 |
 |
Hi Ravi, I assume that you are referring to path distributions calculated by probtrackx? In this case, there are two separate issues (which are pictured in the attached document). 1. Sampling issue: to get an estimate of the 'true' path distribution, you have to calculate a sufficient number of samples. There is no obvious way to say what is a sufficient number. Obviously two samples are not enough. In our experience, 5000 samples seems to give a good approximation. In this case, increasing the number of samples shouldn't change your results dramatically. 2. Binning issue: the true path distribution is continuous. But because we only have access to discrete bins (voxels), there is no obvious way to do quantitative measures on it, for various reasons. For example, if you increase your resolution you decrease the probability of hitting any voxel, and if a path goes through the edge of a voxel it is less probable than if it goes through the centre, etc. A proper way to do quantitative measures using path distribution is to use anatomical masks, as in Behrens NatureNeuroscience paper on thalamic segmentation. Cheers, Saad  On 11 Sep 2007, at 06:08, Ravi wrote: > Hi, I've been having some trouble lately trying to figure out what > the thresholding probabilities mean > exactly. From what I understand(2 mask model) if when threshholding > you have use a min of 10000 > and a max of 20000 that would result in a probability of 50% but > this is a relative probability because > you could set the maximum higher or minimum lower. Is there anyway > to get a sort of normalized > probability? I ask this because I want to compare between subjects, > and believe I could just set all the > subjects probability distributions to the same min and max but in > that case how would i determine > what min and max value to set because 1/2 is different then > 10000/20000 even though they would > yield the same relative probability. Thanks. > > -Ravi ------------------------------------------------------------------------ --- Saad Jbabdi, Postdoctoral Research Assistant, Oxford University FMRIB Centre FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK +44 (0) 1865 222545 (fax 222717) [log in to unmask] http://www.fmrib.ox.ac.uk/~saad ------------------------------------------------------------------------ ---