-field sound sources with different spectral content and temporal envelop structure. Experiment 2: Localization of wideband noise with acoustic effects of the head minimized. Experiment 3: ITDs and ILDs were manipulated in virtual acoustic space.Experiment 1Localization of free-field sound sources with different spectral content and temporal envelop structure.Low-pass noise (LPN): restricted availability of ILD, localization on basis of carrier or envelop ITDsLPN, 200 ms env. – some ILD, carrier ITD, no envelop ITDLPN, pulsed – some ILD, carrier ITD, envelop ITD High-pass noise (HPN): localized on basis of ILD, ITD could contributeHPN, scrambled – ILD, some envelop ITDWide-band noise (WBN): ILD as well as both carrier and envelop ITD cues are availbleWBN, pulse – ILD, carrier ITD, envelop ITDWBN-CI, 200 ms env. - ILD, carrier ITD, some envelop ITDExperiment 1: ResultsBW DFLPN: BW –unable to localizeDF – can localize: LPN with slow envelop changes - relatively well; when envelop ITD cue is strengthened, LPN pulsed, localization improvesExperiment 1: ResultsBW DFWBN: Both are able to localize, BW localizes WBN of either temporal structure equality well; DF can localizes both WBN stimuli well (but does slightly poorer on pulsed WBN stimulus)Experiment 1: ResultsBW DFHPN Scrambled: Both are able to localize, ILD or envelop ITD must have been used.Experiment 2Localization of wideband noise with acoustic effects of the head minimized. Subject:DF Stimulus:WBN Pulsed: ILD as well as both carrier and envelop ITD cues are availableConfigurations:“Together” – processor brought together at the center of the head 18cm above the head (minimum ITD and ILD)“Distanced” – processors were placed exactly above ears, with distance in between them approximate to head diameter (preserve ITD, but give nearly no ILD)“Plate” – processors were placed next to each other but were separated by carton board with area approximate to cross-section of the head. (ITD are nearly absent, while ILD is evoked)Experiment 2: ResultsThe results of “Plate” condition suggest that ILD is a strong cue while results of “Distanced” condition indicates a small contribution of ITD cues. ”Together” – no localization“Distanced” – right – left differentiation“Plate” – localization improved significantlyExperiment 3ITDs and ILDs were manipulated in virtual acoustic spaceSubject :BW Stimulus:WBN Pulsed, HPN, 100 ms env., HPN, pulsedVirtual Acoustics with CIs 1)Head-related transfer functions (HRTFs) were derived; response to maximum length sequences (MLS) were measured via CI’s processor; the processors were set to minimum amplification, thus resulting in least compressive setting.2)The test stimuli were recorded also directly off processors with amplification and compression setting commonly used by subject. The recorded stimuli this contained all the directional information. 3)ITD and ILD bias was introduced by filtering the prerecorded stimulus with ITD- or ILD – only filter.Experiment 3: ResultsPulsed WBN: follows an ILD offset by about 50%. This confirms that ILDs provided the predominant information for localization of wide-band sounds, relative contribution of ITD cues is small. HPN with slow envelope changes: ILD-weight increases and ITD-influence appears absent. Pulsed HPN: The dominance of ILDs did not change. The pulsation emphasized envelope ITDs and their weight increased, but it still remained low.Summary:Both subject consistently rely on ILD for localization of all tested sound types, while ILD contributed small amount of