Thanks for your suggestion.
One correction on the description on what I did.
I generated ROI in the mean_FA_skeleton image. This ROI does not include zero value. And the ROI was not generated by manual drawing. It was generated using threshold (not less than 0.2). In addtion, as far as I understand, FA_skeleton represents a common tract center in all FA images processed. And the tract would not have zero FA value because non-zero threshold option (eg. 0.2) was used. So FA_skeleton of each time points (in my data for longitudinal study) would not have zero value if ROI was drawn using mean_FA_skeleton.
The processing what I intended was, utilizing TBSS process to generated a well registered longitudinal data set and get a set of skeletonised longitudinal FA maps of an individual subject.
I thought, if a FA_skeleton was generated for each time point (in my longitudinal data set), then the FA_skeleton would have representative FA values for major tracts of the subject. After pre-processing, my processed data would be a single 4D longitudinal FA_skeletonised images containing data of each time points . Hence, I would be able to compare data for each time point systematically.
Projection of FA onto mean_FA_skeleton would generate FA_skeleton for each time point and later it will be merged to be a all_FA_skeletonised. Right? As far as I know, max FA value would be searched and would be projected on to the skeleton. If the FA_skeleton contained max FA value, how the FA value cannot be lower than FA values acquired using typical ROI work. That was my initial question and I am still confused looking at my data.
Probably, I do not fully understand FA projection process. Is there any one who has clear explanation on the FA projection onto FA_skeleton?
Thanks in advance,
SC
On Fri, Mar 30, 2012 at 4:40 PM, Kirstie Whitaker
<[log in to unmask]> wrote:
I believe the all_FA_skeletonized dataset is also 4 dimensional with the 4th dimension representing the number of subjects (or scan sessions in the case of longitudinal data).
Here's one possible suggestion for the differences in your ROI values - are you sure you aren't including zeros when you calculate the mean FA values from the skeletonized data? There's a difference between fslstats <file> -m and fslstats <file> -M that wouldn't show up in your ROI for the non-skeletonized data because there are (probably) no zero values in the middle of the corpus callosum!
KxOn Fri, Mar 30, 2012 at 12:11 PM, Christine Zakrzewski
<[log in to unmask]> wrote:
The all_FA file represents a 4 dimensional dataset (x, y, z, time). The value at any one voxel would represent the FA in that particular location, at that point along the time-series. The mean FA file collapses the dataset across time to produce a dataset with a single value at each voxel (the mean value at that location across the time-series), whereas all FA skeletonised collapses datasets across sessions to again produce a datset with a single value at each voxel (the average mean value at each voxel across sessions). I dont think I would worry too much about the differences you observe.
Chrissy
Date: Fri, 30 Mar 2012 14:52:08 -0400
From:
[log in to unmask]Subject: Re: [FSL] FA values in TBSS results
To:
[log in to unmask]Thanks for your reply, Chrissy.
I was unclear about this. I am conducting a test on a longitudinal data from a single subject.
I ran a TBSS process on a set of longitudinal data from subject.
all_FA, all_FA_skeletonised, mean_FA were generated.
I draw a ROI on the mean_FA map.
Then applied the ROI to get FA values from one of the all_FA images and corresponding all_FA_skeletonised images.
What I call typical FA means the FA values from one of the all_FA images.
I hope I made my point on this question.
Thanks,
SC
On Fri, Mar 30, 2012 at 2:41 PM, Christine Zakrzewski
<[log in to unmask]> wrote:
The skeletonised FA map represents a mean of the FA tracts across all individual subjects in a particular group. I do not know what you mean by typical FA values (single subject FA vaules?), but this may account for the differences you are observing.
Hope this helps.
Chrissy
> Date: Fri, 30 Mar 2012 16:27:20 +0100
> From:
[log in to unmask]> Subject: [FSL] FA values in TBSS results
> To:
[log in to unmask]
>
> Dear FSL users,
>
>
> I am conducting FA value comparison between skeletonised FA and typical FA values.
>
> Typical FA values acquired from non-skeletonised FA map using ROI drawing on Corpus Callosum in the midsagittal plane of the mean FA skeleton map.
>
> I expected values between typical FA map and skeletonised FA map are not different drastically.
>
> Interestingly, FA values from skeletonised FA map is approximately 50 % lower than that from typical FA map.
>
> Is there any explanation on this difference?
>
>
> Sincerely,
>
> SC
It's that time again! I'm riding from San Francisco to LA with AIDS/Lifecycle for the fourth time in June 2012.
Please donate anything you can to help me reach my goal: http://www.tofighthiv.org/goto/kirstie
-----------------------
Kirstie Whitaker
Doctoral Candidate
Bunge Laboratory: Building Blocks of Cognition
Helen Wills Neuroscience Institute
University of California at Berkeley
134 Barker Hall, MC 3190
Berkeley, CA, 94707
tel: 510 684 2456
web: bungelab.berkeley.edu
