UP Bioengineering Our people Design and application of - - PowerPoint PPT Presentation

▶
up bioengineering
SMART_READER_LITE
LIVE PREVIEW

UP Bioengineering Our people Design and application of - - PowerPoint PPT Presentation

Aug 14, 2023 232 likes •333 views

2015/07/15 UP Bioengineering Our people Design and application of user-specific models of cochlear implants Tania Hanekom Tiaan K Malherbe, Liezl Gross, Rene Baron, Riaze Asvat, Werner Badenhorst & Johan J Hanekom Larry Schmidt, Tiaan

slide-1
SLIDE 1

2015/07/15 1

Design and application of user-specific models of cochlear implants

Tania Hanekom

Tiaan K Malherbe, Liezl Gross, Rene Baron, Riaze Asvat, Werner Badenhorst & Johan J Hanekom

UP Bioengineering

Our people

CIAP 2015

2

Larry Schmidt, Tiaan Malherbe, Johannes Myburgh, Pieter Venter, Rene Baron, Dirk Oosthuizen, Johanie Roux, Werner Badenhorst, Liza Blignaut, Johan Hanekom, Liezl Gross, Heinrich Crous, Tania Hanekom, Alex Oloo. Insert: Riaze Asvat.

UP Bioengineering

Our place

CIAP 2015

3

www.up.ac.za/eece

UP Bioengineering

is part of

Electrical, Electronic and Computer Engineering

CIAP 2015

4

http://www.ee.up.ac.za/main/emk310/index

slide-2
SLIDE 2

2015/07/15 2

Agenda

  • The development of our user-specific models

– Human and guinea pig model generations – What the models can do – What the models show

  • Translating models into tools

– Research tools – Model-predicted mapping (MPM) – Model-based diagnostics (MBD)

CIAP 2015

5

Progression of UP Bioengineering's volume conduction models

Human Generation 1

HG1: Generalised human cochlea extruded on analytical spiral from 2D section (1996-2001).

CIAP 2015

6

User-specificity in cochlear implants…

CIAP 2015

7

  • Realised that models need to migrate to include user-specific

characteristics

  • Started to work on user-specific models in 2007
  • Micro-CT data from guinea-pig subject plus neural data from same subject

was made available through Russ Snyder/ Ben Bonham from Pat Leake’s lab (Epstein Labs)

Micro-CT of subject’s cochlea Acoustic frequency response areas of the 16 electrode contacts implanted in the inferior colliculus

What qualifies subject-specificity in animals?

CIAP 2015

8

  • Subject-specificity characterised by, among others:

– duration of cochlear implantation and duration of deafness – age of implantation – stimulation mode (BP, MONO, CG) – electrode insertion depth – design of the electrode array – position of the electrode array – neural survival patterns – cochlear morphometry

slide-3
SLIDE 3

2015/07/15 3

Progression of UP Bioengineering's volume conduction models

Guinea pig Generation 1

GPG1: From CT to FEM. Subject-specific cochlear morphometry and electrode location included (2007-2009).

CIAP 2015

9

Progression of UP Bioengineering's volume conduction models

Guinea pig Generation 2

CIAP 2015

10

GPG2: Adding subject-specificity: bone capsule and hook area (cochlear morphometry), return electrode location (2007-2009).

What qualifies user-specificity in humans?

CIAP 2015

11

  • Speech perception variability caused by, among others:

– duration of cochlear implantation and duration of deafness – age of implantation – stimulation mode (BP, MONO, CG) – electrode insertion depth – design of the electrode array – position of the electrode array – neural survival patterns – cochlear morphometry – speech perception before implantation – speech processing algorithm used – unilateral or bilateral implantation – …

From dead guinea pig to live human

CIAP 2015

12

Challenge 1: Resolution of image data VC model Challenge 2: Access to neural response data ANF model

slide-4
SLIDE 4

2015/07/15 4

Progression of UP Bioengineering's volume conduction cochlear models

Human Generation 2

HG2: Realistic generalised human cochlea extruded on spiral derived from mid-modiolar section of cochlea (2009). TEMPLATE MODEL USED AS BASE FOR SUBSEQUENT GENERATIONS

CIAP 2015

13

  • HG3. Person-specific models based on CT data from live

implantees (2010>>).

CIAP 2015

14

Progression of UP Bioengineering's volume conduction cochlear models

Human Generation 3

HG1-3 & GPG1. Cochlea embedded into infinite bone volume (outer surface of sphere/cylinder modelled at infinity).

CIAP 2015

15

Progression of UP Bioengineering's volume conduction cochlear models

Extra-cochlear volume: infinite homogeneous

  • HG4. Cochlea embedded into head-sized ellipsoid bone volume

with accurate description of return electrode.

CIAP 2015

16

Progression of UP Bioengineering's volume conduction cochlear models

Human Generation 4

slide-5
SLIDE 5

2015/07/15 5

CIAP 2015

17

Progression of UP Bioengineering's volume conduction cochlear models

Human Generation 5

  • HG5. Cochlea embedded into skull with brain and scalp and

accurate description of return electrode.

CIAP 2015

18

  • HG5. Detail of return electrode placement.

Progression of UP Bioengineering's volume conduction cochlear models

Human Generation 5

Current status of VC model

CIAP 2015

19

  • User-specific cochlear

morphometry (macro characteristics)

  • User-specific electrode

location

  • Correct return electrode

location

  • Description of skull,

brain and scalp

CIAP 2015

20

2013: Initiated project to create morphometric library of inner structure templates

  • collaborate with Dept Anatomy, Faculty of Health Sciences
  • address low-res/soft tissue problem
  • Improve user-specific model representation of morphometry
  • 60 dry skulls imaged and digitized to date (micro-CT)

Progression of UP Bioengineering's volume conduction cochlear models

NEXT: Human Generation 6

slide-6
SLIDE 6

2015/07/15 6

Progression of UP Bioengineering's auditory nerve fibre (ANF) models ANF models integrate with VC models to predict neural excitation from spread of electrical activity.

We have used / are using

  • GSEF / Hodgkin-Huxley / Rattay (literature)

We have worked on

  • Smith-2·Hanekom model (own)

Problems

  • Single fibre instead of population
  • Electrical stimulation
  • Bottom-line: can't predict absolute thresholds; trends okay

CIAP 2015

21

  • Working to create physiologically-based neural models that

can predict responses to electrical stimulation

– Computationally INTENSIVE! POSTER W26 Development of a voltage dependent current noise algorithm for conductance based stochastic modelling of auditory nerve fibre populations in compound models

Werner Badenhorst

Progression of UP Bioengineering's auditory nerve fibre (ANF) models

CIAP 2015

22

How big is the influence of person- specificity on the periphery?

  • Neural threshold profiles of

– five cochlear models – inserted with identical medial and lateral arrays – stimulated on electrode 4 – inserted at the same angle in all the models.

  • Mean medial-lateral difference in

thresholds: 6.4 dB

  • Mean medial-lateral difference in CFs:

2988 Hz.

Istim [dB re 1 A] S13R S13L S3R S3L S25R 65 70 75 80 85 Lateral Array S13R S13L S3R S3L S25R Medial Array Istim [dB re 1 A] apex base 4000 8000 12000 16000 65 70 75 80 85 Frequency along Organ of Corti [Hz] a) Non-degenerate Neurons b) Degenerate Neurons CIAP 2015 23

Inter-user variability of 3.93 dB (both) Inter-user variability of 2535 Hz (medial) and 1992 Hz (lateral).

But are the models useful?

CIAP 2015

24

– duration of cochlear implantation and duration of deafness – age of implantation – speech perception before implantation – speech processing algorithm used – stimulation mode used – unilateral or bilateral implantation – electrode insertion depth – design of the electrode array – position of the electrode array – cochlear morphometry – neural survival patterns – complications e.g. Scar tissue e.g. Bone impedance ??

Predict characteristics of peripheral neural excitation using a specific speech algorithm and a specific stimulation mode

(e.g. frequency matching between ears) Integral part of the VC models; affect neural excitation. Probe the effect of neural survival on excitation characteristics e.g. FNS

slide-7
SLIDE 7

2015/07/15 7

Translating models into tools

CIAP 2015

25

  • Research tools

– Underlying functioning of auditory system

  • Clinical tools

– Visualization – Model-predicted mapping (MPM) – Models-based diagnostics (MBD)

  • 1. Find ways to build models quickly
  • 2. Augment images to improve representation of

inner structures (HG6)

  • 3. Define what we need from clinicians to enable

us to do this as a routine procedure

– Pre-op & post-op scans, neurophysiological data, psychoacoustic data, etc. – Challenge in SA: User records very difficult to find, e.g. can’t find user records of imaging data older than a couple of months.

CIAP 2015

26

Research Tools

CIAP 2015

27

  • Probe the fundamentals that underpin the

functioning of the auditory system

  • Model parameters

– Quantify characteristics of the system so that we can describe it with mathematics – Example: bone impedance

Research Tools

CIAP 2015

28

  • From our models we know bone impedance affect

spread of electrical activity, i.e. neural excitation

Equipotential surfaces at 0.5 dB below the electrode potential.

slide-8
SLIDE 8

2015/07/15 8

Clinical Tools

Visualization

CIAP 2015

29

Original CT image Image derived from model Scala Vestibule insertion Scala Tympani insertion Possible damage to cochlear wall

Clinical Tools

Model-Predicted Mapping (MPM)

CIAP 2015

30

Example: tool to estimate spread of excitation

  • Commonly accepted 3 dB/mm decay for monopolar stimulation does not

hold.

  • Depends on
  • location in cochlea (basal vs apical)
  • intrascaler location
  • Derive equation to give a “quick-and-dirty” estimation of real current

decay.

  • Useful for mapping
  • Also useful as a research tool in models that use current decay, e.g.

acoustic models.

Scala Vestibuli Reissner’s membrane Stria Vascularis Scala Media Organ of Corti Buffer area Scala tympani Basilar membrane Spiral ligament Axonal nerve Source positions

Clinical Tools

Model-Based Diagnostics (MBD) Example:

A case of facial nerve stimulation

Potential of MBD

  • Investigate the factors that may cause FNS
  • e.g. current paths
  • Investigate the potential effectiveness of interventions

based on the current implanted system

  • e.g. optimal electrode configuration
  • Investigate the potential effectiveness of customized

interventions designed for the individual

  • e.g. alternative stimulation strategies

CIAP 2015

31

For details about model-based tools in development

CIAP 2015

32

POSTER W28 Model-based interventions in cochlear implants

Potential distributions as a result of stimulation with different electrodes Nerve fibre plane Electrode array Electrode array Scala vestibule insertion Auditory nerve Spiral ganglion Red area indicates location of electrode contact

slide-9
SLIDE 9

2015/07/15 9

Other work

CIAP 2015

33

Neurophysiology Perception Computational models

Parameters

1. Model of central processing

  • 2. Acoustic

Models

  • 3. Volume

conduction model

  • 4. Nerve fibre

model

Models of the physical situation Models of perception

  • 1. Psychoacoustics
  • 2. Speech

perception Poster R13 NEW APPROACHES TO FEATURE INFORMATION TRANSMISSION ANALYSIS (FITA) Dirk JJ Oosthuizen, Johan J Hanekom Poster R58 RATE PITCH WITH MULTI-ELECTRODE STIMULATION PATTERNS: CONFOUNDING CUES Pieter J Venter, Johan J Hanekom

Thank you

CIAP 2015

34

For postdoc positions in our group, speak to Prof Johan Hanekom Head of UP Bioengineering Johan.hanekom@up.ac.za

  • r

Prof Tania Hanekom tania.hanekom@up.ac.za