r20200830: Figuring out bad results for MTsat

jcohenadad commented 4 years ago

MTsat data are poorly reproducible:

From: r20200830:

fig_mtsat

The purpose of this issue is to discuss potential causes.

sherbrooke

MTsat=0.000110555005470231.
Data look OK
acq params: -trmt 00645023 -trpd 00645023 -trt1 0015 -famt 9 -fapd 9 -fat1 15 --> ⚠️ should be 64ms (not 645023...).
Getting a R1 values were found to be lower than 0.01. They will be set to inf, producing T1=0 for these voxels. in the log file.

sub-geneva02

MTsat=0.793558347804196
Data look OK
params: -trmt 0035 -trpd 0035 -trt1 0015 -famt 9 -fapd 9 -fat1 15 --> does match the protocol
Getting a R1 values were found to be lower than 0.01. They will be set to inf, producing T1=0 for these voxels. in the log file.

fslAchieva

jcohenadad commented 4 years ago

That's the MTsat map from sub-sherbrooke03 😅 :

jcohenadad commented 4 years ago

So, there is a first identified problem: fetching acquisition parameters from the json file.

Example for sub-sherbrooke, the json is:

    "FlipAngle": 9,
    "EchoTime": 0.002001,
    "RepetitionTime": 0.0645023,

But the extracted param by the code is 00645023 (the comma disappeared).

jcohenadad commented 4 years ago

Updated from r20200904:

fig_mtsat

jcohenadad commented 4 years ago

another strange thing is that there are three spots for dresden, whereas this site only has two subjects 🤨

jcohenadad commented 4 years ago

another strange thing is that there are three spots for dresden, whereas this site only has two subjects 🤨

that was a bug in the job launch script-- now corrected in r20200907:

fig_mtsat

spine-generic / data-multi-subject