[Mrtrix-discussion] streamline numbers at two b-values

[Donald Tournier]

Hi Kerstin,

I've got to say, I'm not sure what's causing these differences. One
possibility is that the sampling is failing too early. That's the part
that selects a new random direction. It's based on rejection sampling,
and by default will stop trying after 50 attempts. There is an option
you can pass to streamtrack called '-trials' - you could trying
increasing that number to 100 or maybe even 1000. You might find that
some of these differences resolve, but using more trials will  slow
down the algorithm somewhat.

Cheers,

Donald.


On Mon, Nov 16, 2009 at 4:42 PM, Kerstin Pannek <k.pannek1 at uq.edu.au> wrote:
> Hi Donald
>
> thanks for your help.
>
> I had a look at the length of the streamlines (see attached histograms).
> I was expecting (maybe naively) to see more short streamlines for the low
> b-value data compared to the high b-value data. I found that the
> distributions look very similar (normalized using the number of accepted
> streamlines), with more short streamlines for the high b-value data.
>
> Is this what you would expect?
>
> Cheers
> Kerstin
>
>
> Donald Tournier wrote:
>>
>> Hi Kerstin,
>>
>> I have to admit, I'm not too sure about that. My best guess is that
>> since the low b-value FODs are (presumably) smoother, more of them
>> start in the 'right' direction, and a greater proportion of them
>> manage to exceed the minimum length threshold. This is just a guess,
>> I'd have to really dig down to give you a definitive answer though...
>>
>> Cheers,
>>
>> Donald.
>>
>>
>> On Fri, Nov 13, 2009 at 3:10 PM, Kerstin Pannek <k.pannek1 at uq.edu.au>
>> wrote:
>>
>>>
>>> Hi all
>>>
>>> I have two DWI datasets acquired per participant, one with b-value 1000,
>>> and
>>> another with b-value 3000. I performed whole brain tractography for both
>>> datasets, seeding 1 streamline per mm3:
>>>
>>> $ streamtrack SD_PROB csd.mif -seed brain_mask.nii -number 103127 -maxnum
>>> 103127 wb.tck
>>>
>>> The number of accepted streamlines is about 30% higher for the b1000 data
>>> than for the b3000 data. This is consistent across all participants.
>>>
>>> I was just wondering why that is?
>>>
>>> Thanks
>>> Kerstin
>>> _______________________________________________
>>> Mrtrix-discussion mailing list
>>> Mrtrix-discussion at www.nitrc.org
>>> http://www.nitrc.org/mailman/listinfo/mrtrix-discussion
>>>
>>>
>>
>>
>>
>>
>
>



-- 
Jacques-Donald Tournier (PhD)
Brain Research Institute, Melbourne, Australia
Tel: +61 (0)3 9496 4078