Hi Christian,
Thank you for the beta version. I still seem to be running into a segmentation issue (though this version does give output that suggests that there is not sufficient difference in the global intensity values).
I am happy to play with the segmentation settings since I am going to be processing more atypical brains, but I was wondering if you knew what setting I might be able to change to help me with this brain.
CAT12 and SPM seem to be having a hard time identifying the skull as well. ARPG leaves most of the skull there, GCUT Removes most of the skull but also some brain, and I think I remember the SPM approach also not working.
Would changing the SPM processing accuracy, APP, and/or LAS potentially improve the segmentation? I saw that the default APP can fail with neonate brains and, while this is not a neonate brain, there might be a reduction in intensity difference between WM/GM that isn't represented visually in the display.
I've pasted the log from when it completed running with skullstrip set to gcut (but the stripping and segmentation were innacurate), in case the reference helps.
Thank you, again!
Bethany
------------------------------------------------------------------------
CAT12.7-Beta r1589: 1/1: ./segmentation/CAT12RUN/rT1W.nii
------------------------------------------------------------------------
SANLM denoising (medium): 8s
APP: Rough bias correction:
Initialize 3s
Estimate background 1s
Initial correction 3s
Refine background 2s
Final correction 2s
Final scaling 3s
18s
Affine registration 17s
Affine registration 6s
SPM preprocessing 1 (estimate 1): 70s
SPM preprocessing 1 (estimate 2): 47s
SPM preprocessing 2 (write):
Write Segmentation 10s
Update Segmentation 19s
Update Skull-Stripping 8s
Update probability maps 7s
44s
Global intensity correction:
WARNING: The contrast between different tissues is relative low!
(BG=694.45, CSF=17678.73, GM=17718.23, WM=27278.31)
8s
SANLM denoising after intensity normalization (medium): 3s
Fast registration 38s
Local adaptive segmentation (LASstr=0.50):
Prepare maps 2s
Prepare partitions 1s
Prepare segments (LASmod = 1.00) 5s
Estimate local tissue thresholds (WM) 7s
Estimate local tissue thresholds (GM) 10s
Estimate local tissue thresholds (CSF/BG) 2s
Intensity transformation 26s
SANLM denoising after LAS (medium) 11s
11s
ROI segmentation (partitioning):
Atlas -> subject space 3s
Major structures 2s
Ventricle detection 7s
Blood vessel detection 6s
WMH detection (WMHCstr=0.50 > WMHCstr'=0.50) 12s
Manual stroke lesion detection 0s
Closing of deep structures 1s
Side alignment 1s
Final corrections 1s
33s
Blood vessel correction (BVCstr=0.50): 0s
Skull-stripping using graph-cut (gcutstr=0.50):
WM initialisation 1s
GM region growing 1s
GM-CSF region growing 2s
CSF region growing 1s
Ventricle filling 1s
5s
Amap using initial SPM12 segmentations (MRF filter strength 0.07): 34s
AMAP peaks: [CSF,GM,WM] = [0.57±0.19,NaN±NaN,1.00±0.04]
Final cleanup (gcutstr=0.25):
Level 1 cleanup (ROI estimation) 1s
Level 1 cleanup (brain masking) 1s
Level 2 cleanup (CSF correction) 0s
Level 3 cleanup (CSF/WM PVE) 0s
3s
Optimized Shooting registration with 2.50:-0.25:1.50 mm (regstr=0.50):
Template: "/home/bsussman/Documents/MATLAB/toolboxes/spm12/toolbox/cat12/templates_volumes/Template_0_IXI555_MNI152_GS.nii"
1 | 2.50 | 0.2066 0.0000 0.2066
2 | 2.50 | 0.1902 0.0043 0.1946
3 | 2.50 | 0.1818 0.0081 0.1898
4 | 2.50 | 0.1748 0.0108 0.1856
5 | 2.50 | 0.1678 0.0133 0.1810
6 | 2.50 | 0.1604 0.0157 0.1761
7 | 2.50 | 0.1510 0.0182 0.1691
8 | 2.50 | 0.1424 0.0206 0.1630
9 | 2.50 | 0.1336 0.0229 0.1565
10 | 2.50 | 0.1247 0.0252 0.1499
11 | 2.50 | 0.1159 0.0274 0.1433
12 | 2.50 | 0.1076 0.0294 0.1370
13 | 2.50 | 0.0998 0.0312 0.1310
14 | 2.50 | 0.0925 0.0329 0.1254
15 | 2.50 | 0.0835 0.0344 0.1180
16 | 2.25 | 0.1004 0.1041 0.2045
17 | 2.25 | 0.0857 0.0453 0.1310
18 | 2.25 | 0.0780 0.0454 0.1234
19 | 2.25 | 0.0726 0.0466 0.1192
20 | 2.25 | 0.0683 0.0476 0.1159
21 | 2.25 | 0.0648 0.0484 0.1132
22 | 2.25 | 0.0618 0.0491 0.1110
30 | 2.00 | 0.0646 0.1378 0.2024
31 | 2.00 | 0.0566 0.0537 0.1103
32 | 2.00 | 0.0530 0.0513 0.1043
33 | 2.00 | 0.0515 0.0516 0.1031
44 | 1.75 | 0.0531 0.1342 0.1873
45 | 1.75 | 0.0491 0.0574 0.1065
46 | 1.75 | 0.0479 0.0541 0.1019
58 | 1.50 | 0.0565 0.1474 0.2040
59 | 1.50 | 0.0497 0.0622 0.1118
60 | 1.50 | 0.0475 0.0591 0.1065
Shooting registration with 2.50:-0.25:1.50 mm takes: 189s
Prepare output 5s
194s
Jacobian determinant (RMS): 0.035 0.973 1.570 1.660 1.883 | 1.918047
Template Matching: 0.620 0.186 0.154 0.144 0.142 | 0.142432
Write result maps: 29s
Surface and thickness estimation:
lh:
Thickness estimation (0.50 mm³):
WM distance: 29s
CSF distance: 14s
PBT2x thickness: 8s
59s
Create initial surface 20s
Topology correction and surface refinement: 37s
Correction of central surface in highly folded areas 6s
Refine central surface 53s
Correction of central surface in highly folded areas 2 3s
Spherical mapping with areal smoothing 81s
Spherical registration 269s
Euler char. / def. number / def. size: -98 / 26 / 9.47%
rh:
Thickness estimation (0.50 mm³):
WM distance: 38s
CSF distance: 15s
PBT2x thickness: 8s
70s
Create initial surface 22s
Topology correction and surface refinement: 35s
Correction of central surface in highly folded areas 5s
Refine central surface 52s
Correction of central surface in highly folded areas 2 2s
Spherical mapping with areal smoothing 83s
Spherical registration 273s
Euler char. / def. number / def. size: -74 / 33 / 2.65%
Final surface processing results:
Average thickness: 2.5523 ± 0.8768 mm
Surface intensity / position RMSE: 0.0903 / 0.1286
Euler char. / def. number / def. size: 90.0 / 29.5 / 6.06%
Display thickness: <a href="matlab:cat_surf_display('/home/bsussman/Documents/MATLAB/research/PCH/SUBJECTS/DMD/HE/segmentation/CAT12RUN/surf/lh.thickness.rT1W');">/home/bsussman/Documents/MATLAB/research/PCH/SUBJECTS/DMD/HE/segmentation/CAT12RUN/surf/lh.thickness.rT1W</a>
Display thickness: <a href="matlab:cat_surf_display('/home/bsussman/Documents/MATLAB/research/PCH/SUBJECTS/DMD/HE/segmentation/CAT12RUN/surf/rh.thickness.rT1W');">/home/bsussman/Documents/MATLAB/research/PCH/SUBJECTS/DMD/HE/segmentation/CAT12RUN/surf/rh.thickness.rT1W</a>
Surface and thickness estimation: 1383s
ROI estimation:
Data mapping to normalized atlas space 26s
ROI estimation of 'cobra' atlas 13s
Data mapping to normalized atlas space 10s
ROI estimation of 'neuromorphometrics' atlas 12s
ROI estimation of 'lpba40' atlas 10s
ROI estimation of 'hammers' atlas 11s
ROI estimation of 'ibsr' atlas 10s
ROI estimation of 'mori' atlas 12s
ROI estimation of 'anatomy' atlas 1s
105s
Quality check: 2s
Developer display mode!
Print 'Graphics' figure to:
/home/bsussman/Documents/MATLAB/research/PCH/SUBJECTS/DMD/HE/segmentation/CAT12RUN/report/catreport_rT1W.pdf
------------------------------------------------------------------------
CAT preprocessing takes 30 minute(s) and 44 second(s).
Image Quality Rating (IQR): 83.17% (B-)
Segmentations are saved in /home/bsussman/Documents/MATLAB/research/PCH/SUBJECTS/DMD/HE/segmentation/CAT12RUN/mri
Reports are saved in /home/bsussman/Documents/MATLAB/research/PCH/SUBJECTS/DMD/HE/segmentation/CAT12RUN/report
Labels are saved in /home/bsussman/Documents/MATLAB/research/PCH/SUBJECTS/DMD/HE/segmentation/CAT12RUN/label
------------------------------------------------------------------------
|