• lots of outputs are not properly documented
• need to show how to plot a network in brain space given output of above

gotta be under 100MB for cran

http://cran.r-project.org/doc/manuals/r-release/R-exts.html#Configure-and-cleanup

For example:

WarpImageMultiTransform()
Error in .Call("WarpImageMultiTransform", as.character(c(...))) :
"WarpImageMultiTransform" not resolved from current namespace (ANTsR)

I think this example is important because right now you can transform into a fixed space, perform operations, but you cannot go back into the moving space afterward.

I installed the latest ANTsR version from github and noticed that selecting any affine transform causes a segfault. I checked using the command line tool built in src/antb against an older version from 2014-09-29 that works:
old version (runs fine):

~/GIT/DEV/ANTSbuild/bin/antsRegistration -d 3 -m mattes[skull1.mha,skull2.mha,1,32,Regular,0.1] -t rigid[ 1.0 ] -c [1000x100x10,1e-8,20] -s 4x2x1vox -f 3x2x1 -u 1 -o [skull1.mha_fixed_skull1.mha_moving_custom,skull1.mha_fixed_skull1.mha_moving_custom_diff.nii.gz,skull1.mha_fixed_skull1.mha_moving_custom_inv.nii.gz]

new version:

~/GIT/DEV/ANTsR/src/antb/bin/antsRegistration -d 3 -m mattes[skull1.mha,skull2.mha,1,32,Regular,0.1] -t rigid[ 1.0 ] -c [1000x100x10,1e-8,20] -s 4x2x1vox -f 3x2x1 -u 1 -o [skull1.mha_fixed_skull1.mha_moving_custom,skull1.mha_fixed_skull1.mha_moving_custom_diff.nii.gz,skull1.mha_fixed_skull1.mha_moving_custom_inv.nii.gz]

fails with

All_Command_lines_OK
Using double precision for computations.
  number of levels = 3
  fixed image: skull1.mha
  moving image: skull2.mha
Dimension = 3
Number of stages = 1
Use Histogram Matching true
Winsorize image intensities false
Lower quantile = 0
Upper quantile = 1
Stage 1 State
   Image metric = Mattes
     Fixed image = Image (0x3c44700)
  RTTI typeinfo:   itk::Image<double, 3u>
  Reference Count: 2
  Modified Time: 639
  Debug: Off
  Object Name: 
  Observers: 
    none
  Source: (none)
  Source output name: (none)
  Release Data: Off
  Data Released: False
  Global Release Data: Off
  PipelineMTime: 0
  UpdateMTime: 448
  RealTimeStamp: 0 seconds 
  LargestPossibleRegion: 
    Dimension: 3
    Index: [0, 0, 0]
    Size: [117, 157, 170]
  BufferedRegion: 
    Dimension: 3
    Index: [0, 0, 0]
    Size: [117, 157, 170]
  RequestedRegion: 
    Dimension: 3
    Index: [0, 0, 0]
    Size: [117, 157, 170]
  Spacing: [1.5, 1.5, 1.5]
  Origin: [-88.5114, -114.68, -157.958]
  Direction: 
1 0 0
0 1 0
0 0 1

  IndexToPointMatrix: 
1.5 0 0
0 1.5 0
0 0 1.5

  PointToIndexMatrix: 
0.666667 0 0
0 0.666667 0
0 0 0.666667

  Inverse Direction: 
1 0 0
0 1 0
0 0 1

  PixelContainer: 
    ImportImageContainer (0x3c449c0)
      RTTI typeinfo:   itk::ImportImageContainer<unsigned long, double>
      Reference Count: 1
      Modified Time: 445
      Debug: Off
      Object Name: 
      Observers: 
        none
      Pointer: 0x7f131f95b010
      Container manages memory: true
      Size: 3122730
      Capacity: 3122730

     Moving image = Image (0x3c48710)
  RTTI typeinfo:   itk::Image<double, 3u>
  Reference Count: 2
  Modified Time: 640
  Debug: Off
  Object Name: 
  Observers: 
    none
  Source: (none)
  Source output name: (none)
  Release Data: Off
  Data Released: False
  Global Release Data: Off
  PipelineMTime: 0
  UpdateMTime: 637
  RealTimeStamp: 0 seconds 
  LargestPossibleRegion: 
    Dimension: 3
    Index: [0, 0, 0]
    Size: [121, 148, 155]
  BufferedRegion: 
    Dimension: 3
    Index: [0, 0, 0]
    Size: [121, 148, 155]
  RequestedRegion: 
    Dimension: 3
    Index: [0, 0, 0]
    Size: [121, 148, 155]
  Spacing: [1.5, 1.5, 1.5]
  Origin: [-85.1487, -103.461, -145.826]
  Direction: 
1 0 0
0 1 0
0 0 1

  IndexToPointMatrix: 
1.5 0 0
0 1.5 0
0 0 1.5

  PointToIndexMatrix: 
0.666667 0 0
0 0.666667 0
0 0 0.666667

  Inverse Direction: 
1 0 0
0 1 0
0 0 1

  PixelContainer: 
    ImportImageContainer (0x3c49c30)
      RTTI typeinfo:   itk::ImportImageContainer<unsigned long, double>
      Reference Count: 1
      Modified Time: 634
      Debug: Off
      Object Name: 
      Observers: 
        none
      Pointer: 0x7f131e42d010
      Container manages memory: true
      Size: 2775740
      Capacity: 2775740

     Weighting = 1
     Sampling strategy = regular
     Number of bins = 32
     Radius = 4
     Sampling percentage  = 0.1
   Transform = Rigid
     Gradient step = 1
     Update field sigma (voxel space) = 0
     Total field sigma (voxel space) = 0
     Update field time sigma = 0
     Total field time sigma  = 0
     Number of time indices = 0
     Number of time point samples = 0
Registration using 1 total stages.

Stage 0
  iterations = 1000x100x10
  convergence threshold = 1e-08
  convergence window size = 20
  number of levels = 3
  using the Mattes MI metric (number of bins = 32, weight = 1)
  preprocessing:  histogram matching the images
  Shrink factors (level 1 out of 3): [3, 3, 3]
  Shrink factors (level 2 out of 3): [2, 2, 2]
  Shrink factors (level 3 out of 3): [1, 1, 1]
  smoothing sigmas per level: [4, 2, 1]
  regular sampling (percentage = 0.1)
Segmentation fault (core dumped)

Am I missing something?

Best
Stefan

I tested the ANTs SmoothImage command and the ANTsR SmoothImage.R function, and I didn't get the same output results with a same sigma in input...

How could this be explained ?

I let you try with my 3D test image ("test.nii") :
http://we.tl/oNkfQirp39

Best regards,
Matthieu

When I installed using
library(devtools)
install_github("ANTsR", "stnava")
the install worked fine on MacOSX 10.9.2, R 3.02.

Linux server didn't have CMake so tried the pre-compiled binaries, but didn't work. Install said it was fine, but when libraried in:

Error in dyn.load(file, DLLpath = DLLpath, ...) :
unable to load shared object '/home/bst/student/jmuschel/R/x86_64-unknown-linux-gnu-library/3.0/ANTsR/libs/ANTsR.so':

Which I think is because of my version of Rcpp:
http://cran.r-project.org/web/packages/Rcpp/news.html
"Functions provided/used by Rcpp are now registered with R and instantiated by client package alleviating the new for explicit linking against libRcpp which is therefore no longer created."

needs work

One of the biggest issues when working with 1mm images is the huge matrix size obtained from a full brain 1mm neighborhood matrix with radius 4. The matrix can be 16gb or more.

To my understanding, with higher resolution the neaghboring voxels are increasingly less informative. What I need is not the immediate surrounding voxel layers but further out. The best would be to get i.e. 1st neighrbohood layer, 3rd layer, etc, thus skipping some neighborhood layers.

Would be great if this is achieved with a "step" parameter, or better if can be defined manually, like c(1,3,4) to get 1st layer + 3rd layer + 4th layer.

Let me know if this is not clear.

see discussion here:

joey711/phyloseq#322

it's not clear why this would happen but in the meantime you can try an install from the binary releases

or by cloning the repository and using R CMD INSTALL on the cloned repo.

Thanks a lot for bringing this functionality to R. During installation (Ubuntu 14.04; R 3.1.0) I noticed a couple of compiler errors:
e.g: error: ‘constexpr’ needed for in-class initialization of static data member ‘const float vnl_numeric_traits::one’ of non-integral type
To make the package compile successfully, I needed to set

 CPPFLAGS=-fpermissive

in ~/.R/Makevars

Best
Stefan

First, I want to say thanks for so quickly resolving my last abpBrainExtraction issue, it worked. But, I used a crappy "tem" and "temmask" so I got a crappy brain mask. In an effort to fix this I am trying to use the MNI standard brain as the template and it's very good brain mask as the template mask. I used the code below:

ImageMath 3 close_MP1.nii.gz ClosestSimplifiedHeaderMatrix MP1.nii.gz
ImageMath 3 close_MNI.nii.gz ClosestSimplifiedHeaderMatrix MNInew.nii.gz
ImageMath 3 close_MNImask.nii.gz ClosestSimplifiedHeaderMatrix brainMask.nii

struct<-antsImageRead("/Users/Bryson/Data/ANTsR_test/close_MP1.nii.gz")
struct<-abpN4(struct)
tem<-antsImageRead("/Users/Bryson/Data/ANTsR_test/close_MNI.nii.gz")
tem<-abpN4(tem)
temmask<-antsImageRead("/Users/Bryson/Data/ANTsR_test/close_MNImask.nii.gz")

bm<-abpBrainExtraction(img=struct,tem=tem,temmask=temmask)

antsImageWrite(bm$bmask, "MP1bmask.nii.gz")
antsImageWrite(bm$brain, "MP1masked.nii.gz")

But the output I got looked more like a head-excluding mask than a brain-including mask.

I've attached the input and output files here:
https://drive.google.com/open?id=0B9oWACI5TJOEfm1KdzAxSG5IdlNZR3QyOEtGa3l2ZEVETHBUQXZyNDJGUGk5YXRNT2N4blk&authuser=0

Thanks again for your help!

this is from R 3.2

Examples

... to test examples by themselves, do the following

cd ANTsR # source dir
ln -s inst/extdata .
R
devtools::run_examples()

when you are done, be sure to remove the linked ext data

current findings:

• aslPerfusion reports "exception caught"
• iMathOps.Rd fails on several examples ... i fixed a couple but had to comment
  k <- eval( parse( text = toString( iMathOps$Example[j] ) ) )
  due to unclear BlobDetector failure ... this really needs to be fixed

 i<-antsImageRead( getANTsRData('r16') , 2 )
 iMath(i,"BlobDetector",100)

• basic erosion and dilation fails

img<-antsImageRead( getANTsRData("r16"))
 mask<-getMask( img )
 mask2<-getMask( img )
 mask<-iMath(mask,'ME',25)
print(paste( sum(mask2==1), sum(mask==1) ))

i added a test for this in iMathOps ...

Vignettes

seem to be ok .... though i did not look at the results visually. they at least build.

Documentation issues

• checking R code for possible problems ... NOTE
  aslCensoring : robSelection: no visible binding for global variable
  ‘mask_img’
  aslCensoring : robSelection: no visible binding for global variable
  ‘dorobust’
  aslCensoring: no visible binding for global variable ‘mat’
• checking for missing documentation entries ... WARNING
  Undocumented code objects:
  %>%

this last issue was resolved with an Imath documentation object

but this was removed for unknown reasons ... i undid this here:

2460a41

so as not to avoid the variety of things that were breaking.

@armaneshaghi - looks like we need several different options for travis builds

• i'd like to be able to automate releases from travis
• would also like to see RMD check results ( has to be done separately from above )
• coverage numbers would be nice - but this seems to time out

seems that , b/c of limited travis time, we can't do all at once ...

any ideas about how to approach this?

It would be extremely useful to have a function that can save the entire workspace in a single line. Right now is very difficult to keep track of what is a pointer and what is not, then save them separately, then write code again to load them on new session. A save-all function can take this burden away, create cleaner code with one-line saving, and offer to restart at different levels of a script. Of course, this should be complemented with a load function.

I am using cropImage:
predlabel = cropImage(predlabel, predlabel, max(predlabel))

At this stage predlabel can be plotted, but it is already corrupt:

"> predlabel
antsImage
Pixel Type : float
Pixel Size :
Dimensions : 32x49x38
Voxel Spacing: 2x2x2
Origin : -93 138.1172 -82
Direction : 1 0 0 0 -1 0 0 0 1
��> predlabel[15,15,15]
index not inside the image : [14, 14, 14]
[1] NA"

After cropping, resampleImage is performed (within mrvnrfs, but I tried manually as well). There is no error but resampleImage produces a corrupt image, it cannot be plotted and getNeirghborhoodInMask is a matrix with 0 columns.

Cropping works fine if I use traditional code:

ttt2 = as.antsImage(as.array(predlabel)[20:80, 30:100, 1:84])
��> ttt2[15,15,15]
, , 1

 [,1]

[1,] 0

plot(ttt2)
NULL
ttt3=resampleImage(ttt2, rep(4,3), 0, 1)
plot(ttt3)
NULL

Hi,
after your recent revamp, "-DVCL_CAN_STATIC_CONST_INIT_FLOAT=0" was dropped from
-PKG_CPPFLAGS. With this option added, everything builds fine again.
This requires to throw in an -fpermissive flag. Could you please readd this, or is there a reason for dropping it?

Best
Stefan

Hi, I am trying to get the affine transformation from a .mat file. I can import them via R.matlab::readMat() and get a list containing $AffineTransform.float.3.3 (a 12x1 matrix) and a 3x1 matrix named $fixed. Is there an easy way to combine this to a transform matrix (4x4 in my case)?
It looks like this:

$AffineTransform.float.3.3
               [,1]
 [1,]    1.05657983
 [2,]    0.04566867
 [3,]    0.03638180
 [4,]   -0.04745625
 [5,]    0.99157244
 [6,]   -0.15884599
 [7,]   -0.01750578
 [8,]    0.08629224
 [9,]    1.02625072
[10,]   -3.50945663
[11,] -185.69580078
[12,]  777.48010254

$fixed
              [,1]
[1,]    -1.3779984
[2,]     0.2889023
[3,] -1098.2500000

attr(,"header")
attr(,"header")$version
[1] "4"

attr(,"header")$endian
[1] "little"

The correct affine matrix (calculated from deformed pointsets using Morpho::computeTransform()) would be:

[1,]  1.05657998  0.04566868 -0.03638175 -36.51162
[2,] -0.04745618  0.99157235  0.15884597 360.21133
[3,]  0.01750584 -0.08629226  1.02625079 806.26085
[4,]  0.00000000  0.00000000  0.00000000   1.00000

of which the upper-left 3x3 submatrix is (apart from the sign) equivalent to matrix(mat$AffineTransform.float.3.3[1:9],3,3,byrow = T)

Thanks
Stefan

predlabel2 = predlabel*0
plot(predlabel2)
Error in if (window.img[1] == window.img[2]) window.img[1] <- min(x) :
missing value where TRUE/FALSE needed

The plot of an image with an overlay shows wrong displacement between the two. Initially I thought this is related to a double conversion antsImage->array->antsImage , but now it's clear that is happening in many other situations, perhaps even with original images. Saving the image and loading in itk-snap shows no displacement between images, so this is a specific problem of the plot function.

@jeffduda @bkandel

ANTsR will be "broken" for a while ... Being tired of this ongoing issue, I did the aggressive thing and removed imageMath and src/ImageMath.cpp and also updated the antstag s.t. ImageMath is not compiled in core. here are the current issues:

* checking R code for possible problems ... NOTE
aslCensoring : aslOutlierRejection: no visible global function
  definition for ‘imageMath’
basicInPaint: no visible global function definition for ‘imageMath’
compcor: no visible global function definition for ‘imageMath’
getCentroids: no visible global function definition for ‘imageMath’
getTemplateCoordinates: no visible global function definition for
  ‘imageMath’
projectImageAlongAxis: no visible global function definition for
  ‘imageMath’
sliceTimingCorrection: no visible global function definition for
  ‘imageMath’
* checking Rd files ... OK

relating to:

ANTsR/R/aslCensoring.R:    imageMath(4, diffs, "TimeSeriesSimpleSubtraction", asl)
ANTsR/R/basicInPaint.R:  imageMath(img@dimension, temp, "MD", temp, 1)
ANTsR/R/basicInPaint.R:  # imageMath(img@dimension,paintMaskUse,'GD',paintMask,1)
ANTsR/R/basicInPaint.R:  imageMath(inpainted@dimension, inpainted, "FastMarchingExtension", speedimage,
ANTsR/R/compcor.R:    imageMath(4, myoutfn, "CompCorrAuto", fmri, mask, ncompcor)
ANTsR/R/getCentroids.R:  imageMath(imagedim, pointfile, "LabelStats", img, img, clustparam)
ANTsR/R/getTemplateCoordinates.R:  imageMath(milab@dimension, pointfile, "LabelStats",
ANTsR/R/pairwiseImageDistanceMatrix.R:        # broken !!  metric <- capture.output(imageMath(dim, 'j', metrictype, i1,
ANTsR/R/projectImageAlongAxis.R:  imageMath(imageND@dimension, downimg, "Project", imageND, axis, projtype)
ANTsR/R/sliceTimingCorrection.R:    imageMath(4, corrected, "SliceTimingCorrection", fmri, sliceTime, interpolation,
ANTsR/R/sliceTimingCorrection.R:    imageMath(4, corrected, "SliceTimingCorrection", fmri, sliceTime, interpolation,
ANTsR/R/sliceTimingCorrection.R:    imageMath(4, corrected, "SliceTimingCorrection", fmri, sliceTime)

here is a commit that deals with some of these:

3713fd8

however it would be good to get the rest as well ...

@jeffduda a few pieces of functionality in iMath

• Project operator from ImageMath
• FastMarchingExtension

@bkandel add x, y, z, t to labelStats ... the output is always 4D centroids ... e.g. in 2D it will contain x_i, y_i, 0, 0 ( or maybe NA, NA ) ....

the other issues should be self explanatory.

to run tests in a different repo

I just installed ANTsR and the problem I am having is with abpBrainExtraction.
I enter the following commands in R in terminal:

struct<-antsImageRead("/Users/Bryson/Data/ANTsR_test/MPRAGE.nii.gz")
tem<-antsImageRead("/Users/Bryson/Data/ANTsR_test/bet.nii.gz")
temmask<-antsImageRead("/Users/Bryson/Data/ANTsR_test/bet_mask.nii.gz")

bm<-abpBrainExtraction(img=struct,tem=tem,temmask=temmask)

And I get the following error:

Problem in arg list
see usage by calling antsApplyTransforms() w/o arguments
Error in thresholdImage(temmaskwarped, 0.5, 1) :
trying to get slot "dimension" from an object of a basic class ("numeric") with no slots

So I called antsApplyTransforms() w/o arguments, but it wasn't helpful to me:

antsApplyTransforms()
warpedimg<-antsApplyTransforms( fixed=img1 , moving=img2 ,
transformlist=list("my0GenericAffine.mat","my1Warp.nii.gz") )

For full mode: use standard ants call as in antsApplyTransforms full mode ...

antsApplyTransforms("-d","2","-i","r64slice.nii.gz","-o","temp.nii.gz","-r","r16slice.nii.gz","-t","./Z0GenericAffine.mat")
for full help:
use .Call( "antsApplyTransforms", .int_antsProcessArguments( c(list("--help")) ), PACKAGE="ANTsR" );

[1] 0

The "img" image is a MPRAGE from the ABIDE dataset.
The "tem" image is the bet brain extracted version of the same image(I wasn't sure what to put here).
And the "temmask" is the brain mask output from bet.
I've run the example in the manual and it runs fine. Also, I tried running abpBrainExtraction.R another time, and although it didn't give me good results (probably from a bad choice of the "tem" image), it did finish without errors. Any ideas on what I can do?

Hi,
as I am mostly dealing with Volume and surface data, I momentarily am using a wrapper called antsTransformPoints* (it is an S3 function) calling the command line tool antsApplyTransformsToPoints to apply the transforms to point sets. It takes care of IJK to RAS transform and works on matrices and meshes of class "mesh3d". Would it be a big issue to implement this in ANTsR?

*) vert2points used from Morpho is simply

vert2points  <- function (mesh) {
    out <- t(mesh$vb[1:3, ])
    return(out)
}

Best
Stefan

see https://github.com/stnava/ANTsR/blob/master/DESCRIPTION#L4-L5

These should be synced. Also you might want to get rid of the date field since it's not necessary and it's easy to forget to update it.

Have a viewer (decided for MRIcron) that can be launched from R to view an antsImage.

@jeffduda
Issue emerged here: #42

Problem: assigning is performed on wrong slices, and may also contaminate other dimensions

Example:
3D file: https://www.dropbox.com/s/ot4ponqk7j221n6/templateBrainMask.nii.gz?dl=0

brainmask=antsImageRead('templateBrainMask.nii.gz',3)
cropmask=brainmask*0
cropmask[1:192,1:256, 83:89] = 1
plot(cropmask, axis=3, slices=82:90) # should show two black slices on the sides, but it doesn't
plot(cropmask, axis=3, slices=83:91) # this shows two black slices on the sides

Plus cropImage thinks there are 8 slices with some value = 1.
cropImage(brainmask,cropmask)
antsImage
Pixel Type : float
Pixel Size :
Dimensions : 192x256x8
Voxel Spacing: 0.9765625x0.9765625x1
Origin : -93 138.1172 1
Direction : 1 0 0 0 -1 0 0 0 1

@ntustison - the violators include

• sccan
• antsRegistration
• Atropos
• antsMotionCorr
• N4BiasFieldCorrection
• .... anything else?

Goal: test ImageMath for addtozero:

from man:

addtozero : add image-b to image-a only over points where image-a has zero values

Reproducible example:

require(ANTsR)
require(testthat)
         img1_array <- array(rep(2, 5*5), dim=c(5,5))
         img1_array[1,2] <- 0
         img1_array[2,3] <- 0
         img1_array[2,2] <- 0
         img1_array[3,3] <- 0
         img1_array[4,3] <- 0
         img1_array[4,3] <- 0
         img1_array[4,4] <- 0
         img1_array[4,3] <- 0
         img1_array[5,1] <- 0
         img2_array <- array(rep(2, 5*5), dim=c(5,5))

so:

> as.array(img1)
     [,1] [,2] [,3] [,4] [,5]
[1,]    2    0    2    2    2
[2,]    2    0    0    2    2
[3,]    2    2    0    2    2
[4,]    2    2    0    0    2
[5,]    0    2    2    2    2

and:

> as.array(img2)
     [,1] [,2] [,3] [,4] [,5]
[1,]    2    2    2    2    2
[2,]    2    2    2    2    2
[3,]    2    2    2    2    2
[4,]    2    2    2    2    2
[5,]    2    2    2    2    2

Moving on:

         img1 <- as.antsImage(img1_array)
         img2 <- as.antsImage(img2_array)
         output <- antsImageClone(img2)
         ImageMath(2, output, 'addtozero', img1, img2)

I would expect to get a homogenous array with 2s, but:

> as.array(output)
     [,1] [,2]          [,3]          [,4]          [,5]
[1,]    2    2  2.000000e+00 1.850677e-316 6.926506e-310
[2,]    2    2  2.000000e+00 3.127860e-316 4.940656e-324
[3,]    2    2  0.000000e+00 4.356597e-316 6.926506e-310
[4,]    2    2 4.431769e-321 1.037538e-321 6.926506e-310
[5,]    2    2 6.926606e-310 3.458460e-323 9.881313e-324

to check if casting could be a problem:

> img1@pixeltype
[1] "double"
> img2@pixeltype
[1] "double"
> output@pixeltype
[1] "double"

ImageMath.cxx:

 else if( strcmp(operation.c_str(), "addtozero") == 0 && pix1 == 0 )
      {
      result = pix1 + pix2;
      }
    else if( strcmp(operation.c_str(), "addtozero") == 0 && pix1 != 0 )
      {
      result = pix1;

Ideas?

Arman

Write wrapper testANTsRExample function to do stopifnot on examples.

An aesthetic issue:
Perhaps would be better to have consistent name for make, read, and write image. Maybe is better to have makeImage, readImage, writeImage.

@bkandel thought this is bug.

I have a crop mask with 7 slices (83:89), but crop is producing an image with 8 slices.

Get file here:
https://www.dropbox.com/s/ot4ponqk7j221n6/templateBrainMask.nii.gz?dl=0

Run:
brainmask=antsImageRead('templateBrainMask.nii.gz',3)
cropmask=brainmask*0
cropmask[1:192,1:256, 83:89] = 1
cropImage(brainmask,cropmask)
antsImage
Pixel Type : float
Pixel Size :
Dimensions : 192x256x8
Voxel Spacing: 0.9765625x0.9765625x1
Origin : -93 138.1172 1
Direction : 1 0 0 0 -1 0 0 0 1

Started getting this earlier today on the master branch. But the file seems to exist

src/itkb/itkinst/include/ITK-4.8/itkAddImageFilter.h
src/itks/Modules/Filtering/ImageIntensity/include/itkAddImageFilter.h

Still, R CMD INSTALL ANTsR quickly fails with

antsImage.cpp:6:10: fatal error: 'itkAddImageFilter.h' file not found

include "itkAddImageFilter.h"

     ^

1 error generated.
make: *** [antsImage.o] Error 1
ERROR: compilation failed for package ‘ANTsR’

• removing ‘/Users/pcook/Library/R/3.1/library/ANTsR’
• restoring previous ‘/Users/pcook/Library/R/3.1/library/ANTsR’

Trying to complete the documentation for LabelClustersUniquely, can anyone confirm my understanding based on the source code:

• Takes a statistical map (e.g. T map)?
• Takes a threshold value, based on which clusters are formed
• Outputs a sequentially re-labeld image based on the thresholded value

Where exactly have you used this before?

This is a really weird error. I want to use foreach loops with multiple processors. If I run a single loop I can capture the output and everything is fine. If I use more than one loop, the output is corrupt and R crashes. Here is the example:

img1 = antsImageRead(getANTsRData('r16'), 2)
img2 = antsImageRead(getANTsRData('r27'), 2)
temp=list(img1,img2)

library(foreach)
library(doMC)
registerDoMC(2)

single loop

output = foreach(i=1:1) %dopar% {
kmeansSegmentation(temp[[i]], 3, getMask(temp[[i]]))$segmentation
}

output
[[1]]
antsImage
Pixel Type : float
Pixel Size :
Dimensions : 256x256
Voxel Spacing: 1x1
Origin : 0 0
Direction : 1 0 0 1

plot(output[[1]])

two loops

output = foreach(i=1:2) %dopar% {
kmeansSegmentation(temp[[i]], 3, getMask(temp[[i]]))$segmentation
}

output
[[1]]
antsImage
Pixel Type : float
Pixel Size :

*** caught segfault ***
address (nil), cause 'memory not mapped'

Traceback:
1: .Call("antsImage_dim", x, PACKAGE = "ANTsR")
2: dim(object)
3: dim(object)
4: paste(dim(object), collapse = "x")
5: cat(" Dimensions :", paste(dim(object), collapse = "x"), "\n")
6: function (object) standardGeneric("show")(<S4 object of class "antsImage">)
7: function (object) standardGeneric("show")(<S4 object of class "antsImage">)
8: print.default(<S4 object of class "antsImage">)
9: print(<S4 object of class "antsImage">)

Possible actions:
1: abort (with core dump, if enabled)
2: normal R exit
3: exit R without saving workspace
4: exit R saving workspace

We could get a DOI for ANTsR and make the whole software citable. Objections?

Step by step guide:

https://guides.github.com/activities/citable-code/

It doesn't seem the variable "template" is needed for anything.

average = antsAverageImages(trn.imgCon, normalize = T)
Error: object 'template' not found

function (imageList, normalize = FALSE)
{
avg <- imageList[[1]] * 0
for (i in imageList) {
if (normalize) {
i <- i/mean(i)
}
template <- template + i
}
avg <- avg/length(imageList)
return(avg)
}
<environment: namespace:ANTsR>

Remove need for reference, just specify slices and axis.

In advance of the CRAN submission, it would be nice to pare down the functions to emphasize the core utility. Including functions in CRAN submissions implies a commitment to maintaining the function for the foreseeable future. Here is a list of functions that look to me like they either duplicate functionality or can be deleted:
maskImage
quantifySNPs (useful, but not sure it's suitable for ANTsR)
filterfMRIForNetworkAnalysis (overlap with frequencyFilterfMRI? is this still used?)
exemplarInpainting vs. basicInPainting -- why have both? Shouldn't we just have inPaint and have methods for exemplar and basic inpainting?
jointIntensityFusion and ""3D: Why have two functions for different dimensionalities?
renderNetwork and plotBasicNetwork: Can these be combined?
smoothImage -- I have always thought this should be part of iMath. Is there a consensus on this? Would this be too annoying?
visualizeBlob -- looks to me like it's been superseded by the various rendering functions.

causes seg fault - likely type issue

library(ANTsR)
boldfn <- getANTsRData( "pcasl" )
bold <- antsImageRead( boldfn , 4 )
fixed<-getAverageOfTimeSeries( bold )
mask<-getMask( fixed )
antsMotionCalculation( bold, mask=mask, fixed=fixed, moreaccurate=0)
antsMotionCalculation( bold, mask=mask, fixed=fixed, moreaccurate=0)
kk<-antsMotionCalculation( bold, mask=mask, fixed=fixed, moreaccurate=0)

I don't want to add this as a direct pool request as it may break some code already, but I think you want do this for Ops in general (or at least Arith, Math, and Summary)
https://github.com/muschellij2/ANTsR/blob/master/R/antsImage_Operators.R

I'm not clear as to how you store logicals, and these may cause some problems. It is not as flexible as the operators you've implemented in that you cannot pass the … into as.antsImage (such as mask), but I think that can be added.

Also, you may want to define:
is.antsImage <- function(x){
inherits(x, 'antsImage')
}
or simply:
is.antsImage <- function(x){
is(x, 'antsImage')
}
in case someone extends your class.

@ntustison , @bkandel , @dorianps, @jeffduda , @cookpa , @armaneshaghi, @muschellij2

hoping to work toward a CRAN submission for ANTsR - some updates on this topic:

i ran rd2roxygen and dealt with most of the issues. hopefully wont have

to do that again ... so, in the future, we should write documentation

using the roxygen2 style and just roxygenize() regularly.

a couple other changes

• enabled plot( antsImage ) so we dont need to type plotANTsImage
• added resampleImage
• updated invariantImageSimilarity to use reflection as well as rotation, only tested in 2D so far
• added CreateJacobianDeterminantImage
• added a simple kmeansSegmentation filter ...
• would like to start running R CMD CHECK and get failures to a minimum ...
• need to write a quick vignette, possibly based on the README.md ...

@muschellij2 - would appreciate any advice / help on this ....

some advice from wickham: http://www.rstudio.com/products/rpackages/devtools/

it's a very good page with strategies that will help with most of the issues we've had in the past ...

any thoughts appreciated.

works with + or -
temp = ConAvgDiff.t1[[k]] %>% iMath('Normalize') %>% - brainmask

doesn't work with * and /
temp = ConAvgDiff.t1[[k]] %>% iMath('Normalize') %>% * brainmask
Error: unexpected '*' in "temp = ConAvgDiff.t1[[k]] %>% iMath('Normalize') %>% *"

Example:
img <- antsImageRead(getANTsRData('r16'),2)

img[,,]
Error en typeof(i) : el argumento "i" está ausente, sin valor por omisión
img[,1:20]
Error en typeof(i) : el argumento "i" está ausente, sin valor por omisión
img[1:20,]
[1] "'mask' provided is not of type 'logical'"
NULL

The activity should match that of R arrays, where img[, ,] should return all the values in the image.

labelsimage won't work, due to missing negation and calling missing object limg.

sort of a dual bug ... apply.antsImage gives you some junky header info and plot fails when it gets this - presumably actually cropImage fails ....

but quite nasty because of the segfault

tslice=apply.antsImage( img, c(1,3,4), mean )
plot(tslice)
ITK ExceptionObject caught !

itk::ExceptionObject (0x7faac52003c8)
Location: "unknown" 
File: /Users/stnava/code/RLibs/ITKR/libs/include/ITK-4.8/itkImageBase.hxx
Line: 193
Description: itk::ERROR: Image(0x7faac3dbbe70): Bad direction, determinant is 0. Direction is -0 -1 0
0 0 0
0 0 1



libc++abi.dylib: terminating with uncaught exception of type Rcpp::exception: ITK exception caught
Abort trap: 6
MacBook-Pro:wetransfer-2bfbfc stnava$ plot(tslice)
-bash: syntax error near unexpected token `tslice'

I am having this issue all the time. AntsImages are unusuable if saved in .R files and then reloaded.

For example, I load antsImages in workspace, save the workspace, and reload it again later. What happens is that antsImages seem to be there, but are compromised. If I try to use them R goes in segmentaiton fault.
There are two examples below, first I load an .R file, use mask from within it and R crashes. Second I remove mask and reload it again, and all works.
Is this a bug or antsImages S4 objects simply cannot be saved?

Welcome! Loading ANTsR...
ANTsR is loaded.

load(file="subjectdata.R")
mask
An object of class "antsImage"
Slot "pixeltype":
[1] "float"

Slot "dimension":
[1] 3

Slot "pointer":
<pointer: (nil)>

ls()
[1] "imageList" "mask" "mydata" "thick" "workfolder"
mat<-imagesToMatrix(imageList, mask)

*** caught segfault ***
address (nil), cause 'memory not mapped'

Traceback:
1: .Call("antsImage_RelationalOperators", e1, e2, region, operator, PACKAGE = "ANTsR")
2: mask > 0
3: mask > 0
4: which(mask > 0)
5: imagesToMatrix(imageList, mask)

Possible actions:
1: abort (with core dump, if enabled)
2: normal R exit
3: exit R without saving workspace
4: exit R saving workspace
Selection: q()
Selection: 3

working example

Welcome! Loading ANTsR...
ANTsR is loaded.

load(file="subjectdata.R")
rm(mask)
mask<-antsImageRead('processing/smo_4_NormCortThick/mask.nii.gz', 3)
mat<-imagesToMatrix(imageList, mask)

Wrap PrintHeader from ANTs into Rcpp.

Why can this not be installed using the "standard" github installer:

library(devtools)
install_github("ANTsR", "stnava")

?

1. Enable adding a title to the plot.
2. Crop better the image so it doesn't plot in two rows when not necessary.
3. Fix slices='84x84x1' so it gets simply slices=84, or better slices=84:90, or slices=c(20,32,56) for multiple images.

antsImageRead/Write does currently not expand path (i.e. antsImageRead("~/myimage.nii.gz") leads to a segfault, the same is if the image to import does not exist.