Meeting 1: Summary and slides
Presentations
Petra: A plan of action
[[1]] This presentation summarizes the key metrics used so far in the litterature to quantify the extend of motion as well as its confounding effect. It also lists:
- all datasets (and associated people) where the problem may exit
- all methods so far put forward to tackle the problem (and associated people)
The aim is to annotate these lists with our findings over the next few weeks.
Simon: Motion artifacts in drug endophenotypes study with170 subjects
[[2]] This presentation repeats the Peterson study on a large dataset related to addiction. The motion problem is found to affect connectivity significantly in the gourp of drug users. Simon's implementation of the Power et al analyses is in MATLAB. Simon also highlighted a series of preprocessing issues which will need to be examined in parallel. Finally, Simon suggested testing a 'soft scrubbing' method based on frame-weighting rather than frame-deletion.
Tim: Wavelet Filtering (I)
[[3]] This presentation summarizes Tim's work on designing a wavelet-based filtering method to denoise the data. Tim used Prantik's %DVARS code as well as R scripts to generate plots such as Fig 1 in Power et al. His code packaged into a single python script (which also executes the R commands) is availablle in the 'Useful Code' [[4]]
Zac: Wavelet Filtering (II)
[[5]] This presentation summarizes Zac's work on designing a wavelet-based filtering method to denoise the data.
Ameera: An improved scrubbing method
This presentation summarizes Ammera's work on designing a modified 'scrubbing' method to denoise the data. Ameera's code is MATLAB based.
John: Spatial maps of DVARS
John suggested creating spatial maps that may reveal whether the DVARS effect is localizd in particular brain regions.
Outcomes
- Ameera and Petra (possibly with Prantik's help) will work on developping a 'diagnostics' package which everyone can simply and efficiently apply to their data to quantify the effects of motion.
- Ed, John and Simon will look further into the addiction data (diagnostics, effect of wavelet decomposition, possibility of using these data as a benchmark).
