Transcript Papademetras

Papademetras
• Image-guided Navigation Systems
– Structural and Functional images registered to patient
space
• Research workstation is interfaced to
commercial surgical workstation for Epilepsy
surgery
– Allows stable, reliable system for clinical surgery
– VVLink now part of BrainLab system
• Includes ability to track surgical tools
• MODELS OF TUMOR GROWTH ARE CLOSEST
Papademetras
• Brain Deformation is compensated for by
stereo cameras and modeling
• In cases where surgical resection is not
possible, neurostimulators are used (in
functionally sensitive areas)
• How does one leverage prior information
from normal controls
• How do we emphasize robustness over
accuracy?
•
•
•
•
•
•
Coulon
Cortical localization and inter-subject surface matching
For morphometric or functional studies of populations
Volume-based and sphere transformations have been
used in the past
Cortical organization based on “sulcal roots” is an
alternative model
Sulcal landmarks are (semi-automatically?) defined for
tranform to standard space
Application can be for parcellation of brain anatomy and
QUESTIONS: How does one deal with anatomical
variability.
– Can this generalized to maps of brain function (e.g. language
defined in surgical stimulation studies
• Should multiple atlases be developed?
• See: brainvisa.info
Clatz
Modeling brain tumors for patient-specific therapy
• Tumor growth and mass effects are
simulated through computer (linear elastic)
models
• Goal is to use models for patient-specific
treatments
Leiphart
Neurosurgery for brain implants
• Technology for seizure detection
Paradigm shift from exec (localized) center to
[network] of anatomically distinct centers
Synchrony precedes 70% of seizures
Advanced algorithms and computation needed
to predict seizures (shift from electrocorticog)
Reisenhuber
• How does object (face) recognition work?
• Is there an “Alice Cooper neuron” in the
brain?
• Question: How nonlinear is the fMRI/EEG
brain? Is there a functional role for
oscillations…?
Brette
• Electrode introduces bias in single neuron
recording
• Active electrode Compensation is a
computer model to remove this bias.
• Message: Model based approaches can
extend possibilities of standard electrode
recording, for example measuring
excitatory and inhibitory synaptic
conductance.
Schmueli
• Physiological noise (nuisance variables)
can be removed from fMRI
– Heart rate
– Magnetic susceptibility corrections also shown
Marrelec
• How do we infer interactions in fMRI
– Motor network is well understood
– Information exchange between brain units –
definition of functional interactions
• Can this be defined mathematically
• Can causality be inferred from fMRI?
Pessiglione
functional imaging/pharm challenges
• DA involved in reward,validated in monkeys,
mimicked by engineered models
• Behavioral experiment in choice experiment
rewarded by money (or punished by loss)
• L-dopa associated with faster learning than
Haldol
• Model/theory ventral striatum associated with
critic. Anterior putamen
• Are [decisions] subconscious?
• Can brain reading predict decisions?
Gaillard
Epilepsy, functional imaging and language
• Auditory Description Decision Task
(ADDT) fMRI
• Stroke or epilepsy result in relocation of
language function in R hemisphere
• Exploratory data analysis (no a priori
hypothesis) was done in epilepsy pts.
• Posterior margin of Wernicke’s is utilized
• Anterior only group vs. anterior and
posterior moves to contralateral side
Nadja
CNMC Neuroradiology Research
National Children’s Center, Wash DC
• MRI scanners, technologies and neuroradiologists
available for collab.
• Areas of ongoing research
– Craniofacial phenomics, head shape modeling
– Molecular imaging of pediatric brain tumors
• Superparamagnetic nanoparticles for MRI and optical fluorescence
imaging
–
–
–
–
–
Tumor volumetry
Infant MRS and DTI
Fetal MRI and 3D U/S
ASL/Susceptibility weighted imaging in stroke (sickle cell)
Small animal imaging
Michael Levene
• Multiphoton microscopy
– Single molecule sensitivity
• Sub micron resolution
• Two photon excitation, unlike single, has a single focal
point of excitation
• Near IR light scatters less than blue light
• Intrinsic fluorescence of UV is possible too
• Hypometabolism measured in epilepsy with
neurosurgeon
– NADH is fluorescent, NAD is not
• Long lens might be used to penetrate 3.5 mm of tissue
for optical biopsy
• What are clinical applications for MPM and optical imgs?
Olivier David
EEG/fMRI/Epilepsy
•
•
•
•
•
•
Small animal up to 7T
Human 3T scanner
fMRI studies of unilateral epilepsy
Rat model (genetic absence
fMRI of visual stimulation (thalamic relay)
Implanted epileptic patients are used in studies of 1 Hz
neurostimulation of hippocampus
• Antiepileptics studied in animal (rat) models
• Microelectronics developed for neural implants and
adaptive control of epilepsy
• Stimulation of network may be alternative model for TX
– Experimental work needed prior to clinical trials
– Animal studies are vital
Christian Habeck
Multivariate neuroimage analysis
• Univariate stats (SPM) are straightforward parametric
and use stringent type-I error correction
• Multivariate – non-parametric, better suited for
distributed analysis
–
data dredging?
• Covariance analysis of ASL MRI in Alzheimer’s appears
to discriminate
• Multivariate techniques could be used in meta-analyses
• Where does Multivariate analysis show new insight over
univariate analysis?
• Epiphenomena or does Multivariate analysis really
provide insights
Thirion
Group Analysis of fMRI
• Problems with standard inference
– Small samples
– Spatial normalization –precision about 1cm
• Critical questions
– Which structures should be studied?
– How to detect false positives
– Which structures are cross-correlated?
– Question: Build an anatomo-functional atlas
(integrated and fully Bayesian)
– Brainvisa software available soon.
Baillet
Time resolved imaging MEG/EEG
• Atlas of basal ganglia might be used for DBS targeting – plug in for
Papademetras?
• Inverse problem – go from signal to map of brain (function, anatomy)
• What can time perspective bring you for understanding brain
function?
– 400 ms duration map of tennis ball catch
– Similarly conscious perception moves from visual to ventral processing
of information
– Optical flow (borrowed from weather models) can be used to
understand dynamics of brain activity
– State 1 flows to state 2
• Challenges: Reduce theory/computational gap
• Make tools clinically useful tools
• Richard Leahy USC collaborator