Closed 7alberto closed 5 years ago
Will give it a try to see what's wrong. How old is your LINDA/ANTsR installation?
On Wed, Jul 4, 2018, 12:22 PM 7alberto notifications@github.com wrote:
Software versions:
- R (3.4.4)
- ANTsR (0.6.1) [also tried 0.3.2 and 0.4.0]
- randomForest (4.6-12) [also tried 4.6-14]
- LINDA (v.0.2.6)
When sourcing the "linda_predict" script, everything works fine up to the 1st prediction step:
16:47 Starting LINDA v0.2.6 ... Loading required package: ANTsRCore randomForest 4.6-12 Enter file name: T1.nii.gz 16:48 Loading file: T1.nii.gz 16:48 Creating folder: ./linda 16:48 Loading template... 16:48 Skull stripping... (long process) 17:06 Saving skull stripped files 17:06 Loading LINDA model 17:06 Computing asymmetry mask... 17:08 Saving asymmetry mask... 17:08 Running 1st registration... 17:11 Feature calculation... 17:15 Running 1st prediction... Error in which(submask > 0) : argument to 'which' is not logical
It seems that, inside the "mrvnrfs.predict_chunks" function, the data type for the submask cannot be used by the "which" function. I did some debugging, and it turns out that "submask" is actually of type "antsImage", with max = 1 and min = 0:
antsImage Pixel Type : unsigned char Components Per Pixel: 1 Dimensions : 256x256x192 Voxel Spacing : 1x1x0.999995172023773 Origin : 103.2539 -120.9078 -117.9969 Direction : -0.008816802 0.9962615 0.08593764 -0.01123489 -0.08603425 0.9962288 -0.999898 -0.007818053 -0.01195143
I cannot figure out what's wrong here. Any hint? I attach here the T1 example I am currently using.
Thanks a lot, Alberto T1.nii.gz https://github.com/dorianps/LINDA/files/2163838/T1.nii.gz
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/dorianps/LINDA/issues/9, or mute the thread https://github.com/notifications/unsubscribe-auth/AIqafQ5pE3sFD1m3PYiBVXaH97XJI6nXks5uDOuvgaJpZM4VC0up .
Apologies, just noticed you already put the versions.
On Wed, Jul 4, 2018, 1:21 PM Dorian Pustina albnet@gmail.com wrote:
Will give it a try to see what's wrong. How old is your LINDA/ANTsR installation?
On Wed, Jul 4, 2018, 12:22 PM 7alberto notifications@github.com wrote:
Software versions:
- R (3.4.4)
- ANTsR (0.6.1) [also tried 0.3.2 and 0.4.0]
- randomForest (4.6-12) [also tried 4.6-14]
- LINDA (v.0.2.6)
When sourcing the "linda_predict" script, everything works fine up to the 1st prediction step:
16:47 Starting LINDA v0.2.6 ... Loading required package: ANTsRCore randomForest 4.6-12 Enter file name: T1.nii.gz 16:48 Loading file: T1.nii.gz 16:48 Creating folder: ./linda 16:48 Loading template... 16:48 Skull stripping... (long process) 17:06 Saving skull stripped files 17:06 Loading LINDA model 17:06 Computing asymmetry mask... 17:08 Saving asymmetry mask... 17:08 Running 1st registration... 17:11 Feature calculation... 17:15 Running 1st prediction... Error in which(submask > 0) : argument to 'which' is not logical
It seems that, inside the "mrvnrfs.predict_chunks" function, the data type for the submask cannot be used by the "which" function. I did some debugging, and it turns out that "submask" is actually of type "antsImage", with max = 1 and min = 0:
antsImage Pixel Type : unsigned char Components Per Pixel: 1 Dimensions : 256x256x192 Voxel Spacing : 1x1x0.999995172023773 Origin : 103.2539 -120.9078 -117.9969 Direction : -0.008816802 0.9962615 0.08593764 -0.01123489 -0.08603425 0.9962288 -0.999898 -0.007818053 -0.01195143
I cannot figure out what's wrong here. Any hint? I attach here the T1 example I am currently using.
Thanks a lot, Alberto T1.nii.gz https://github.com/dorianps/LINDA/files/2163838/T1.nii.gz
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/dorianps/LINDA/issues/9, or mute the thread https://github.com/notifications/unsubscribe-auth/AIqafQ5pE3sFD1m3PYiBVXaH97XJI6nXks5uDOuvgaJpZM4VC0up .
Alberto, version 0.2.6 might be the problem. I think I fixed this problem in 0.2.7. Get it here https://github.com/dorianps/LINDA/releases/download/0.2.7/LINDA_v0.2.7.zip
On Wed, Jul 4, 2018, 1:34 PM Dorian Pustina albnet@gmail.com wrote:
Apologies, just noticed you already put the versions.
On Wed, Jul 4, 2018, 1:21 PM Dorian Pustina albnet@gmail.com wrote:
Will give it a try to see what's wrong. How old is your LINDA/ANTsR installation?
On Wed, Jul 4, 2018, 12:22 PM 7alberto notifications@github.com wrote:
Software versions:
- R (3.4.4)
- ANTsR (0.6.1) [also tried 0.3.2 and 0.4.0]
- randomForest (4.6-12) [also tried 4.6-14]
- LINDA (v.0.2.6)
When sourcing the "linda_predict" script, everything works fine up to the 1st prediction step:
16:47 Starting LINDA v0.2.6 ... Loading required package: ANTsRCore randomForest 4.6-12 Enter file name: T1.nii.gz 16:48 Loading file: T1.nii.gz 16:48 Creating folder: ./linda 16:48 Loading template... 16:48 Skull stripping... (long process) 17:06 Saving skull stripped files 17:06 Loading LINDA model 17:06 Computing asymmetry mask... 17:08 Saving asymmetry mask... 17:08 Running 1st registration... 17:11 Feature calculation... 17:15 Running 1st prediction... Error in which(submask > 0) : argument to 'which' is not logical
It seems that, inside the "mrvnrfs.predict_chunks" function, the data type for the submask cannot be used by the "which" function. I did some debugging, and it turns out that "submask" is actually of type "antsImage", with max = 1 and min = 0:
antsImage Pixel Type : unsigned char Components Per Pixel: 1 Dimensions : 256x256x192 Voxel Spacing : 1x1x0.999995172023773 Origin : 103.2539 -120.9078 -117.9969 Direction : -0.008816802 0.9962615 0.08593764 -0.01123489 -0.08603425 0.9962288 -0.999898 -0.007818053 -0.01195143
I cannot figure out what's wrong here. Any hint? I attach here the T1 example I am currently using.
Thanks a lot, Alberto T1.nii.gz https://github.com/dorianps/LINDA/files/2163838/T1.nii.gz
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/dorianps/LINDA/issues/9, or mute the thread https://github.com/notifications/unsubscribe-auth/AIqafQ5pE3sFD1m3PYiBVXaH97XJI6nXks5uDOuvgaJpZM4VC0up .
The new version of LINDA made it! It seems to be working properly. Thanks so much for your support, Dorian.
Best, Alberto
Software versions:
When sourcing the "linda_predict" script, everything works fine up to the 1st prediction step:
It seems that, inside the "mrvnrfs.predict_chunks" function, the data type for the submask cannot be used by the "which" function. I did some debugging, and it turns out that "submask" is actually of type "antsImage", with max = 1 and min = 0:
I cannot figure out what's wrong here. Any hint? I attach here the T1 example I am currently using.
Thanks a lot, Alberto T1.nii.gz