vim-2 data preprocessing / estimation meaning

Hi,

I am working with the vim-2 data and had some questions about it. The data description says- " Peak BOLD responses to each timepoint in the stimuli were estimated from the preprocessed data. "

I was wondering what this line exactly means. In my understanding it has been corrected for motion changes, however has this data been also corrected for the fMRI delay? If yes, then how much of the delay has been taken into account (6s or 4s or something else?). Also, is the delay same for all regions?

Also, what other preprocessing/estimation has been done? Is this just evoked data or is it fitted data to a GLM or an HRF? 

Please let me know as soon as possible.

Thanks,

Sahil

Hi Sahil,

All details about data preprocessing can be found in Nishimoto et al., 2011, where these data were originally used:

http://gallantlab.org/index.php/publications/nishimoto-et-al-2011/

Please feel free to ask if you still have questions after reading the details in the paper.

Best,

Natalia

Hi Natalia,

I am still unclear whether BOLD activations have been deconvolved/time corrected. The original article uses the uncorrected BOLD activations and estimates the HRFs using the images at t-3 to t-6 secs,  but the phrase ''Peak BOLD responses to each timepoint in the stimuli were estimated from the preprocessed data." is quite similar to a sentence in the documentation for vim-1 and that dataset was deconvolved as in Kay et al., 2008, which could have been used for vim-2, too. Plus, my models' fit is only slightly better if I correct for time delays by shifting the training data by some seconds. I'm inclined to believe that it has not been deconvolved, but I am not sure.

Thank you,

Erick

Previously Natalia Bilenko wrote:

Hi Sahil,

All details about data preprocessing can be found in Nishimoto et al., 2011, where these data were originally used:

http://gallantlab.org/index.php/publications/nishimoto-et-al-2011/

Please feel free to ask if you still have questions after reading the details in the paper.

Best,

Natalia

Hi Erick,

The BOLD responses have not been deconvolved. In Nishimoto et al. 2011, the stimuli were convolved with hemodynamic response filters after the motion energy filters to incorporate the hemodynamic response function in model fitting. I would not suggest simply time shifting the stimuli, since the hemodynamic response function spans multiple TRs. You can find the details of the convolution of the stimuli with the hemodynamic response filters in the Supplemental Experimental Procedures section of Nishimoto et al, 2011 under the headings "Motion-Energy Encoding Model" and "Model Fitting".

Hope this helps!

Best,

Natalia