When performing tract-based analysis, I met a problem that I couldn’t find the option to set the minimum threshold of streamline number such as 5 streamlines. Or, is it only accessible when the track is transformed into fixel-wise?
Another question, do I need to set the threshold of FOD amplitude again when transforming the tract into fixel-wise (though I have set cut-off FOD amplitude = 0.1 during the step of generating whole-brain tractogram)?
I read the method from this paper: doi: 10.1093/cercor/bhy348
“we extracted these ﬁber bundles individually from the template tractogram…The streamlines corresponding to these bundles of interest were then concatenated together and used to generate a ﬁxel analysis mask that included only those ﬁxels through which at least 5 streamlines passed and that had an FOD amplitude of at least 0.1…”
Can you help me to make through this step?
I think what you’re looking for is a fixel-wise analysis mask, where only those streamlines containing at least 5 streamlines are present in the mask. The number of streamlines attributed to each fixel can be generated using the
tck2fixel command; the resulting fixel data file can be turned into a mask based on a threshold using the
The FOD amplitude can actually be thought of as being applied in two different ways. Initially, a peak amplitude threshold is applied during the
fod2fixel stage (when run on the FOD template image), since FOD lobes with a peak amplitude below that threshold will be omitted from the fixel template. However it’s also possible to subsequently apply an FOD peak amplitude threshold when generating a fixel processing mask. If, during the
fod2fixel step, you export a fixel data file containing the peak amplitude of each fixel (using the
-peak_amp option), you can then apply a threshold to that fixel data file that may be different to the one used to exclude fixels from the template, and use that data in production of the fixel analysis mask (e.g. by multiplying the resulting mask with the one derived from the fixel streamline count data). The difference is subtle, but might assist in understanding the wording of that reference.