[Camino-users] Camino qball reconstruction and tracking

[Neher, Peter]

Hi everyone,

I wanted to use camino to perform ball reconstruction (Descoteaux, not the Aganj CSA version) and probabilistic tracking. Is it normal that the reconstruction takes a VERY long time (hours to days) on a relatively small dataset (FiberCup hardware phantom, 64x64x3). Maybe I did something wrong. Here are the commands I used:

qballmx -schemefile $filename"_SCHEMEFILE.b" -basistype sh -order $shOrder > $filename“_camino_qballMatrix.Bdouble"

sfpicocalibdata -schemefile $filename"_SCHEMEFILE.b" -snr $SNR -infooutputfile $filename"_CalibData.info" > $filename“_CalibData.Bfloat"

linrecon $filename".Bfloat" $filename"_SCHEMEFILE.b" $filename"_camino_qballMatrix.Bdouble" -normalize > $filename“_camino_CalibDataQBall.Bdouble"

sfpeaks -numpds 3 -density 100 -searchradius 1.0 -inputmodel sh -order $shOrder -inputdatatype double < $filename"_camino_CalibDataQBall.Bdouble" > $filename“_camino_CalibDataQBall_PDs.Bdouble"

sflutgen -infofile $filename"_CalibData.info" -outputstem $filename"_camino_QBall_LUT" -pdf watson < $filename“_camino_CalibDataQBall_PDs.Bdouble“

picopdfs -inputmodel pds -numpds 3 -pdf watson -luts $filename"_LUT_oneFibreLineCoeffs.Bdouble" $filename"_LUT_twoFibreLineCoeffs.Bdouble" $filename"_LUT_twoFibreLineCoeffs.Bdouble" < $filename"_camino_CalibDataQBall_PDs.Bdouble" > $filename“_QBallPDsWatsonPDFs.Bdouble“

track -inputmodel pico -pdf watson -numpds 3 -iterations 10 -stepsize $stepSize -interpolator linear -interpolate -seedfile $filename"_SEED.nii" -outputfile $filename"_camino_fibers.Bfloat" < $filename"_QBallPDsWatsonPDFs.Bdouble" -anisthresh $globalFAThres -anisfile $filename"_FA.nii"

I think it gets stuck at sfpeaks. Not sure though if the succeeding commands are much faster.

Best,
Peter