I do not know the answer to this. But I would assume for every registration, you could use a different threshold value, since each mask will be interpolated differently.

Perhaps, if I am incorrect, an expert can comment on this.

Cheers,

Bryson

On Thu, Sep 25, 2014 at 12:57 PM, Alshikho, Mohamad J. <[log in to unmask]> wrote:
Thank you very very much. This is helpful
One  more question. In a list of subjects (two groups) can I use different thresholds in fslmaths i.e every subject has it's own threashold?? or it should be the same value for all the subjects?




From: FSL - FMRIB's Software Library [[log in to unmask]] on behalf of Bryson Dietz [[log in to unmask]]
Sent: Thursday, September 25, 2014 12:44 PM

To: [log in to unmask]
Subject: Re: [FSL] Binarize a mask

0.9 and 0.1 should result in extremely different masks, so you should find very different statistics.

When you visualize the masks in FSLVIEW, which one appears to overlay the white matter more closely?

I am saying for your white matter mask, perhaps try 0.7, 0.75, 0.8, 0.85, 0.9, 0.95 and see which one appears more accurate (i.e. do not rerun your analysis for every mask).

The thresholding is used to re-binarize your mask following the interpolation and blurring from registration.

On Thu, Sep 25, 2014 at 12:36 PM, Alshikho, Mohamad J. <[log in to unmask]> wrote:
Hi Bryson,
The reason why I am not sure about the suitable threshold, because it can change the final results dramatically from significant to non significant. i.e I have 50 subjects and I want to check if there is any significant difference in the mean FA between two groups (23 Patients and 27 controls); I created a mask for the left hemispheric white matter and I registered (FLIRT) this mask to the FA map for every subject then I used fslmaths (after FLIRT) with a threashold 0.9  for all the subjects. I repeated this process for the threashold 0.1 in fslmaths. I used fslstats dti_FA -K <mask> -M for the subjects with 0.9 and then for the subjects with 0.1. Now I have two lists of mean FAs (for the two groups) the first one for fslmaths ( 0.9 ) and the second one for fslmaths (0.1). I ran simple two tailed t.test for the first list (0.9) and for the second list (0.1) . I found that the results are significant in 0.9 and not significant for 0.1!!!

I can't understand this point. The threshold in fslmaths can affect the final results
kindly do you have any interpretation for this difference in the results?
Thanks 
Mohamad 
 


From: FSL - FMRIB's Software Library [[log in to unmask]] on behalf of Bryson Dietz [[log in to unmask]]
Sent: Thursday, September 25, 2014 12:03 PM

To: [log in to unmask]
Subject: Re: [FSL] Binarize a mask

This may not be the correct answer (I have never done this myself), but it will be trial and error. Try several values and find the one that has the best fit. Applying fslmaths only takes a few seconds so I would just try a bunch of values.

Citing the website again: "For example, if the mask is being used to quantify values within an ROI and it is important to not include contamination from surrounding areas, then a high threshold should be used".

The threshold value will differ depending on the interpolation and blurring from registration.

Hope this helps,

Bryson

On Thu, Sep 25, 2014 at 11:56 AM, Alshikho, Mohamad J. <[log in to unmask]> wrote:
Hi Micheal and Bryson,
Thank you very much for your quick response.
In FSLWiki they mentioned the following "The threshold used (with fslmaths) should be set depending on the intended use of the output mask. " Let's say that I want to binarize a mask for the left hemispheric white matter and another mask for the cerebellum and I want to calculate the mean FA using fslstats with the flag K. What is the suitable threshold for my masks here is it 0.9, 0.5 or 0.1? 

Looking forward to learn from you guys

thanks,
Mohamad  

From: FSL - FMRIB's Software Library [[log in to unmask]] on behalf of Michael Dwyer [[log in to unmask]]
Sent: Thursday, September 25, 2014 10:42 AM
To: [log in to unmask]
Subject: Re: [FSL] Binarize a mask

Hi Mohamad,

Take a look at the following entry in the FLIRT FAQ:

http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FLIRT/FAQ#How_do_I_transform_a_mask_with_FLIRT_from_one_space_to_another.3F

On Thu, Sep 25, 2014 at 8:30 AM, Bryson Dietz <[log in to unmask]> wrote:
-thr <value> does not have to be between 0-1. You specify the cut off value i.e. 750 => every pixel below this intensity = 0.

On Wed, Sep 24, 2014 at 10:58 PM, M. alshikho <[log in to unmask]> wrote:
Hi FSL experts,
After FLIRT I am using fslmaths to binarize my mask using the following command line :
fslmaths mask -thr <value> -bin mask
my question is about the threshold value? I know that this value should be between 0-1 but I am wondering how can I choose this threshold correctly? what can control this value up or down between 0-1 ??

Thanks in advance
Mohamad




--
Michael G. Dwyer, Ph.D.
Assistant Professor of Neurology and Biomedical Informatics
Director of Technical Imaging Development
Buffalo Neuroimaging Analysis Center
University at Buffalo
100 High St. Buffalo NY 14203
[log in to unmask]
(716) 859-7065

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.