Table of Contents
HELP:
DATA: no data.
The purpose of this section is for instance to compare the ROI measurement for both hemispheres following the realignment of the brain by using a symmetric axe:
Draw a ROI on the T1 MRI and export it as a mask to work with a .ima file (image) and not a .arg (graph): roi.ima
Use the AimsMidPlaneAlign command line to realign the image and compute the transformation (superposition of the interhemispheric plane with the plane x=dimX/2).
prompt% AimsMidPlaneAlign -i rmiT1.ima -o align_rmiT1.ima
NOTE: the transformation matrix is located in the input file directory with the following name rmiT1.ima_TO_align.ima.trm.
Do a linear combination if the ROI is a binary image :
prompt% AimsLinearComb -i roi.ima -o linearComb_roi.ima -a 16000
NOTE: please refer to the table in NOTE_2 if the image is not binary.
Resample the ROI with the previously calculated transformation:
prompt% AimsResample -i linearComb_roi.ima -o resample_roi.ima -m rmiT1.ima_TO_align.ima.trm
Perform a threshold to 8000 to preserve a correct volume because the resampling widely extend the symmetric roi volume:
prompt% AimsThreshold -i resample_roi.ima -o threshold_roi.ima -m ge -t 8000
Get the symmetrical ROI by using AimsFlip as follows:
prompt% AimsFlip -i threshold_roi.ima -o sym_roi.ima -m XX
Each ROI can be in both referentials which are T1 and T1_align. In order to change the coordinate system, you apply a .trm. For instance, if you want the ROIs in T1 referential, you must resample the sym_roi.ima with the inverse of T1MRI.ima_TO_align.ima.
prompt% AimsInvertTransformation -i rmiT1.ima_TO_align.ima.trm -o align.ima_TO_rmiT1.ima.trm
Then, resample the sym_roi.ima:
prompt% AimsResample -i sym_roi.ima -o sym_roi_RT1.ima -m align.ima_TO_rmiT1.ima.trm
Analyze/compare the ROIs by using AimsRoiFeatures.
NOTE_1: be aware that the procedure presented below is not formal. In fact, many variations can be processed, the modality (PET, CT ...), how the ROI is obtained (draw on the original referential, or after the realignment) or where it comes from (i.e. created by an other process), what is the type of ROI value (binary, label image ...).
NOTE_2: here is a summary to help you compute and/or do a threshold of your ROI to preserve a correct volume (the resampling leads to volume changes):
Table 9.1. Summary to preserve the ROI volume
| Max value | commande line |
|---|---|
Binary image (max=1) |
prompt% AimsLinearComb -i roi.ima -o roi.ima -a 16000 prompt% AimsResample -i roi.ima -o roi.ima -m motion.trm prompt% AimsThreshold -i roi.ima -o roi.ima -m ge -t 8000 |
Max=max_value | prompt% AimsLinearComb -i roi.ima -o roi.ima -a 16000 -b max_value prompt% AimsResample -i roi.ima -o roi.ima -m motion.trm prompt% AimsThreshold -i roi.ima -o roi.ima -m ge -t 8000 |
NOTE_3: for more information on the algorithm used by AimsMidPlaneAlign, please refer to Prima S, Ourselin S, and Ayache N. Computation of the mid-sagittal plane in 3-D brain images. IEEE Trans Med Imaging. 2002 Feb;21(2):122-38.