Theorie und Experiment in der klinischen Forschung

1
(No Transcript)
2
(No Transcript)
3
(No Transcript)
4
NeuroimplantsResearch application
? from VDE

• FIAS, 1.2.2008
• Dr. rer. nat. Dipl.- Phys. cand. med. Andreas
  Bahmer
• HNO, Universitätsklinikum Frankfurt am Main

5
Contents

• Aspects of Theoretical Neuroscience
• A role model for an oscillatory pacemaker
  Simulation of oscillating neurons in the auditory
  system
• Neuroimplants
• Perspectives in medical research

6
Theory Experiment

• Theories without a neurobiological substrate are
  not relevant.
• However, a theory need not be based on
  experimentally facts alone but can also reveal
  mathematical principles.
• Neurobiology in the 21st century should tightly
  connect theoretical and experimental
  neuroscience.

van Hemmen, 2006, Editorial of Biological
Cybernetics
7
Neuron
8
Neuron
9
Andi
10
Andi
11
Relevant Information ?
Andi
12
Neuron
13
Neuron
14
HEBB Learning Rule
Neuron
What fires fogether, wires together
15
Andi
16
Andi
17
Lost in details
Can you see me?

• Not see the woods for the trees..
• Sixty four dollar question.
• ...better unsharp ...?

18
A way...
Nachdem sich die Wissenschaft bisher vornehmlich
damit befasst hat, die Welt in ihre Komponenten
zu zerlegen, müssen jetzt die vielfach sehr gut
beschriebenen Bausteine in ihrem Zusammenwirken
betrachtet und besser verstanden
werden, formulieren Singer und Greiner das
Programm des FIAS.
Die ZEIT 2005
The BRAIN
19
Similar (Micro)-Circuits
Similar architecture Cerebellum, Hippocampus, Dors
al cochlear nucleus e.g. Oertel What makes
a cerebellar architecture in the DCN? Form
follows functions? Same pattern?
TINS,Grillner 2005
20
Top down Bottom up Middle sidewards
Top-down
Middle sidewards
Bottom-up
De Schutter 2005
21
My Paradigm No paradigm
Time (Oscillations) Rate (PSTH, Averaging,
fMRI) Everything
a brain like display follows
22
Oscillations
By definition, oscillations are temporal periodic
changes in the state of a system. In nonlinear
systems like the brain, oscillations define a
stable state. Some of theses stable states are
speculated to be the equivalent for short term
memories and play a role in decision making
(Basar-Eroglu et al., 1992). Some authors have
described a theory of memory that is equivalent
to the optical recording technique that is called
holography (Longuet-Higgins, 1968 Gabor,
1968a,b Westlake, 1970). In their theories
memory processes are based on the coherent
interplay of many neurons, similar to holography
where the coherence of many light waves forms a
pattern that is stored.
PhD Thesis, Bahmer 2007
23
Oscillations
By definition, oscillations are temporal periodic
changes in the state of a system. In nonlinear
systems like the brain, oscillations define a
stable state. Some of theses stable states are
speculated to be the equivalent for short term
memories and play a role in decision making
(Basar-Eroglu et al., 1992). Some authors have
described a theory of memory that is equivalent
to the optical recording technique that is called
holography (Longuet-Higgins, 1968 Gabor,
1968a,b Westlake, 1970). In their theories
memory processes are based on the coherent
interplay of many neurons, similar to holography
where the coherence of many light waves forms a
pattern that is stored.
Gabor D (1968a) Holographic model of temporal
recall. Nature 217584
-gt see diffractive optics holographic giant
data store
PhD Thesis, Bahmer 2007
24
Oscillations some historical facts
Oscillations occur in different sensory systems,
like the visual, olfactory, motor, and auditory
system.

In the midbrain, the first functional description
of neural oscillations in electro-physiological
recordings in the auditory system was by Langner
(1978), which led to a model of auditory temporal
processing and neural oscillators (Langner,
1981).

Later, sensory segmentation with
coupled neural oscillators were described by van
der Malsburg (von der Malsburg and Schneider,
1986 van der Malsburg, 1992). He linked the
binding problem with neural oscillators. In this
case binding means that oscillations of different
neural ensembles representing different features
of an auditory object like the timbre or the
pitch, synchronize (Cocktail-Party" effect).

Subsequently, in 1990 neural
oscillations became a hot topic in the visual
system. Studies of Gray and Singer (Gray and
Singer, 1989 Gray, 1994), and others (Eckhorn et
al.,1988) associated oscillations in the visual
system with the binding problem.
PhD Thesis, Bahmer 2007
25
Oscillations some historical facts
Oscillations occur in different sensory systems,
like the visual, olfactory, motor, and auditory
system.

In the midbrain, the first functional description
of neural oscillations in electro-physiological
recordings in the auditory system was by Langner
(1978), which led to a model of auditory temporal
processing and neural oscillators (Langner,
1981).

Later, sensory segmentation with
coupled neural oscillators were described by van
der Malsburg (von der Malsburg and Schneider,
1986 van der Malsburg, 1992). He linked the
binding problem with neural oscillators. In this
case binding means that oscillations of different
neural ensembles representing different features
of an auditory object like the timbre or the
pitch, synchronize (Cocktail-Party" effect).

Subsequently, in 1990 neural
oscillations became a hot topic in the visual
system. Studies of Gray and Singer (Gray and
Singer, 1989 Gray, 1994), and others (Eckhorn et
al.,1988) associated oscillations in the visual
system with the binding problem.
PhD Thesis, Bahmer 2007
26
Oscillations in medicine
In medicine, large scale neuronal synchronization
was found to be a reason for epileptical
seizures, which are apparently based on the
mutual excitation between neurons (Traub and
Wong, 1982). Epileptic seizures can be triggered
by various factors. Video screens, including
television, video games, and computer displays,
are the most common environmental triggers of
photosensitive epileptic seizures. Interestingly,
in patients with history of photosensitive
epileptic seizures outbreaks occurred when
certain flashing or patterned images have been
broadcast (Fylan et al., 1999 Zifkin and
Trenite, 2000). It has always been a dream to
interface the brain with a computer in order to
record signals of the brain by a computer and
control brain functions with signals from
outside. In a study of Pesaran et al. (2002)
neural oscillations were suggested as a control
signal because in monkeys oscillations changed
while preparation of movements (see also Andersen
et al., 2004). In the auditory system
computer-brain interfaces have already become
reality in the form of cochlea and brainstem
implants. Cochlea implants stimulate the auditory
nerve in the cochlea with electrical impulses,
brainstem implants are located in the
cochlear nucleus. The implants are still the aim
of research and understanding the role of the
oscillations in the cochlear nucleus might be
important to improve the performance of these
medical aids.
PhD Thesis, Bahmer 2007
27
Oscillations in medicine
In medicine, large scale neuronal synchronization
was found to be a reason for epileptical
seizures, which are apparently based on the
mutual excitation between neurons (Traub and
Wong, 1982). Epileptic seizures can be triggered
by various factors. Video screens, including
television, video games, and computer displays,
are the most common environmental triggers of
photosensitive epileptic seizures. Interestingly,
in patients with history of photosensitive
epileptic seizures outbreaks occurred when
certain flashing or patterned images have been
broadcast (Fylan et al., 1999 Zifkin and
Trenite, 2000). It has always been a dream to
interface the brain with a computer in order to
record signals of the brain by a computer and
control brain functions with signals from
outside. In a study of Pesaran et al. (2002)
neural oscillations were suggested as a control
signal because in monkeys oscillations changed
while preparation of movements (see also Andersen
et al., 2004). In the auditory system
brain-computer interfaces have already become
reality in the form of cochlea and brainstem
implants. Cochlea implants stimulate the auditory
nerve in the cochlea with electrical impulses,
brainstem implants are located in the
cochlear nucleus. The implants are still the aim
of research and understanding the role of the
oscillations in the cochlear nucleus might be
important to improve the performance of these
medical aids.
PhD Thesis, Bahmer 2007
28

• Lets go

29
Chopper neurons a role model for an oscillatory
pacemaker

• Auditory system and temporal processing
• Properties of chopper neurons
• Topology of chopper model
• Results of the simulations
• Unification of broad- and narrow-band analysis
• Hodgkin-Huxley like neurons in the model

30
Hearing system
31
Hearing system
32
Anatomy of the Nucleus cochlearis (CN)
anterior
ventral
dorsal
posterior
auditory nerve
33
Cell types of the Nucleus cochlearis
34
Cell types of the Nucleus cochlearis
35
Cell types of the Nucleus cochlearis
Chopper (Oscillations)
response
time
36
Point plot PSTHTemporal precision of Chopper
neurons
S
37
Temporal precisionSimulation auditory nerve
38
Precision50 times one channel
Ergodic theory (Zeitmittel gleich
Scharmittel) Robustness!
39
The Periodicty modelanatomical motivation
Langner, 1981
40
The Periodicity model function
t
Input periodical signal e.g. from speech
modulation period Pitch
41
The Periodicity model function
t
Input periodical signal e.g. from speech
modulation period Pitch
Modulations- periode Tonhöhe
t
c
carrier period
42
The Periodicity model function
t
Input periodical signal e.g. from speech
t
c
carrier period
Correlation analysis of modulation- and carrier
period
43
The Periodicity model function
t
Neuronal periodicity analysis
Periodicity equation
44
The Periodicity model function
t
45
Simulation correlation analysis
256 combinations, best 61,16 cf, 16 mf
Voutsas et al. 2005
46
ElectrophysiologyTopology of the model
47
Preferred oscillations intervals in the auditory
midbrain
Preference for multiples of 0.4 ms in
oscillations of neurons in the inferior
colliculus of the cat (Langner and Schreiner,
1988)
48
Preferred oscillations intervals in the auditory
midbrain
Preference for multiples of 0.4 ms in
oscillations of neurons in the inferior
colliculus of the cat (Langner and Schreiner,
1988)
49
Interspike intervals (ISIs) of VCN neurons
(Young et al. 1988, cat)
Preferred oscillations intervals in the auditory
brainstem
50
Histogram of the interspike intervals
Bahmer and Langner I, Biol Cybern, 2006
Number of intervals
Interspike intervals ms/0.4
The preference for multiples of 0.4 ms is
statistically significant !
51
Anatomical evidences
Ferragamo et al. 1998
52
Circular topologywith constant synaptic delay
0.4 ms
0.4 ms
0.4 ms
0.4 ms
0.4 ms
53
Physiological propertiesof chopper neurons
PSTH Tuning Periodicity coding
not comparable to nerve fibers
comparable to nerve fibers
highly regular
54
Tuning vs. periodicity encoding at low intensity
55
Tuning vs. periodicity encodingat high intensity

56
Tuning vs. periodicity encodingat high intensity

57
Anatomical properties of chopper neurons
Narrow band input from the cochlea via 5 synapses
(Ferragamo et al. 1998)
58
Anatomical properties of chopper neurons
Narrow band input from the cochlea via 5 synapses
Choppers are interconnected
(Ferragamo et al. 1998)
59
Bahmer and Langner , Proc. ISH, 2007
Topology
(Ferragamo et al. 1998)
Long intervals require an unrealistic high number
of neurons
60
Topology of an improved model
pace maker slave
61
Technical details of the simulation
Model of inner outer ear Mex-file
Onset neuron Mex-file
Chopper neurons Script-file

• HH-like membrane model
• gaussian shaped
• integration of broad band input from ANFs.

• Wave-digital filter model consists of
• Second order resonators coupled by fluid mass
  (Strube 1985)
• Model of inner hair cells

• LIF with integrating synapses
• Input from onset neuron
• Input from ANF

Implemented by W. Hemmert (Infineon Technologies)
Bahmer and Langner I, Biol Cybern, 2006 Bahmer
and Langner II, Biol Cybern, 2006
MATLAB
62
Chopper responses
Simulation I
µ
Bahmer and Langner II, Biol Cybern, 2006
Physiological data
Blackburn and Sachs 1989
63
Chopper responses
Simulation II
Bahmer and Langner II, Biol Cybern, 2006
Physiological data
Blackburn and Sachs 1989
64
Synchronization ofsimulated choppers
Bahmer and Langner II, Biol Cybern, 2006
65
Synchronization ofsimulated choppers
Bahmer and Langner II, Biol Cybern, 2006
with onset
narrow integration
without onset
66
Interspike intervals of simulated
choppers
67
Representation auf resolved harmonics in the model
Bahmer and Langner , 2008 subm.
68
Representation auf resolved harmonics in the model
broadband integration
narrowband integration
norm. Spikerate
norm. Spikerate
frequency Hz
frequency Hz
Frequency resolution
Frequency resolution
69
Representation of periodicity in the model
Bahmer and Langner , 2008 subm.
High level
periodicity
integration width
70
Adaptation-problem at high levels
Frequency resolution
Integration
Periodicity coding
71
Adaptation-problem at high levels
Frequency resolution
Integration
Periodicity coding
72
Adaptation-problem at high levels
Frequency resolution
Integration
Periodicity coding
?
or
73
Adaptation-problem at high levels
Adaptation of integrationswidths of
,Onset-neuron
or
could explain the pitch ambiguity found in
psychoacoustic tests (Schneider et al., 2005).
Important role for signal processing in the
entire auditory system (brainstem effect)
74
Interested in Hodgkin-Huxley?
75
Hodgkin-Huxley Model Channel Kinetics
76
Rothman and Manis Model VCNLow Threshold Channel
77
The truthor fact and fancy
Rothman and Manis, 2003b
78
It is so hot !
K Channel VCN Cao and Oertel, 2005
79
Real curve fitting expression
Differencies between fit and real curve
(flatline)
Rothman and Manis, 2003
80
Tribute I to Prof. Donata Oertel
Dear Andi,How wonderful that you take my
comments with such wonderful spirit.  I guess it
is your Bavarian background that makes you see
things positively!  I, too, enjoy intellectual
arguments!
To a biologist like me, the functional
consequences of a model of a neuronal circuit
that does not exist is like dreaming and is not
terribly interesting or meaningful.  To me the
value of a model lies in summarizing the
functional consequences of experimental
observations that are too complicated or too
numerous to imagine.  I do hope that I have not
offended you by being too critical Professor
Donata Oertel, Wisconson
Sorry, not for me
81
Tribute II to Prof. Donata Oertel
The results of Cao and Oertel (2005) are in some
ways contradicting to the findings of Rothman and
Manis (2003b) concerning the fittings of the
conductances. It was stated (Oertel, D., personal
communication) It is not always easy to decide
how to fit exponentials to currents. People
usually try fitting a single exponential first.
If the fit is terrible, they go to two
exponentials. The fit is invariably better but is
it good enough? Those are to some extent
arbitrary judgments. Professor Donata Oertel,
Wisconson in PhD Thesis Bahmer 2007
Real data are the truth?
82
Simulations with ,Rothman-Modell
83
One cell in NEURON
84
Two cells in NEURON
85
The network stabilizes interspike intervals (ISI)
and spikerate
The network stabilizes the ISI
Too slow (for me) -gt modification
86
,Rothman-Model
Two important channels
87
,Rothman-Model
Factors l and k (Langrange)
88
,Rothman-Model
Optimizing k Temperature change
89
Optimization with genetic algorithms
Bahmer and Langner , Biol Cybern 2008, in prep.
Changes in the capacity (cell diameter) are in a
physiological range. Changes in the time
constants are in a physiological range.
90
Accelerated model in network of two neurons
Bahmer and Langner , Biol Cybern 2008, in prep.
Input
PSC
intervals
AP
time ms
0.8 ms
Zeit ms
91
Multi-Oscillator
Bahmer and Langner , Proc. ISH 2007
t
Pacemaker Slow neuron
nt
92
Simulation of Multi-Oscillator
93
ModelDB

• Where can we collect and share data?
• ModelDB Collection of Neuronal Models
  (University of Yale)
• Part of NEURON project

94
(No Transcript)
95
(No Transcript)
96
Final Tribute to Prof. Donata Oertel
Yes, cochlear implants are marvelous for many
reasons.  It is a very successful interface with
the brain that has caused a revolution in the way
deaf children can learn language and function in
the general society.  It is also remarkable for
what it says about the nervous system. The
stimulation strategies are so primitive and yet
many people do so very well with them. It is by
thinking about cochlear implants that I have come
to think that choppers are so important for
understanding speech. This is great fun. 
Thank you! Professor D. Oertel, Wisconson
97
Auditory Implants
Cochlear Implants
Colliculus Inf. Implants
Brainstem Implants
98
Cochlear implant
99
One problemDamage of Serve and volley
nerve fibers
gt 1ms
spikes
Projection
lt 1ms

• Nerve can temporally encode up to 5 kHz
  (Volley principle, statistical but phase
  locked).
• Entrainment of nerve by electrode stops Volley
  principle.

100
Deep Brain Stimulation
Find oscillations in electrophysiology in vivo!

• Electrode is inserted in deep brain regions
• Therapy for Parkinson, Epilepsy, and Depression.

101
Difficulties in Clinical research
Clinic
Doctor
Research
102
Neuroimplant Research
Basic research
Medical technic
Neuroimplant
Knowledge Transfer
103
Neuroimplant Clinic
Neuroimplant
Diagnosis
Implantation
Rehabilitation
Screening
104
Research Clinic
Basic research
Medical technic
Neuroimplant
Diagnosis
Implantation
Rehabilitation
Screening
Knowledge Transfer
105

• Dissertation
• Prof. Dr. rer. nat. Langner, AG Langner, Prof.
  Dr. med. Galuske
• Prof. Dr. Ing. W. Hemmert und Dipl. Ing. M.
  Holmberg, Infineon Technologies
• Prof. Dr. Ing. Adamy Dr. Ing. Voutsas, Inst. f.
  Robotik, TU Darmstadt
• Oscillating Discussions
• Dr. Raul Muresan, MPI Brain Research, FIAS,
  Coneural Romania

• Cochlea Implantate
• Prof. Dr. Ing. U. Baumann, Prof. Gstöttner, Dr.
  med. Silke Helbig, Dipl.-Ing. T. Rader,
  Otolaryngology University Frankfurt
• Prof. Dr. med. Klinke, Prof. Smolders, Dr.
  Hartmann, Dr. Susanne Braun, Neurophysiology
  Frankfurt
• Firma Med-El, Innsbruck/Starnberg
• Deep Brain Stimulation
• Prof. Dr. Dr. Tass, Forschungszentrum Jülich
• Prof. Steinmetz, University Clinic Frankfurt,
  Neurology

106
Artikel Referenzen

• University of Yale, Model by Bahmer in ModelDB
  http//senselab.med.yale.edu/modeldb/SearchByAutho
  r.asp?authorStrbahmer
• A. Bahmer and G. Langner, Oscillating neurons in
  the cochlear nucleus I. Experimental basis of a
  simulation paradigm, Biological Cybernetics,
  95371-379, 2006
• A. Bahmer and G. Langner, Oscillating neurons in
  the cochlear nucleus II. Simulation results,
  Biological Cybernetics, 95381-392, 2006
• A. Bahmer and G. Langner, Simulation of
  oscillating neurons in the cochlear nucleus a
  possible role for neural nets, onset cells, and
  synaptic delays, Hearing - from basic research to
  applications, Springer, Heidelberg, 2007
• A. Bahmer and G. Langer, A simulation of chopper
  neurons in the cochlear nucleus with wideband
  input from onset neurons, Biol Cyber 2007
  submitted
• A. Bahmer and G. Langner, Networks of
  Hodgkin-Huxley-like neuron models for the
  simulation of oscillating neurons in the cochlear
  nucleus, in preparation.
• A. Bahmer and G. Langner, Oscillating neurons in
  the cochlear nucleus Experimental evidences for
  a new simulation topology, simulation results,
  and consequences for pitch perception Proceedings
  31st Göttingen Neurobiology Conference, German
  Neuroscience Society
• A. Bahmer and G. Langner, Modeling intrinsic
  oscillations in the auditory system a neuronal
  mechanism for quantal pitch shifts and absolute
  pitch? Annals New York Academy of Sciences, Vol.
  1060, 2005.
• http//home.arcor.de/a.bahmer/

107
Question to you ?

• What is the detector of synchronized
  oscillations?
• How can we filter relevant information?
• In the simulation, a slight parameter change can
  lead to a completely different result.
• How will theoretical Neuroscience proceed?
• Your Daisy architecture, your GABA group (Global
  appr.)

108
HH-like Rothman und Manis Modell
109
EEG-type BERA
It is your brainstem, not an Alien!
Moller, 2006