Thesis Updates
Purpose of Meeting
- Update progress on Aim 1
- Rationale for revisions with Aim 2
- and deletion of Aim 3
- Suggestions/Concerns for revisions
- Goal for Graduation (Summer 2020)
Overall Thematic Changes
- Increased focus on validation of the software (Aim 1)
- More explicit theoretical justification of the software (Aim 1)
- New dataset (Aim 2)
- More straight-forward analysis (Aim 2)
Aim 1:
Create and validate NiBetaSeries to perform beta series correlations
Methods
- Create NiBetaSeries
- Simulate correlations
- Task versus Rest Data
- Task versus "Pseudo Task" Data
NiBetaSeries
- Generates beta maps and creates correlations
- multiple tests to ensure functionality (88% coverage of code)
- documentation (and a tutorial)
- deployed with python and dockerhub
- contributions from around the world
- published in journal of open source software (JOSS)
Simulations
Task versus Rest
- Question: Do beta series correlations detect task condition differences in a cognitive control task?
- 61 older adults (65+)
- Task Switch Task
- ~40 switch trials
- ~40 repeat trials
- Intertrial interval ~ 3.5 seconds
Task versus Rest
Task versus Rest
Task versus Rest
Task versus Rest
-
-
~ 0.1
~ 0.0
Power Analysis: Aim 1
Task versus "Pseudo-Task"
- Question: Are beta series correlations only driven by the magnitude of the BOLD response?
- 61 older adults (65+)
- Task Switch Task
- ~40 switch trials
- ~40 repeat trials
Task versus "Pseudo Task"
Task versus "Pseudo Task":
Model the average response to each task condition for every voxel
Task versus "Pseudo Task"
+
+
+
+
+
+
+
Task versus "Pseudo Task"
Task versus "Pseudo Task" versus Rest: Aim 1
Aim 1: Questions?
Aim 2:
Characterize the relationship between beta series correlations and residual correlations during a flanker task.
Methods
- OpenNeuro ds001751
- Flanker Task
- 4 trial types
- MI-C: 48 trials
- MI-I: 192 trials
- MC-C: 192 trials
- MC-I: 48 trials
- Inter-Trial Interval ~ 3.0 seconds
- 4 trial types
Methods Cont.
- Fronto-Parietal Network
- Beta Series Correlations
- (MI-I) - (MC-I) = negative number
- Residual Correlations
- (MI - MC) ~ 0?
- Beta Series Correlations
- Cingulo-Opercular Network
- Beta Series Correlations
- (MI-I) - (MC-I) ~ 0?
- Residual Correlations
- (MI - MC) = positive number
- Beta Series Correlations
Residual Correlations
Residual Correlations
Task Onsets
Task Beta Estimates
Residual Correlations
-
=
Residual Correlations
Predictions: Aim 2
Quesions: Aim 2?
What about Aim 3?
Next Steps
- Follow up on simulation findings
- create the pseudo data and analyze
- write and submit paper
- "PLOS-ONE"
- "Frontiers in Systems Neuroscience"
- run ds0001751
- through fmriprep
- through nibetaseries
- Write Thesis