On Wed, May 22, 2013 at 11:41 PM, Ce Mo <[log in to unmask]> wrote:

Dear Karsten,

Thank you so much for your kind and insightful thoughts.

"As you observed correctly, the movement within a session looks almost okay to me. "

That's good to hear. However, my biggest concern is that monotonically increasing head movement with scanning time, which seems to me that the subject kept moving her head towards a certain direction. As a result, the motion parameter estimates become quite large by the end of each session. This problem also presents in another experimental data of ours. I tried registering the images to the mean image instead of the 1st image but it didn't improve much. Is it a rare and serious problem? If so, is there a way to tackle it?

>>> There is no way to eliminate a unrestrained subject from moving in the scanner. The amount of movement you are seeing is extremely small. In the lab I work in, we use the following criteria: Total movement -- 5mm, Total Rotation -- 5 degrees. Bad timepoints -- .75mm movement or 1.5 degrees rotation (TR-TR difference in parameters). If there are more than 15% bad points, then the run is bad. If the mean motion (TR-TR difference) is greater than 0.2, then the run is bad. Depending on the subject population, you might need to relax the last value slightly.

Aligning to the mean will not change the the overall effect of motion as motion is motion. The idea of registering to the mean is that you will move the data less overall.

"I guess you haven't entered the sessions not in a chronological order, isn't it?"

I entered the sessions in chronological order in accordance with the timing of our data acquisition, i.e. (Run1 Run2 Run 3...) Will there be improvement if I do it differently?

>>> No improvement.

"Was it a block design, you have used?"

We used a fast ER design.

P.S. According to your expertise, what are the acceptible maximum within-session head movements?

>> See above.

Many thanks and Best regards,

Ce

----- 原始邮件 -----
发件人：Karsten Specht <[log in to unmask]>
收件人：[log in to unmask]
主题：Re: [SPM] Weird excessive head movement in realigning multiple sessions
日期：2013年05月23日 00点32分