External Email - Use Caution
Dear Freesurfer community,
I computed T1 maps from Siemens VIBE (a spoiled 3D GRE variant; instead of Siemens FLASH (RF-Spoiled GRE)). I utilized mri_ms_fitparms to compute the T1 map (which looks ok) and compared the result to a T1 map (inline) from MP2RAGE. However, T1 values based on Siemens VIBE (FA: 4, 8, 13, 21, 34) are much lower than from MP2RAGE:
(results from the same subject, same day/position/setup scan, same ROIs)
WM/Corpus callosum: 652 ms (VIBE) vs 859 ms (MP2RAGE) Cerebellaer WM: 583 ms vs 834 ms --- GM/Cortex: 845 ms vs 1184 ms Deep/Inner GM: 780 ms vs 920 ms --- Liquor: 2222 or 3121 vs 2758 or 3652 ms Eye-Bulb: 112 vs 588
The literature suggests (mouse brain, 3T): Corpus callosum 1108±9 Internal capsule 913±16 Hippocampus 1310±15 Cortex 1246±28 (DOI: 10.1118/1.2968092)
The literature suggests (rat brain, 3T): WM: 1084 or 1110 ms GM: 1820 or 1470 ms (DOI: 10.1002/mrm.20605)
However, it appears that T1 values can be quite different (Fig. 3 from 10.1002/mrm.21313)
#-----------#
Therefore, I would like to ask:
(1) Is it to be expected that T1 values differ dependent on the method used at same field strength and same scanner/setup?
(2) Considering that mri_ms_fitparms has not option for B1 correction, can I "optimize" the maps in a different manner? Would it make sense to run N4B on the respective VIBE scans (before map computation) or maybe directly on the T1 map?
(3) I would assume that the application of the VIBE sequence should be acceptable. Are there reasons not to believe so?
Thanks in advance
Hi Joe
do you know what the forward model for VIBE is? mri_ms_fitparms uses the FLASH forward model of image formation then does a max likelihood estimation of the T1/T2*/PD using that model. If VIBE is a different model then it certainly won't work.
cheers Bruce
On Sun, 5 Jan 2020, joe vanderlo wrote:
External Email - Use Caution
Dear Freesurfer community,
I computed T1 maps from Siemens VIBE (a spoiled 3D GRE variant; instead of Siemens FLASH (RF-Spoiled GRE)). I utilized mri_ms_fitparms to compute the T1 map (which looks ok) and compared the result to a T1 map (inline) from MP2RAGE. However, T1 values based on Siemens VIBE (FA: 4, 8, 13, 21, 34) are much lower than from MP2RAGE:
(results from the same subject, same day/position/setup scan, same ROIs)
WM/Corpus callosum: 652 ms (VIBE) vs 859 ms (MP2RAGE) Cerebellaer WM: 583 ms vs 834 ms
GM/Cortex: 845 ms vs 1184 ms Deep/Inner GM: 780 ms vs 920 ms
Liquor: 2222 or 3121 vs 2758 or 3652 ms Eye-Bulb: 112 vs 588
The literature suggests (mouse brain, 3T): Corpus callosum 1108±9 Internal capsule 913±16 Hippocampus 1310±15 Cortex 1246±28 (DOI: 10.1118/1.2968092)
The literature suggests (rat brain, 3T): WM: 1084 or 1110 ms GM: 1820 or 1470 ms (DOI: 10.1002/mrm.20605)
However, it appears that T1 values can be quite different (Fig. 3 from 10.1002/mrm.21313)
#-----------#
Therefore, I would like to ask:
(1) Is it to be expected that T1 values differ dependent on the method used at same field strength and same scanner/setup?
(2) Considering that mri_ms_fitparms has not option for B1 correction, can I "optimize" the maps in a different manner? Would it make sense to run N4B on the respective VIBE scans (before map computation) or maybe directly on the T1 map?
(3) I would assume that the application of the VIBE sequence should be acceptable. Are there reasons not to believe so?
Thanks in advance
External Email - Use Caution
Thanks for the reply, Bruce! I've done some research and found the following publication showing how T1 quantification based on the VIBE sequence can be done offline, however, it doesn't seem to be straightforward as hoped: Rapid T1 quantification from high resolution 3D data with model‐based reconstruction 10.1002/mrm.27502
Most publications I found so far use the inline map computation, which I also tried, but got me even more inaccurate T1 values than mri_ms_fitparms (and DESPOT1). Considering the correlation shown above, would you think it is something that can be easily implemented?
On Sun, Jan 5, 2020 at 4:33 PM Bruce Fischl fischl@nmr.mgh.harvard.edu wrote:
Hi Joe
do you know what the forward model for VIBE is? mri_ms_fitparms uses the FLASH forward model of image formation then does a max likelihood estimation of the T1/T2*/PD using that model. If VIBE is a different model then it certainly won't work.
cheers Bruce
On Sun, 5 Jan 2020, joe vanderlo wrote:
External Email - Use CautionDear Freesurfer community,
I computed T1 maps from Siemens VIBE (a spoiled 3D GRE variant; instead
of
Siemens FLASH (RF-Spoiled GRE)). I utilized mri_ms_fitparms to compute
the
T1 map (which looks ok) and compared the result to a T1 map (inline) from MP2RAGE. However, T1 values based on Siemens VIBE (FA: 4, 8, 13, 21, 34) are much lower than from MP2RAGE:
(results from the same subject, same day/position/setup scan, same ROIs)
WM/Corpus callosum: 652 ms (VIBE) vs 859 ms (MP2RAGE) Cerebellaer WM: 583 ms vs 834 ms
GM/Cortex: 845 ms vs 1184 ms Deep/Inner GM: 780 ms vs 920 ms
Liquor: 2222 or 3121 vs 2758 or 3652 ms Eye-Bulb: 112 vs 588
The literature suggests (mouse brain, 3T): Corpus callosum 1108±9 Internal capsule 913±16 Hippocampus 1310±15 Cortex 1246±28 (DOI: 10.1118/1.2968092)
The literature suggests (rat brain, 3T): WM: 1084 or 1110 ms GM: 1820 or 1470 ms (DOI: 10.1002/mrm.20605)
However, it appears that T1 values can be quite different (Fig. 3 from 10.1002/mrm.21313)
#-----------#
Therefore, I would like to ask:
(1) Is it to be expected that T1 values differ dependent on the method
used
at same field strength and same scanner/setup?
(2) Considering that mri_ms_fitparms has not option for B1 correction,
can I
"optimize" the maps in a different manner? Would it make sense to run N4B on the respective VIBE scans (before map computation) or maybe directly on the T1 map?
(3) I would assume that the application of the VIBE sequence should be acceptable. Are there reasons not to believe so?
Thanks in advance
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Bruce Fischl -
joe vanderlo