Upload
avon
View
43
Download
0
Tags:
Embed Size (px)
DESCRIPTION
Exploring Magnetoencephalography (MEG ) Data Acquisition and Analysis Techniques Rosalia F. Tungaraza , Ph.D. Anthony Kelly, B.A. Ajay Niranjan, M.D., MBA July 22, 2010. What does MEG Measure?. Data Acquisition and Analysis Steps. Acquisition. Coregistration. Preprocessing. Averaging. - PowerPoint PPT Presentation
Citation preview
Exploring Magnetoencephalography (MEG)Data Acquisition and Analysis Techniques
Rosalia F. Tungaraza, Ph.D.
Anthony Kelly, B.A.
Ajay Niranjan, M.D., MBA
July 22, 2010
1
2
What does MEG Measure?
Data Acquisition and Analysis Steps
3
Acquisition
Coregistration
Preprocessing
Averaging
Localization
Data Acquisition and Analysis Steps
4
Fitting isotrack points to structural MRI
Fitting a sphere
Acquisition
Coregistration
Preprocessing
Averaging
Localization
Data Acquisition and Analysis Steps
5
Temporal FilteringBandpass 0.5 – 40Hz
Maxwell’s Filtering
Signal Source Projection (SSP)
Acquisition
Coregistration
Preprocessing
Averaging
Localization
Raw Continuous Data
Time frame: 10000 ms 6
Elekta's Maxwell Filtering Method
7
After Maxwell Filtering
Time frame: 10000 ms 8
Is the Filtering Process Worth it?
9
10
Reduction in Covariance of Magnetometer Signals
Data Acquisition and Analysis Steps
11
Selecting a time region and baseline
Averaging by condition
Acquisition
Coregistration
Preprocessing
Averaging
Localization
Retrieving Trials
12
Baseline: 200ms
1000ms :Time-frame
Median Nerve Stimulation
● Random 10ms long electrical stimulation of same voltage
● Purpose:● localizer● explore effect of number of trials on results
13
Median Nerve Stimulation Average
14180 Trials
Hearing Your Voice in Real Time
Purpose:● explore limitations of MEG (jaw movement, head
motion, activates many simultaneous brain regions e.t.c.) 15
+ house
…+ water
- Press a button- Read word out loud
+ book
Hearing Your Voice in Real Time Stimulation Average
16
- Complex task: elicits response from multiple brain regions- Variations in subject’s response
60 Trials
Data Acquisition and Analysis Steps
17
Dipole Fitting
Acquisition
Coregistration
Preprocessing
Averaging
Localization
18
Source Localization
?
ForwardProblem
InverseProblem
Solvable!
Always anestimate
19
?
Dipole Fitting Technique
1. Pick a subset of sensors (ROI)
2. Select a time point
3. Dipole fit estimates magnetic field given the two parameters
PROCEDURE
Estimated Signals Measured Signals
4.
20
Goodness-of-fit a measure that shows how much the estimated signal matches the measured signal
Confidence Volume the volume within which the dipole fitting method is confident the dipole exists
Measures of Quality
21
22ms 52ms 83ms
7 sensors
42 sensors
92 sensors
99.7% 99.2% 98.2%
84.6% 97.6% 85.8%
84.6% 97.6% 85.8%
Median Nerve Dipole Fitting Results
22
Hearing Your Voice in Real Time Dipole Fitting Results
• Fits dipole in improbable locations (single dipole insufficient)
• Needs a different source localization technique
Brain Stem Post-central GyrusFace Area
61.0% 80.3% 82.2%
23
SummaryWe have explored the following steps in MEG image acquisition and analysis:
• Acquisition of the MEG signals
• Coregistration of MEG isotrack data with structural MRI
• Preprocessing of the MEG signals
• Averaging the preprocessed signals
• Source localization of the averaged waveforms
We found both strengths and weaknesses, which the user must take into account before making inferences from their analyzed data.
Acknowledgements● MEG
– Erika Laing – Anna Haridis– Dr. Ajay Niranjan
● MNTP– Drs. Seong-Gi Kim and Bill Eddy– Tomika Cohen and Rebecca Clark
● All other MNTP participants
24