That is correct, yes.
Saad
On 31 Jan 2013, at 16:39, Marenco, Stefano (NIH/NIMH) [E] wrote:
> In other words, we cannot use network mode if we want to threshold based on waytotal (which is necessary to compare tract volumes across subjects). Is that the case?
>
> -----Original Message-----
> From: Saad Jbabdi [mailto:[log in to unmask]]
> Sent: Thursday, January 31, 2013 11:35 AM
> To: [log in to unmask]
> Subject: Re: [FSL] probtrackx2 thresholding in network mode
>
> Hi Stefano
>
> No I don't think you can add the two waytotals because these are not samples from the same distribution.
> If you want to calculate the two probabilities separately (and then add them or something), then I'm afraid you'll have to run protbrackx twice to get the separate fdt_paths.
>
> Cheers
> Saad
>
> On 31 Jan 2013, at 16:23, Marenco, Stefano (NIH/NIMH) [E] wrote:
>
>> Saad, the problem is that when we had 2 fdt_paths images we could threshold each one by the corresponding waytotal, but in network mode we have one fdt_paths image so how do we threshold as a percent of the two waytotals? If the output image of network mode is a simple sum of the two back-and-forth runs, then I guess I would add the two waytotals to use as my denominator for thresholding. Is this what you would do? Stefano
>>
>> -----Original Message-----
>> From: Saad Jbabdi [mailto:[log in to unmask]]
>> Sent: Thursday, January 31, 2013 11:05 AM
>> To: [log in to unmask]
>> Subject: Re: [FSL] probtrackx2 thresholding in network mode
>>
>> Hi
>> When you use network mode with two ROIs, probtrackx internally does the same thing as you would do when you were running probtrackx twice, swapping seed and waypoint. The only difference is that it doesn't reload the bedpostx samples. Therefore, it shouldn't make a difference to what you do with the waytotal.
>>
>> Saad
>>
>> On 30 Jan 2013, at 19:32, Anna Slagle wrote:
>>
>>> Hello,
>>>
>>> In the past, we have run tractography twice, once from roi1 to roi2 and once from roi2 to roi1, in seed to target mode. We then used a percentage of each waytotal to threshold the fdt_paths before combining the two directions.
>>>
>>> We are now attempting to use network mode with both roi1 and roi2 as seeds in an effort to simplify (half as many runs, no need to combine them). I have two questions regarding this:
>>> Have people obtained better results using network mode rather than performing two runs in different directions and combining them or are the two methods fairly comparable?
>>> Because the network mode waytotal now includes two values, how do you suggest we go about the thresholding? Do we add the two values and threshold as a percentage of the total?
>>>
>>> Thank you for your help,
>>> Anna
>>>
>>> ____
>>> Anna Slagle
>>> Post-Baccaulaureate IRTA
>>> Unit on Multimodal Imaging Genetics (UMIG) CBDB-GCAP-NIMH-NIH
>>> 10 Center Dr,
>>> Rm. 3C205
>>> Bethesda, MD 20892
>>> Phone: (301) 435-4599
>>>
>>
>> --
>> Saad Jbabdi
>> University of Oxford, FMRIB Centre
>>
>> JR Hospital, Headington, OX3 9DU, UK
>> (+44)1865-222466 (fax 717)
>> www.fmrib.ox.ac.uk/~saad
>>
>
> --
> Saad Jbabdi
> University of Oxford, FMRIB Centre
>
> JR Hospital, Headington, OX3 9DU, UK
> (+44)1865-222466 (fax 717)
> www.fmrib.ox.ac.uk/~saad
>
--
Saad Jbabdi
University of Oxford, FMRIB Centre
JR Hospital, Headington, OX3 9DU, UK
(+44)1865-222466 (fax 717)
www.fmrib.ox.ac.uk/~saad
|