jefffessler / mirt Goto Github PK

• 3d radial with simple DCF for mri_example_3d
• CI e.g. per https://github.com/mathworks/ci-configuration-examples
• mexPrintf so that mex errors go to console not terminal

MH reports that putting images from J-SENSE recon code output
https://sigpy.readthedocs.io/en/latest/generated/sigpy.mri.app.JsenseRecon.html#sigpy.mri.app.JsenseRecon
into mri_field_map_reg.m causes amplification of pixels outside the head in this line:

Masking the fieldmaps (using ESPIRT masks) addressed the issue.
Fixing this is a low priority because I recommend using this code instead anyway:
https://github.com/ClaireYLin/regularized-field-map-estimation

Thanks for this amazing toolbox!
I encountered a problem, when I tried to use nufft in Octave 7.1.0.

nufft 19: Starting nufft() self test; after a wait, shows small errors
nufft 37: err alf1 0.00553546 best 0.000491849
warning: inline is obsolete; use anonymous functions instead
error: 't' undefined near line 1, column 23
error: called from
    feval>@<anonymous> at line 1 column 6
    feval at line 34 column 10
    subsref at line 44 column 14
    nufft_alpha_kb_fit at line 38 column 12
    nufft_init at line 171 column 31
    nufft_table_init at line 38 column 4
    nufft_init at line 94 column 5
    nufft at line 66 column 8

I was able to track this problem down, to this line:

Just changing inline to handle there, solves the problem for me.

I have just started researching magnetic resonance fingerprinting, which involves creating a spiral trajectory to sample k-space. I don't understand how to do this. I have read relevant papers that mention nufft, but I don't seem to be able to find the inverse process of nufft, which means I don't know how to restore the sampled k-space to the image domain.

After I run setup.m and ir_mex_build.m，I still have some question to run cone-beam-ct_example.m。The fail message is as following:
Fail: Gcone_which_mex 232: cannot find jf_mex or cbct_mex
So where I can get jf_mex or cbct_mex?
Thanks.