Thanks for your help Martin.

However I don't have just a single subject with that discrepancy. My data has 18 subjects and most of the results obtained with Freesurfer are bigger than FSL.

Thank you again,

Ana

Hi,

this is not a direct comparison but at least indicates similar volumes in FSL-FIRST and Freesurfer.

br J Psychiatry. 2012 Jan;200(1):30-6.

hth, -joost

On Sun, May 27, 2012 at 2:09 PM, Martin Reuter mreuter@nmr.mgh.harvard.eduwrote:

Hi Ana,

Sorry I can't help more. Have never compared fsl and FS directly. Keep in mind that any consistent differences may also depend on the software version. Maybe other people on the list have done some comparisons.

Best Martin

On May 26, 2012, at 13:59, Ana Arruda anisabel.arruda@gmail.com wrote:

...

Thanks for your help Martin.

However I don't have just a single subject with that discrepancy. My

data has 18 subjects and most of the results obtained with Freesurfer are bigger than FSL.

...

Thank you again,

Ana

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer

The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.

Hi Ana

you can check out:

http://surfer.nmr.mgh.harvard.edu/fswiki#Articles.validation

there are a bunch of comparison studies there.

cheers Bruce

On Mon, 28 May 2012, Ana Arruda wrote:

I really appreciate your help. Thank you.

Has anyone on the list ever compared FSL and Freesurfer?

Thanks,

Ana

Hi Ana

it can depend on your manual tracing protocol. For example: are fimbria/alveus part of your definition of hippocampus? For us they are.

cheers Bruce

On Tue, 2 Oct 2012, Ana Arruda wrote:

Dear experts,

About the differences between FSL and Freesurfer, I've noticed that the left-hippocampus volumes measured by Freesurfer are always bigger than the left-hippocampus volumes measured by FSL. Is there any reason that can explain that?

According to the literature, the percent volume difference between Freesurfer and manual tracing for left-hippocampus is always smaller than the percent volume difference between FSL and manual tracing and in my case, it happens exactly the opposite.

Would you mind helping me?

Thank you for your attention,

Ana Arruda

if their definition of what is and is not hippocampus differs from the one we use then you will get different volumes. You need to find out what they defined as part of it and see if it matches our protocol

cheers Bruce

On Tue, 2 Oct 2012, Ana Arruda wrote:

The manual protocol was made by doctors.They traced the hippocampus segmentation based on their knowledge of anatomical structures.

For automatic segmentation I've used the following command: recon-all -i input_suject_name -autorecon1 -autorecon2 -s output_subject_name

and after this simulation I've been working with the values given in the terminal X11 for left-hippocampus and right-hippocampus.

Am I doing anything wrong? Or am I missing something?

maybe the fimbria/alveus is different on the left than the right? Not sure, but if the protocols are different it's very difficult to make any meaningful comparison Bruce On Tue, 2 Oct 2012, Ana Arruda wrote:

That fact can explain the differences between manual and automatic segmentations.  But how can I explain the differences just for the left-hippocampus values in Freesurfer? Comparing percent volume difference between Freesurfer-Manual and FSL-Manual, the percent is bigger in left-hippocampus for Freesurfer-Manual. And according to the literature (Morey et al.) the percent volume difference for Freesurfer-Manual is always smaller than FSL-Manual as you can see in Fig. 5 of that article.

FSL                          Manual                        Freesurfer HD HE HD HE HD HE 3810.741698 3382.601803   3355.5 3007.5 3705 3721 3409.829467 2860.187224 2868 2653.5 3382 3331             3306.409891 3381.885156 3025.5 3001.5 3358 3575 3794.767147 2907.559116 2880 2769 3651 3474 4521.450897 4587.515346 3642 3468 4049 4113 3794.638317 3841.111348 3042 3066 4061 4242

4772.512671 4028.949924 3570 3496.5 4812 4801             4025.037115 3833.11259 3772.5 3513 4094 4224             ? ? 3196.5 2704.5 ? ?             2900.838938 2705.713765 2526 2466 3184 2820 3607.996048 3061.21013 2749.5 2589 3852 3591             3694.401205 3343.946802 2682 2559 3530 3507 4562.195757 4484.215251 3156 3286.5 4114 4599

Hi Ana

yes, our Neuron (2002) and NeuroImage (2004) papers describe the techniques.

cheers Bruce On Wed, 3 Oct 2012, Ana Arruda wrote:

I'm not sure about this, but probably the doctors use the same methods for left-hippocampus and right-hippocampus. The softwares,  FSL and Freesurfer, don't have a different away to operate in either hippocampus. Do you agree? That way of operating confuses me on how to interpret the differences found only for the left hippocampus.

Anyway, thank you for your help.

Another thing, in my thesis I'll have to describe what I've done in Freesurfer to obtain the hippocampus segmentation. Is there any article that describes in a general and brief way  all the steps to be followed towards the final goal?

Thank you all, again.

kind of. There are natural lateralizations in the brain that are taken into account. Not sure what is known about lateralizations of hippocampal subfield and white matter components though Bruce

On Wed, 3 Oct 2012, Ana Arruda wrote:

Thank you very much!

About the previous question, do the softwares use the same criteria for each hippocampus?

Hi Ana

the point is that even if you aren't segmenting individual structures, if some are not in your manual protocol but are in ours, then lateralization in them will cause lateralization differences in the degree of overlap between the two

cheers Bruce

On Wed, 3 Oct 2012, Ana Arruda wrote:

I haven't gone so far. I've just done the simplest way. I've search the proper commands just to do the hippocampus segmentation, but not its subfields. All my analysis was made with the left and right-hippocampus volumes from the aseg.stats file.

Dear experts,

About the previous subject I have one more question.

Based on Morey, et al. article we can say that there is better results for the right hippocampus. Generally, it is known that right hippocampus volume is often bigger than left hippocampus. So, is there any possible explanation for my volumes, measured by Freesurfer, for left-hippocampus be often bigger than the right hippocampus?

Best regards,

Ana Arruda

Can you please take a look to my previous posted hippocampi values? What do you think about them?

Thank you for your prompt help.

Best regards,

Ana Arruda

Hi Darshan

sounds like something pretty big is wrong. If you upload the dataset we'll take a look Bruce

On Tue, 9 Oct 2012, Darshan P wrote:

I have one MRI and I run freesurfer on it . It runs through completely . When I see the output though , the surface does not include a huge part of the temporal lobe. I looked at the brain mask, and the deskulling is perfect. But the wm.mgz does not include temporal regions and hence the surface drops off much before it reaches the temporal lobe. This also affects the pial surface.

I tried to manually add some voxels in the wm.mgz without success. I read some slides which talk about changing the brainmask, but as I said before the brain mask is perfect.

Not sure what else I can do ?

Regards DP