Hi Ettore,
qboot represents all the estimated function using a linear spherical harmonics basis. This is quite standard for ODF/fODF reconstuction. The coefficients represent "weights" for this basis. Therefore, the coefficients give you all the information you need to reconstruct the ODF, since the spherical harmonics are known functions.
At some point I need to provide a script to do that, but for the time being you can have a look e.g. at Sotiropoulos et al, OHBM, 2011 or Aganj et al, MRM, 2010 to get a feeling of what the basis is.
The humber of coefficients reflect the spherical harmonic order that have been requested. The higher the order, the more detail the ODF can capture, but also the more sensitive results are to noise. You get 15 coefficients, because the default harmonic order is 4, a good compromise for standard diffusion datasets.
From your question, it sounds like the information you are after is already provided by qboot, so no need to do anything extra. Qboot obtains samples of ODF shapes for each voxel given the data. Then, it computes the mean ODF shape (of these samples) and finds its peaks (dyads). The relative strength of each dyad is described in the mean_fsamples files. The values in these files represent the value of the mean ODF along the orientation described by the respective dyad. Also, dyads are sorted in descending order, such that dyad1 corresponds to the peak with the highest ODF value.
Hope this helps,
Stam
On 1 Apr 2012, at 16:55, SUBSCRIBE FSL Ettore Accolla wrote:
> Hi,
> I am exploring qboot and I have some very basic questions I would like to better understand, as a user with no great familiarity with the methods.
> I would like to be able to know how much the peaks are represented, per each voxel - or to compare the relative strength of different fibre orientations.
> I run qboot with reconstruction of CSA ODF, and with the option --savemeancoeff.
> As expected I obtain, beside the dyads files corresponding to the peaks, a file called meanSHcoeff.nii.
> From what I can see, this file is a 4D image made of 15 volumes.
> My questions are:
> -could anybody give me an easy explanation of the meaning of these coefficients? can I use them to estimate how much a peak is represented in the ODF per each voxel?
> -what are exactly these 15 volumes referring to?
>
> Thanks a million for your help,
> Ettore
>
|