Hi. Happy New Year!
On p. 8 of
a directory named AFNI_data2 is referenced, but
find /home/picchionid/afni/CD/ -type d -name ‘AFNI_data2’
yields zero results. I also tried to search for some text in the script on p. 9 with
grep -rli ‘ToolMovie’ ~/afni/CD/
But this also yielded zero results. I also skimmed the slides in
[AFNI Academy] AFNI Time Series Analysis (part 4/8): FMRI Variable-Shape Models
and could not find any additional information. Am I missing something?
Sorry for the confusion. We should probably remove that handout. But if you want access to the data, it is at AFNI_data2.tgz
But it might be preferable to go through the new handout, which corresponds with Bob’s video series. See:
which corresponds with the video series under
or perhaps more easily to use, see it at:
youtube: Time series analysis
Hi! How are you, sir? Thanks for all your afni_proc.py videos. They were very clear and helpful.
I will take your advice and use the latest handout. And thanks for the data link.
Hi again. I actually really like both handouts! afni_TimeSeriesAnalysis.pdf gives helpful didactic details on the experiment, the use of fixed-shape to variable-shape event-related analyses, and the math. afni06_decon.pdf gives a nice hands-on exercise with actual data in a more bite-sized approach and has code that incorporates using afni_proc.py. I for one would love to see one day an update of afni06_decon.pdf and s1.afni_proc.command, a hands-on video to go with them, and the reinclusion of AFNI_data2 in the CD.
I was, in essence, able to recreate the results in terms of the images from p. 19 of afni06_decon.pdf (p. 67 of afni_TimeSeriesAnalysis.pdf), but I am having trouble recreating the graphs from the same page. Can you help me with that? This is didactic, but I am doing it because I will soon have EEG-fMRI data from patients with epilepsy sleeping in the scanner and will be performing an event-related analysis of EEG interictal epileptiform discharges, so I want to be prepared when the data comes in.
Those plots showing multiple estimated iresp (impulse response) plots are done with the Dataset #N plugin. If one such dataset is the underlay, one can load many time series at once. Note that the scaling for the graph window (which you can adjust, of course) is based on the main Underlay dataset, so some datasets are not good for viewing together (like fitts and errts, one hovers around 100, one around 0).
Examples of that can be seen in (at least) a couple of videos.
Linear regression hands-on 5/5, starting at time 3:20
Start to Finish hands-on 15/17, starting at time 17:50
The key is to start with one iresp underlay, and then load other ones using that plugin.
Does that seem like what you want?
Yes! I was able to recreate the graphs from the handouts. Thanks.