Dear DCMers,
We perform pilot studies for a motor connectivity analysis with DCM. Polit subjects performed isometric handgrips left / right for 6s holding period with null events (30-30-30) in a randomized, conterbalanced paradigm. 1st level SPM gives wonderful EoI (F-test) with activations over both primary and secondary motor areas. We'd like to infer connectivity using DCM10.
Model specifcation: 2x4 motor sites bilaterally (=8), DCM include grip right and left (individual conditions). While earlier trails with similar event related paradigms (but without holding period of 6s) showed converging fully connected models after about 30-40 iterations (questionble results), now DCM seems to stop after 128 iterations. However DCM.Ep. with A B C have been created, the results in general look reasonable.
1) What happened here? If I reduce the regions to 6 and alter C, time for one iteration (40s instead of 130s) decreases. Though again, after 128 iterat. DCM finishes without telling "convergence" like before. Has it correctly finished? Are Ep.values final? Can I use them?
2) I know about A (endogenous) B and C matrices. 'A' means the intrinsic connectivity under both grip left and right combined?
3) B means how e.g. grip_left alters A-connections? Would this mean that 'A'+'B_right' would give the actual connectivity while subjects perform a right sided hand grip?
I hope you can help me with these questions. I am looking forward to your answers.
Best wishes
Robert
Dpt. of Neurology
University Center Hamburg Eppendorf
BINS Lab
|