Tuesday, December 6, 2011

data dict

Global var [scannerdata]

  • Attributes of each trial: 
    • [iArtifacts]: if the current trial is considered as a "bad" trial due to artifacts, ∈ (0: good trial; 1: bad)
    • [iTrialType]: trial type {1,2,3,4,-1}
    • [iInstructionLength]: length of the instruction period, ∈(100:1000 ms), source: presentation log 5xxx.
    • [iOffIndex]: the block index number for current trials's light off in eegData02, ∈(1:191)
    • [iOffOffset]: the offset index number for current trials's light off  in each block in eegData02, ∈ (1:12500)
    • [iOnIndex]: the block index number for current trials's light on in eegData02, ∈(1:191)
    • [iOnOffset]: the offset index number for current trials's light on  in each block in eegData02, ∈ (1:12500)
  • Attributes with respect to the raw markers (markerData(1, x).trial & time)
    • [lightOnTimeClean]: clean events of light on events (with double trigger events manually removed)
    • [lightOnCleanIndex]: index of clean events in the originally markers -- gData.markerData(1,1).trial
    • [lightOffTimeClean]: clean events of light off events (with double trigger events manually removed)
    • [lightOffCleanIndex]: index of clean events in the originally markers -- gData.markerData(1,2).trial
  • EEG channel working data
    • [eegData02]: EEG working data. {Column 1: left&Right. Column2: up&down. Column3: info channel}
      • Column3 info channel data format: [condition].[iTrialtype]  e.g. 54.1 = light off for proleft, 52.3 = light on for antileft

Sunday, December 4, 2011

MEGANTI - Preprocessing

[Presentation sce file]
In total 14 blocks, each block has 80 trials, each trial has following four events recorded in the log file:

  1. [Start]: duration 1000 ms, code 0, port_code 1
  2. [Pro] or [Anti] initiation: LightOn, (green pro, red anti) duration 100 ms, code 54, port_code 54, 
  3. [pro5xxx] or [anti5xxx]: randomized length, code: 5+ [100:50:1000], e.g. 5100, 5250..., port_code [101:1:117]
  4. [GO cue ∈ 1 | 2 | 3 | 4] :  duration 20 ms, LightOff
    • 1: greyboxproleft, port_code 201 -- prostim left
    • 2 :greyboxproright, port_code 202  -- prostim right
    • 3: greyboxantileft, port_code 203 -- antistim left
    • 4: greyboxantiright, port_code 204 - -- antistim right
Therefore(in preprocessing_subject.m), the duration of instruction (green or red box) = 100 ms (initiation) + x ms (x is indicated by event2 in matlab code).  The actual difference between light on and off is around 117-119 ms with some trials had more than 135 ms lag (var: timeLag).

SW - [2010-08-11] - Old system no parallel code
raw: S02_MEG54_VEF_02.ds
stim log (raw): sw32-110810-SO2-Pro+Anti-saccade task.log
Problems in the data:
  1. The stim log has 195 missing event condition, all replaced with -1.
  2. The lightOn marker has 130 extra double triggers - fixed (lightOnTimeClean:1121)
  3. The lightOff marker has 3 extra  double triggers - fixed (lightOffTimeClean:1120)
  4. Verify by comparing the time diff between on and off with the event code, 121.40 

Saturday, October 15, 2011

re: Thesis Overview


Supervisor: Dr. JFX DeSouza (link)
MEG Analyses Mentor: Dr. P Ferrari

Research questions

Part 1: Motion perception (fast & slow moving dots)
  • Is there different cortical representation for fast & slow motion?
  • Are there different temporal dynamics for fast & slow motion?
Part 2: Decision-making
  • Subjects need to detect a change in the motion velocity and respond with button press
  • When and where in the brain is a visual perception (velocity change) transformed into the neural signals for action?

  • Prior to the imaging sessions, I conducted a perceptual thresholds test (N=22) to acquire the minimal detectable increase and decrease in motion velocity for the subjects
  • I set up the experimental environment and collected MEG & MRI data (N=12) with the helps from S. Bells and M. Lalancette at Toronto Sickkids hospital.
    • Velocity change in the experiment was set according to each subject's perceptual thresholds, so the correct and incorrect responses could be compared
    • A delayed motor response paradigm was used to separate decision-making signals from motor related signals.



Part 1 - Motion Perception
No difference in the MT+ locations (Talairach) for fast and slow motion was found (Hotelling's T2 for two multivariate independent samples)
Fig - MT+ locations
Fig - comparing the MT+ locations with previous studies

No temporal dynamics for source peak amplitude and latency found
  • Velocity and visual display had no effects (3-way ANOVA)
  • The amplitude and latency btw the three ROIs were different (P<.05)
  • Multiple-comparison showed cuneus had higher amplitude & earlier latency than V3A & MT+

Part 2 - Decision-making
Comparing the grand average event-related beamformer images: 
  • In the correct responses, the frontoparietal sources were observed at various time points after the velocity change onset (left column, perm-test, P<.05)
  • No significant frontoparietal activations were observed in the incorrect responses (right column)

Comparing the time-frequency plots for correct and incorrect responses from IPL and SMC sources:
  • IPL source showed a beta power difference between correct and incorrect responses from 200 - 400 ms (area encircled in dotted line)
  • The initiation of IPL beta ERS was aligned with the high beta ERD increase in SMC after 400 ms (dotted line)

Friday, May 27, 2011

MT+ location

Left MT+

 Right MT+

Problem: # 12 subject

sub#12 has noisy sensor data, but the beamformer reported good MT activation.

(-30    -68    1)

(-35    -66    -3)

Other subjects' RMS can be found here

Scatter plot: outlier sub#12 is in the black box

When sub#12 is excluded :

Individual virtual sensor time course at MT+, please look at sub#12:

Left Fast
Left Slow
Right Fast
 Right Slow


Left Fast

 Left Slow

Right Fast

Right Fast

Sunday, May 22, 2011



Monday, May 16, 2011

Glassbrain movies

[A001] Aligned to [motion stimulus] onset
All movies are made with neurological orientation in Tal coordinates from onset (0 ms) of visual stim to 500 ms after onset.



Left_Fast: weak MT+

Left_Fast: interesting, contraL cuneus -> contraL V3A -> ipsiL MT+ -> contraL MT+


Left_Fast:contraL MT+ then moved to ipsiL MT+

Left_Fast: V3A prior to V1 activation


ST (this subject had no strong V3A activation)



Left_Fast: V3A prior to cuneus

[B002] Group avg for velocity change onset

[A002] Group avg for GO cue onset

[C] Group avg for button press
Incorrect response