 |
 |
Dear DCM experts, We would appreciate your advice regarding applying DCM to multi-echo fMRI datasets. There are two particular scenarios in which we would like to apply DCM: (1) resting-state multi-echo fMRI, for analysis with stochastic or spectral DCM (2) task-based multi-echo fMRI, for analysis with deterministic or network discovery DCM We are keen to use a multi-echo sequence as we are collecting data in patient and control populations, and we expect the patients to have more motion artifact. We plan to average the echoes, and then use ICA denoising to identify BOLD and non-BOLD components (e.g. as described here http://www.ncbi.nlm.nih.gov/pubmed/22209809 and here http://www.ncbi.nlm.nih.gov/pubmed/24038744). Our sequence uses a TR of 2.57s with 3 TEs of 15ms, 34ms, and 54ms, on a Siemens 1.5T Avanto. I have two questions in particular: (1) Is it acceptable to average the echoes and apply DCM to the averaged data? (as mentioned previously on the SPM maillist here https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=spm;46e18ff9.1502) If so, would one give the average TE value during model specification? (2) Would you recommend extracting timeseries data for DCM from the data representing only the BOLD components post-ICA denoising, or to extract timeseries from the ‘raw’ data (after averaging the echoes) with non-BOLD components from ICA added in the design matrix as regressors (as one would do with the traditional 6 movement parameters)? Any other advice or points you have regarding application of DCM to multi-echo fMRI would be much appreciated. With many thanks Charlotte --------------------------------------------------------- Dr Charlotte Rae Brighton & Sussex Medical School Sackler Centre for Consciousness Science [log in to unmask] ---------------------------------------------------------