def writenifti(image, filename, info):
"""Write nifti file."""
writer = vtk.vtkNIFTIImageWriter()
writer.SetInputData(image)
writer.SetFileName(filename)
writer.SetInformation(info)
writer.Write()
def vtk_write_mask_as_nifty(mask, M, image_fn, mask_fn):
reader = vtk.vtkNIFTIImageReader()
reader.SetFileName(image_fn)
reader.Update()
writer = vtk.vtkNIFTIImageWriter()
M.Invert()
if reader.GetQFac() == -1:
for i in range(3):
temp = M.GetElement(i, 2)
M.SetElement(i, 2, temp * -1)
reslice = vtk.vtkImageReslice()
reslice.SetInputData(mask)
reslice.SetResliceAxes(M)
reslice.SetInterpolationModeToNearestNeighbor()
reslice.Update()
mask = reslice.GetOutput()
mask.SetOrigin([0., 0., 0.])
writer.SetInputData(mask)
writer.SetFileName(mask_fn)
writer.SetQFac(reader.GetQFac())
q_mat = reader.GetQFormMatrix()
writer.SetQFormMatrix(q_mat)
s_mat = reader.GetSFormMatrix()
writer.SetSFormMatrix(s_mat)
writer.Write()
return
def vtk_write_mask_as_nifty(mask, image_fn, mask_fn):
import vtk
origin = mask.GetOrigin()
reader = vtk.vtkNIFTIImageReader()
reader.SetFileName(image_fn)
reader.Update()
writer = vtk.vtkNIFTIImageWriter()
Sign = vtk.vtkMatrix4x4()
Sign.Identity()
Sign.SetElement(0, 0, -1)
Sign.SetElement(1, 1, -1)
M = reader.GetQFormMatrix()
if M is None:
M = reader.GetSFormMatrix()
M2 = vtk.vtkMatrix4x4()
M2.Multiply4x4(Sign, M, M2)
reslice = vtk.vtkImageReslice()
reslice.SetInputData(mask)
reslice.SetResliceAxes(M2)
reslice.SetInterpolationModeToNearestNeighbor()
reslice.Update()
mask = reslice.GetOutput()
mask.SetOrigin([0., 0., 0.])
writer.SetInputData(mask)
writer.SetFileName(mask_fn)
writer.SetQFac(reader.GetQFac())
q_mat = reader.GetQFormMatrix()
writer.SetQFormMatrix(q_mat)
s_mat = reader.GetSFormMatrix()
writer.SetSFormMatrix(s_mat)
writer.Write()
return
def write_image(_data, path_to_image):
# == make sure data type is unsigned char
cast = vtk.vtkImageCast()
if vtk.VTK_MAJOR_VERSION <= 5:
cast.SetInput(_data)
else:
cast.SetInputData(_data)
cast.SetOutputScalarTypeToUnsignedChar()
cast.Update()
_data = cast.GetOutput()
# == write
file_ext = fu.get_file_extension(path_to_image)
if (file_ext == "vti"):
print("Writing as '.vti' file.")
image_writer = vtk.vtkXMLImageDataWriter()
elif (file_ext == "nii"):
print("Writing as '.nii' file.")
image_writer = vtk.vtkNIFTIImageWriter()
print(
"VTK seems to change image orientation of NIFTI. Make sure to check image orientation relative to original image"
)
elif (file_ext == "mhd" or file_ext == "mha"):
print("Writing as .mhd/.raw or .mha image.")
image_writer = vtk.vtkMetaImageWriter()
image_writer.SetCompression(False)
else:
print("No valid image file extension specified!")
if vtk.VTK_MAJOR_VERSION <= 5:
image_writer.SetInput(_data)
else:
image_writer.SetInputData(_data)
image_writer.SetFileName(path_to_image)
# image_writer.Update()
image_writer.Write()
print("Image has been saved as %s " % (path_to_image))
def test_get_nii_gz(self):
'''compressed nifti file returns correct reader'''
extension='.nii.gz'
expected = type(vtk.vtkNIFTIImageReader())
writer = vtk.vtkNIFTIImageWriter()
filename = os.path.join(self.test_dir, 'file'+extension)
self.generate_image(filename, writer)
self.assertEqual(type(get_vtk_reader(filename)), expected)
def ImageCropper(Image,X1, X2,Y1,Y2, Z1,Z2):
CroppedImage = vtk.vtkExtractVOI()
CroppedImage.SetInputData(Image)
CroppedImage.SetVOI(X1, X2,Y1,Y2, Z1,Z2)
CroppedImage.Update()
writer=vtk.vtkNIFTIImageWriter()
writer.SetInputConnection(CroppedImage.GetOutputPort())
writer.SetFileName("CroppedImageout.nii")
writer.Write()
reader=vtk.vtkNIFTIImageReader()
reader.SetFileName("CroppedImageout.nii")
reader.Update()
image=reader.GetOutput()
return(image)
def TestReadWriteRead(infile, outfile):
"""Read, write, and re-read a file, return difference."""
inpath = os.path.join(str(VTK_DATA_ROOT), "Data", infile)
outpath = os.path.join(str(VTK_TEMP_DIR), outfile)
# read a NIFTI file
reader = vtk.vtkNIFTIImageReader()
reader.SetFileName(inpath)
reader.TimeAsVectorOn()
reader.Update()
writer = vtk.vtkNIFTIImageWriter()
writer.SetInputConnection(reader.GetOutputPort())
writer.SetFileName(outpath)
# copy most information directoy from the header
writer.SetNIFTIHeader(reader.GetNIFTIHeader())
# this information will override the reader's header
writer.SetQFac(reader.GetQFac())
writer.SetTimeDimension(reader.GetTimeDimension())
writer.SetQFormMatrix(reader.GetQFormMatrix())
writer.SetSFormMatrix(reader.GetSFormMatrix())
writer.Write()
reader2 = vtk.vtkNIFTIImageReader()
reader2.SetFileName(outpath)
reader2.TimeAsVectorOn()
reader2.Update()
diff = vtk.vtkImageMathematics()
diff.SetOperationToSubtract()
diff.SetInputConnection(0,reader.GetOutputPort())
diff.SetInputConnection(1,reader2.GetOutputPort())
diff.Update()
diffrange = diff.GetOutput().GetScalarRange()
differr = diffrange[0]**2 + diffrange[1]**2
return differr
def write(objct, fileoutput, binary=True):
"""
Write 3D object to file. (same as `save()`).
Possile extensions are:
- vtk, vti, npy, ply, obj, stl, byu, vtp, vti, mhd, xyz, tif, png, bmp.
"""
obj = objct
if isinstance(obj, Mesh): # picks transformation
obj = objct.polydata(True)
elif isinstance(obj, (vtk.vtkActor, vtk.vtkVolume)):
obj = objct.GetMapper().GetInput()
elif isinstance(obj, (vtk.vtkPolyData, vtk.vtkImageData)):
obj = objct
fr = fileoutput.lower()
if fr.endswith(".vtk"):
writer = vtk.vtkPolyDataWriter()
elif fr.endswith(".ply"):
writer = vtk.vtkPLYWriter()
pscal = obj.GetPointData().GetScalars()
if not pscal:
pscal = obj.GetCellData().GetScalars()
if pscal and pscal.GetName():
writer.SetArrayName(pscal.GetName())
#writer.SetColorMode(0)
lut = objct.GetMapper().GetLookupTable()
if lut:
writer.SetLookupTable(lut)
elif fr.endswith(".stl"):
writer = vtk.vtkSTLWriter()
elif fr.endswith(".vtp"):
writer = vtk.vtkXMLPolyDataWriter()
elif fr.endswith(".vtm"):
g = vtk.vtkMultiBlockDataGroupFilter()
for ob in objct:
if isinstance(ob, Mesh): # picks transformation
ob = ob.polydata(True)
elif isinstance(ob, (vtk.vtkActor, vtk.vtkVolume)):
ob = ob.GetMapper().GetInput()
g.AddInputData(ob)
g.Update()
mb = g.GetOutputDataObject(0)
wri = vtk.vtkXMLMultiBlockDataWriter()
wri.SetInputData(mb)
wri.SetFileName(fileoutput)
wri.Write()
return mb
elif fr.endswith(".xyz"):
writer = vtk.vtkSimplePointsWriter()
elif fr.endswith(".facet"):
writer = vtk.vtkFacetWriter()
elif fr.endswith(".tif"):
writer = vtk.vtkTIFFWriter()
writer.SetFileDimensionality(len(obj.GetDimensions()))
elif fr.endswith(".vti"):
writer = vtk.vtkXMLImageDataWriter()
elif fr.endswith(".mhd"):
writer = vtk.vtkMetaImageWriter()
elif fr.endswith(".nii"):
writer = vtk.vtkNIFTIImageWriter()
elif fr.endswith(".png"):
writer = vtk.vtkPNGWriter()
elif fr.endswith(".jpg"):
writer = vtk.vtkJPEGWriter()
elif fr.endswith(".bmp"):
writer = vtk.vtkBMPWriter()
elif fr.endswith(".npy"):
if utils.isSequence(objct):
objslist = objct
else:
objslist = [objct]
dicts2save = []
for obj in objslist:
dicts2save.append( _np_dump(obj) )
np.save(fileoutput, dicts2save)
return dicts2save
elif fr.endswith(".obj"):
outF = open(fileoutput, "w")
outF.write('# OBJ file format with ext .obj\n')
outF.write('# File Created by vtkplotter\n')
cobjct = objct.clone().clean()
for p in cobjct.points():
outF.write('v '+ str(p[0]) +" "+ str(p[1])+" "+ str(p[2])+'\n')
for vn in cobjct.normals(cells=False):
outF.write('vn '+str(vn[0])+" "+str(vn[1])+" "+str(vn[2])+'\n')
#pdata = cobjct.polydata().GetPointData().GetScalars()
#if pdata:
# ndata = vtk_to_numpy(pdata)
# for vd in ndata:
# outF.write('vp '+ str(vd) +'\n')
#ptxt = cobjct.polydata().GetPointData().GetTCoords() # not working
#if ptxt:
# ntxt = vtk_to_numpy(ptxt)
# print(len(cobjct.faces()), cobjct.points().shape, ntxt.shape)
# for vt in ntxt:
# outF.write('vt '+ str(vt[0]) +" "+ str(vt[1])+ ' 0\n')
for f in cobjct.faces():
fs = ''
for fi in f:
fs += " "+str(fi+1)
outF.write('f' + fs + '\n')
#ldata = cobjct.polydata().GetLines().GetData()
#print(cobjct.polydata().GetLines())
#if ldata:
# ndata = vtk_to_numpy(ldata)
# print(ndata)
# for l in ndata:
# ls = ''
# for li in l:
# ls += str(li+1)+" "
# outF.write('l '+ ls + '\n')
outF.close()
return objct
elif fr.endswith(".xml"): # write tetrahedral dolfin xml
vertices = objct.points().astype(str)
faces = np.array(objct.faces()).astype(str)
ncoords = vertices.shape[0]
outF = open(fileoutput, "w")
outF.write('<?xml version="1.0" encoding="UTF-8"?>\n')
outF.write('<dolfin xmlns:dolfin="http://www.fenicsproject.org">\n')
if len(faces[0]) == 4:# write tetrahedral mesh
ntets = faces.shape[0]
outF.write(' <mesh celltype="tetrahedron" dim="3">\n')
outF.write(' <vertices size="' + str(ncoords) + '">\n')
for i in range(ncoords):
x, y, z = vertices[i]
outF.write(' <vertex index="'+str(i)+'" x="'+x+'" y="'+y+'" z="'+z+'"/>\n')
outF.write(' </vertices>\n')
outF.write(' <cells size="' + str(ntets) + '">\n')
for i in range(ntets):
v0, v1, v2, v3 = faces[i]
outF.write(' <tetrahedron index="'+str(i)
+ '" v0="'+v0+'" v1="'+v1+'" v2="'+v2+'" v3="'+v3+'"/>\n')
elif len(faces[0]) == 3:# write triangle mesh
ntri = faces.shape[0]
outF.write(' <mesh celltype="triangle" dim="2">\n')
outF.write(' <vertices size="' + str(ncoords) + '">\n')
for i in range(ncoords):
x, y, dummy_z = vertices[i]
outF.write(' <vertex index="'+str(i)+'" x="'+x+'" y="'+y+'"/>\n')
outF.write(' </vertices>\n')
outF.write(' <cells size="' + str(ntri) + '">\n')
for i in range(ntri):
v0, v1, v2 = faces[i]
outF.write(' <triangle index="'+str(i)+'" v0="'+v0+'" v1="'+v1+'" v2="'+v2+'"/>\n')
outF.write(' </cells>\n')
outF.write(" </mesh>\n")
outF.write("</dolfin>\n")
outF.close()
return objct
else:
colors.printc("~noentry Unknown format", fileoutput, "file not saved.", c="r")
return objct
try:
if hasattr(writer, 'SetFileTypeToBinary'):
if binary:
writer.SetFileTypeToBinary()
else:
writer.SetFileTypeToASCII()
writer.SetInputData(obj)
writer.SetFileName(fileoutput)
writer.Write()
except Exception as e:
colors.printc("~noentry Error saving: " + fileoutput, "\n", e, c="r")
return objct
BrainMask.SetScalarComponentFromFloat(x, y, z, 0, 1)
else:
BrainMask.SetScalarComponentFromFloat(x, y, z, 0, 0)
voxelValueBrainMask = BrainMask.GetScalarComponentAsFloat(
x, y, z, 0)
if voxelValueBrainMask == 1:
HeadMask.SetScalarComponentFromFloat(x, y, z, 0, 0)
voxelValueHipoLeft = HipoLeft.GetOutput(
).GetScalarComponentAsFloat(x, y, z, 0)
voxelValueHipoRight = HipoRight.GetOutput(
).GetScalarComponentAsFloat(x, y, z, 0)
if voxelValueHipoLeft != 0 or voxelValueHipoRight != 0:
BrainMask.SetScalarComponentFromFloat(x, y, z, 0, 0)
#............................................................................................
#Write resulting masks
writer1 = vtk.vtkNIFTIImageWriter()
writer1.SetInputData(HeadMask)
writer1.SetFileName("HeadMask.nii")
writer1.Write()
writer2 = vtk.vtkNIFTIImageWriter()
writer2.SetInputData(BrainMask)
writer2.SetFileName("BrainMask.nii")
writer2.Write()
#Define a function that takes an image and returns a same structured image but with zero values everywhere
def ZeroImage(image):
image1 = vtk.vtkImageData()
image1.DeepCopy(image)
for z in range(0, image.GetDimensions()[2]):
for y in range(0, image.GetDimensions()[1]):
rightExtractor.SetVOI(rightVOI)
print 'Extractiong right subvolume'
rightExtractor.Update()
# Extract right
leftExtractor = vtk.vtkExtractVOI()
leftExtractor.SetInputConnection(reader.GetOutputPort())
leftExtractor.SetVOI(leftVOI)
print 'Extractiong left subvolume'
leftExtractor.Update()
# Flip image (left becomes right)
leftFlipper = vtk.vtkImageFlip()
leftFlipper.SetInputConnection(leftExtractor.GetOutputPort())
leftFlipper.SetFilteredAxis(0)
print 'Flipping left subvolume'
leftFlipper.Update()
# Write data out
rightWriter = vtk.vtkNIFTIImageWriter()
rightWriter.SetInputConnection(rightExtractor.GetOutputPort())
rightWriter.SetFileName(args.outputRightFemurImageFile)
print "Writing {fileName}".format(fileName=args.outputRightFemurImageFile)
rightWriter.Write()
leftWriter = vtk.vtkNIFTIImageWriter()
leftWriter.SetInputConnection(leftFlipper.GetOutputPort())
leftWriter.SetFileName(args.outputLeftFemurImageFile)
print "Writing {fileName}".format(fileName=args.outputLeftFemurImageFile)
leftWriter.Write()
print('\tmu_scaling: {}'.format(mu_scaling))
print('\thu_mu_water: {}'.format(hu_mu_water))
# First convert Native -> mu (linear attenuation)
# mu = Native / mu_scaling
# Second convert mu -> HU
# HU = 1000 * (mu - mu_water) / (mu_water - mu_air)
m = 1000.0 / mu_scaling / (hu_mu_water - hu_mu_air)
b = -1000.0 * hu_mu_water / (hu_mu_water - hu_mu_air)
print "Proc. log eq: {0} * x {1:+f}".format(m, b)
print "Proc. log eq: {0} * (x {1:+f})".format(m, b / m)
# Apply shift
caster = vtk.vtkImageShiftScale()
caster.SetOutputScalarTypeToFloat()
caster.ClampOverflowOn()
caster.SetInputConnection(reader.GetOutputPort())
caster.SetShift(b / m)
caster.SetScale(m)
print("Shifting and scaling input...")
caster.Update()
# Write outputImage
writer = vtk.vtkNIFTIImageWriter()
writer.SetFileName(args.outputImage)
writer.SetInputConnection(caster.GetOutputPort())
print("Writing to {}".format(args.outputImage))
writer.Update()
def filterSmooth(inputReader):
inputImage = inputReader.GetOutput()
imagePath = imageFile
# determes the filename for the output file based on filter type and input
if (os.path.isdir(imagePath)) and filterType == "gaussian":
# outfile = "Thorax_GaussianSmoothed"
outfile = str(os.path.basename(imagePath)) + '_GaussianSmoothed.nii'
elif (os.path.isdir(imagePath)) and filterType == "median":
# outfile = "Thorax_MedianSmoothed"
outfile = str(os.path.basename(imagePath)) + '_MedianSmoothed.nii'
elif filterType == "gaussian":
outfile = str(os.path.splitext(imagePath)[0]) + '_GaussianSmoothed.nii'
elif filterType == "median":
outfile = str(os.path.splitext(imagePath)[0]) + '_MedianSmoothed.nii'
## used for debugging
# outpath = os.path.join(os.getcwd(), outfile)
# print('\nOutfile: ' + outfile)
# print('\nOutpath: ' + outpath + '\n')
# make a copy of the input image(s) to work with and maintain original data
copyImage = vtk.vtkImageData()
copyImage.DeepCopy(inputReader.GetOutput())
# get the image dimensions
imageDim = inputImage.GetDimensions()
imageExt = inputImage.GetExtent()
kernelSize = 3 # could make this an input arg in the future
imageDimMin = np.empty(
3, dtype=int
) # will have the starting coordinates for the image we want to smooth (padded)
imageDimMax = np.empty(
3, dtype=int
) # will have the ending coordinates for the images we want to smooth (padded)
## used for debugging
# print("imagedim= " + str(imageDim))
# print("image extents= " + str(imageExt))
# print("length of image ext= " + str(len(imageExt)))
# dummy variables
a = 0
b = 0
# set up the dimenions of the image we are analyzing to help iterate over the image's data
# accounts for the size of the smoothing kernel (i.e. implementing padding)
for i in range(
0, len(imageExt), 2
): # 0 to 4 by twos to get all even numbers, corresponds to x y z min coords
imageDimMin[a] = imageExt[i] + (
kernelSize - kernelSize // 2
) # creates a list of the image's min coordinates to start smoothing from
a += 1
for i in range(
1, (len(imageExt) + 1), 2
): # 1 to 5 by twos to get all uneven numbers, corresponds to x y z max coords
imageDimMax[b] = imageExt[i] - (
kernelSize - kernelSize // 2
) # creates a list of the image's max coordinates to stop smoothing at
b += 1
## used for debugging
# print("image dim min, max= " + str(imageDimMin) + str(imageDimMax))
# 3D Gaussian kernel, hard coded. Eventually change this to match the size of the array requested i.e. 3x3, 5x5, etc
kernel = np.array([[[1, 2, 1], [2, 4, 2], [1, 2, 1]],
[[2, 4, 2], [4, 16, 4], [2, 4, 2]],
[[1, 2, 1], [2, 4, 2], [1, 2, 1]]])
# kernelNormal normalizes each entry in the array. If all normalized entries are summed, the answer is 1.0
# will be used as the gaussian kernel for fitering
kernelNormal = kernel / np.sum(kernel)
# used for testing to speed up processing. Set start and end Z-Slices to filter.
imageDimMin[2] = 120
imageDimMax[2] = 130
# Itereate over all admissiblie pixels in a given image
for x in range(imageDimMin[0] - 1, imageDimMax[0] + 1):
for y in range(imageDimMin[1] - 1, imageDimMax[1] + 1):
for z in range(imageDimMin[2] - 1, imageDimMax[2]):
# a sub array of the image being filtered. Should be the size of the kernel
subArray = np.empty((kernelSize, kernelSize, kernelSize),
dtype=float)
# iterate through values to append to the subArray based on 3x3x3 filter
# the created sub array is composed to original image values and will be used to
# apply the kernel to, and determine the filtered voxel value(s)
l = 0 # dummy i index value
for i in range(x - 1, x + 2):
m = 0 # dummy j index value
l += 1
for j in range(y - 1, y + 2):
n = 0 # dummy k index value
m += 1
for k in range(z - 1, z + 2):
n += 1
voxelValue = inputImage.GetScalarComponentAsFloat(
i, j, k, 0)
subArray[l - 1][m - 1][n - 1] = voxelValue
if filterType == "gaussian":
# gives the value of the central voxel by summing all the values from the kernelNormal*subArray convolution
# this is the gaussian blurring or smoothing step
voxelBlurred = np.sum(kernelNormal * subArray)
copyImage.SetScalarComponentFromFloat(
x, y, z, 0, voxelBlurred)
elif filterType == "median":
# gives the value of the central voxel by determining the median value of the subArray
# this is the median blurring or smoothing step
voxelMedian = np.median(subArray)
copyImage.SetScalarComponentFromFloat(
x, y, z, 0, voxelMedian)
# print('. ')
# write the new filtered image as a NIFTI filetype to the working directory
# I'm not sure how to do this for DICOM
writer = vtk.vtkNIFTIImageWriter()
writer.SetFileName(outfile)
writer.SetInputData(copyImage)
writer.Write()
# updates the vtk reader to the new filtered image
smoothedImage = vtk.vtkNIFTIImageReader()
smoothedImage.SetFileName(outfile)
smoothedImage.Update()
return (smoothedImage)
reader = vtk.vtkNIFTIImageReader()
reader.SetFileName(VTK_DATA_ROOT + "/PASIM_N002.nii")
reader.Update()
size = reader.GetOutput().GetDimensions()
# Create Threshold
Threshold = vtk.vtkImageThreshold()
Threshold.SetInputConnection(reader.GetOutputPort())
Threshold.ThresholdBetween(IntensityValue, IntensityValue)
Threshold.SetInValue(255)
Threshold.SetOutValue(0)
Threshold.SetOutputScalarTypeToUnsignedChar()
Threshold.Update()
# Write Threshold Image
writer = vtk.vtkNIFTIImageWriter()
writer.SetInputConnection(Threshold.GetOutputPort())
writer.SetFileName("BinaryImage.nii")
writer.Write()
# Calculate Euclidean distance
dist = vtk.vtkImageEuclideanDistance()
dist.SetInputConnection(Threshold.GetOutputPort())
dist.SetAlgorithmToSaitoCached()
dist.Update()
# Performs Square root to speed up processing time
mathSqrt = vtk.vtkImageMathematics()
mathSqrt.SetInputConnection(dist.GetOutputPort())
mathSqrt.SetOperationToSquareRoot()
mathSqrt.Update()
def test_get_nii_gz(self):
'''nii.gz filetype returns None'''
expected = type(vtk.vtkNIFTIImageWriter())
self.assertEqual(type(get_vtk_writer('test.nIi.gZ')), expected)
reader = vtk.vtkNIFTIImageReader()
reader.SetFileName(name)
reader.Update()
print "type is: ",reader.GetDataScalarType()
matrix = reader.GetQFormMatrix()
print "Ori QForm Matrix is",matrix
caster = vtk.vtkImageCast()
caster.SetInputData(reader.GetOutput())
caster.SetOutputScalarTypeToInt()
caster.Update()
i2 = "new"+i
print "writing to", i2
writer = vtk.vtkNIFTIImageWriter()
#writer.SetQFac(-1)
writer.SetQFormMatrix(matrix)
writer.SetFileName(folder_path+i2)
writer.SetInputData(caster.GetOutput())
writer.Update()
else:
pass
print "Done!!"
HipoCrop = ImageCropper(image, Array[0], Array[1], Array[2], Array[3],
Array[4], Array[5])
print("croppedimage..............")
print(HipoCrop.GetExtent())
print("origImage.................")
print(image.GetExtent())
pad = vtk.vtkImageConstantPad()
pad.SetInputData(HipoCrop)
pad.SetOutputWholeExtent(-Array[0],
image.GetExtent()[1] - Array[0], -Array[2],
image.GetExtent()[3] - Array[2], -Array[4],
image.GetExtent()[5] - Array[4])
pad.SetConstant(0)
pad.Update()
writer6 = vtk.vtkNIFTIImageWriter()
writer6.SetInputConnection(pad.GetOutputPort())
writer6.SetFileName("SegmentedNoPostProcess_Hipo.nii")
writer6.Write()
reader1 = vtk.vtkNIFTIImageReader()
reader1.SetFileName("SegmentedNoPostProcess_Hipo.nii")
reader1.Update()
image1 = reader.GetOutput()
print("finalImage.................")
print(reader1.GetOutput().GetExtent())
reslice = vtk.vtkImageReslice()
reslice.SetInputConnection(reader1.GetOutputPort())
#reslice.SetInformationInput(reader.GetOutput())
ymax=max(yvector) Array=[xmin,xmax,ymin,ymax,zmin,zmax] return(Array) Array=TightBounder(im1) X1=Array[0] X2=Array[1] Y1=Array[2] Y2=Array[3] Z1=Array[4] Z2=Array[5] im1crop=ImageCropper(im1,X1,X2,Y1,Y2,Z1,Z2) im2crop=ImageCropper(im2,X1,X2,Y1,Y2,Z1,Z2) #im3crop=ImageCropper(im3,X1,X2,Y1,Y2,Z1,Z2) writer1=vtk.vtkNIFTIImageWriter() writer1.SetInputData(im1crop) writer1.SetFileName(args.output1) writer1.Write() writer2=vtk.vtkNIFTIImageWriter() writer2.SetInputData(im2crop) writer2.SetFileName(args.output2) writer2.Write()
def mesh_2_mask(
inputMesh, outputImage, inputImage=None, superRes=False, spacing=(1.0, 1.0, 1.0)
):
"""
This program takes in a closed 3D surface, vtkPolyData, and converts it into volume
representation (vtkImageData) where the foreground voxels are 1 and the background voxels
are 0. Internally vtkPolyDataToImageStencil is utilized. The resultant image is saved to disk
in NIFTIimage file format. SimpleITK is used to convert images to standard orientation used
for 3D medical images.
:param inputMesh: a vtkPolyData file of a 3D surface
:param outputImage: output file path for NIFTI image
:param inputImage: reference image to get desired spacing, origin, and direction.
:param superRes:
:param spacing:
:return:
"""
VTK_MAJOR_VERSION = str(vtk.vtkVersion().GetVTKVersion()).split(".")[0]
outputImage = str(outputImage)
if inputImage:
inputImage = str(inputImage)
# check output image extension
out_ext = outputImage.split(".", 1)[-1]
if "nii" in out_ext:
writer = vtk.vtkNIFTIImageWriter()
else:
print(
"ERROR: Output must be NIFTI image file. \n\tUnrecognized extension: {0}".format(
out_ext
)
)
return sys.exit(os.EX_IOERR)
if inputImage:
image = read_vtk_image(inputImage)
else:
image = None
# read the mesh in
pd = read_poly_data(inputMesh)
pd = preprocess_mesh(pd)
# allocate whiteImage
whiteImage = vtk.vtkImageData()
# polygonal data -. image stencil:
pol2stenc = vtk.vtkPolyDataToImageStencil()
if VTK_MAJOR_VERSION <= 5:
pol2stenc.SetInput(pd)
else:
pol2stenc.SetInputData(pd)
# get the bounds
bounds = pd.GetBounds()
if image:
spacing = image.GetSpacing()
dim = image.GetDimensions()
# set VTK direction to RAS
vtk_direction = (-1.0, 0.0, 0.0, 0.0, -1.0, 0.0, 0.0, 0.0, 1.0)
# vtk does not get the correct origin
# origin = image.GetOrigin ()
# use SimpleITK instead
image_sitk = sitk.ReadImage(inputImage)
origin = image_sitk.GetOrigin()
direction = image_sitk.GetDirection()
print(direction)
print(origin)
print(spacing)
# superRes slows down the script quite a bit
if superRes:
"""Creates an image with pixels a fourth the size of the original
This helps allivaite some of the partial voluming effect that
can take place."""
denom = 2
spacing = (
spacing[0] / float(denom),
spacing[1] / float(denom),
spacing[2] / float(denom),
)
dim = (dim[0] * denom, dim[1] * denom, dim[2] * denom)
# VTKImages seem to always have the same direction
origin = (
origin[0] * vtk_direction[0],
origin[1] * vtk_direction[4],
origin[2] * vtk_direction[8],
)
if direction != vtk_direction:
if direction == (1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0):
origin = (
origin[0] - spacing[0] * (dim[0] - 1),
origin[1] - spacing[1] * (dim[1] - 1),
origin[2],
)
else:
print("ERROR: Not sure how to handle input image direction")
sys.exit()
print(origin)
else:
if superRes:
spacing = (
spacing[0] / float(2),
spacing[1] / float(2),
spacing[2] / float(2),
)
# compute dimensions
dim = [0, 0, 0]
for i in range(3):
dim[i] = int(math.ceil((bounds[i * 2 + 1] - bounds[i * 2]) / spacing[i]))
dim = tuple(dim)
# get origin
origin = [0, 0, 0]
origin[0] = bounds[0] + spacing[0] / float(2)
origin[1] = bounds[2] + spacing[1] / float(2)
origin[2] = bounds[4] + spacing[2] / float(2)
origin = tuple(origin)
whiteImage.SetSpacing(spacing)
whiteImage.SetOrigin(origin)
pol2stenc.SetOutputOrigin(origin)
pol2stenc.SetOutputSpacing(spacing)
# set dimensions
whiteImage.SetDimensions(dim)
whiteImage.SetExtent(0, dim[0] - 1, 0, dim[1] - 1, 0, dim[2] - 1)
if VTK_MAJOR_VERSION <= 5:
whiteImage.SetScalarTypeToUnsignedChar()
whiteImage.AllocateScalars()
else:
whiteImage.AllocateScalars(vtk.VTK_UNSIGNED_CHAR, 1)
# fill the image with foreground voxels:
inval = 1
outval = 0
count = whiteImage.GetNumberOfPoints()
for i in range(count):
whiteImage.GetPointData().GetScalars().SetTuple1(i, inval)
# update pol2stenc
pol2stenc.SetOutputWholeExtent(whiteImage.GetExtent())
pol2stenc.Update()
# cut the corresponding white image and set the background:
imgstenc = vtk.vtkImageStencil()
if VTK_MAJOR_VERSION <= 5:
imgstenc.SetInput(whiteImage)
imgstenc.SetStencil(pol2stenc.GetOutput())
else:
imgstenc.SetInputData(whiteImage)
imgstenc.SetStencilConnection(pol2stenc.GetOutputPort())
imgstenc.ReverseStencilOff()
imgstenc.SetBackgroundValue(outval)
imgstenc.Update()
# write image to file
writer.SetFileName(outputImage)
if inputImage != None and direction != vtk_direction:
flipFilter = vtk.vtkImageFlip()
flipFilter.SetFilteredAxis(1) # flip y axis
flipFilter.SetInputData(imgstenc.GetOutput())
flipFilter.SetFlipAboutOrigin(1)
flipFilter.Update()
flipFilter2 = vtk.vtkImageFlip()
flipFilter2.SetFilteredAxis(0) # flip x axis
flipFilter2.SetInputData(flipFilter.GetOutput())
flipFilter2.SetFlipAboutOrigin(1)
flipFilter2.Update()
if VTK_MAJOR_VERSION <= 5:
writer.SetInput(flipFilter2.GetOutput())
else:
writer.SetInputData(flipFilter2.GetOutput())
writer.Write()
itk_image = sitk.ReadImage(inputImage)
out_image = sitk.ReadImage(outputImage)
out_image.SetDirection(itk_image.GetDirection())
out_image.SetOrigin(itk_image.GetOrigin())
sitk.WriteImage(out_image, outputImage)
else:
if VTK_MAJOR_VERSION <= 5:
writer.SetInput(imgstenc.GetOutput())
else:
writer.SetInputData(imgstenc.GetOutput())
writer.Write()
return os.path.abspath(outputImage)
def vtkExtractOuterSurface(filename):
reader = vtk.vtkNIFTIImageReader()
reader.SetFileName(filename)
reader.Update()
image1 = reader.GetOutput()
dim = image1.GetDimensions()
print dim
for x in [0, dim[0] - 1]:
for y in [0, dim[1] - 1]:
image1.SetScalarComponentFromFloat(x, y, dim[2] - 1, 0, 0.0)
image1.SetScalarComponentFromFloat(x, y, dim[2] - 2, 0, 0.0)
image1.SetScalarComponentFromFloat(x, y, dim[2] - 3, 0, 0.0)
image1.SetScalarComponentFromFloat(x, y, dim[2] - 4, 0, 0.0)
image1.SetScalarComponentFromFloat(x, y, 0, 0, 0.0)
image1.SetScalarComponentFromFloat(x, y, 1, 0, 0.0)
image1.SetScalarComponentFromFloat(x, y, 2, 0, 0.0)
image1.SetScalarComponentFromFloat(x, y, 3, 0, 0.0)
for y in [0, dim[1] - 1]:
for z in [0, dim[2] - 1]:
image1.SetScalarComponentFromFloat(dim[0] - 1, y, z, 0, 0.0)
image1.SetScalarComponentFromFloat(dim[0] - 2, y, z, 0, 0.0)
image1.SetScalarComponentFromFloat(dim[0] - 3, y, z, 0, 0.0)
image1.SetScalarComponentFromFloat(dim[0] - 4, y, z, 0, 0.0)
image1.SetScalarComponentFromFloat(dim[0] - 5, y, z, 0, 0.0)
image1.SetScalarComponentFromFloat(0, y, z, 0, 0.0)
image1.SetScalarComponentFromFloat(1, y, z, 0, 0.0)
image1.SetScalarComponentFromFloat(2, y, z, 0, 0.0)
image1.SetScalarComponentFromFloat(3, y, z, 0, 0.0)
for x in [0, dim[0] - 1]:
for z in [0, dim[2] - 1]:
image1.SetScalarComponentFromFloat(x, 0, z, 0, 0.0)
image1.SetScalarComponentFromFloat(x, 1, z, 0, 0.0)
image1.SetScalarComponentFromFloat(x, 2, z, 0, 0.0)
image1.SetScalarComponentFromFloat(x, 3, z, 0, 0.0)
image1.SetScalarComponentFromFloat(x, 4, z, 0, 0.0)
image1.SetScalarComponentFromFloat(x, 5, z, 0, 0.0)
image1.SetScalarComponentFromFloat(x, dim[1] - 1, z, 0, 0.0)
image1.SetScalarComponentFromFloat(x, dim[1] - 2, z, 0, 0.0)
image1.SetScalarComponentFromFloat(x, dim[1] - 3, z, 0, 0.0)
image1.SetScalarComponentFromFloat(x, dim[1] - 4, z, 0, 0.0)
image1.SetScalarComponentFromFloat(x, dim[1] - 5, z, 0, 0.0)
# threshold
threshold = vtk.vtkImageThreshold()
threshold.SetInputData(image1)
threshold.ThresholdBetween(15, 1000)
threshold.ReplaceInOn()
threshold.SetInValue(200)
threshold.ReplaceOutOn()
threshold.SetOutValue(0)
threshold.Update()
image = threshold.GetOutput()
# write to image
writer = vtk.vtkNIFTIImageWriter()
writer.SetFileName("C:/Users/QIN Shuo/Desktop/test.nii")
writer.SetInputData(image)
writer.Update()
# marching here
marching = vtk.vtkMarchingCubes()
marching.SetInputData(image)
marching.SetValue(0, 8)
marching.Update()
skin = marching.GetOutput()
connector = vtk.vtkConnectivityFilter()
connector.SetInputData(skin)
connector.SetExtractionModeToLargestRegion()
connector.Update()
geo = vtk.vtkGeometryFilter()
geo.SetInputData(connector.GetOutput())
geo.Update()
area = geo.GetOutput()
## fill holes
# filler = vtk.vtkFillHolesFilter()
# filler.SetInputData(area)
# filler.SetHoleSize(1E6)
# filler.Update()
# area = filler.GetOutput()
# filter = vtk.vtkPolyDataConnectivityFilter()
# filter.SetInputData(area)
# filter.SetExtractionModeToSpecifiedRegions()
# filter.AddSpecifiedRegion(0)
# filter.Update()
# area = filter.GetOutput()
# visualization
print "visualizing..."
mapper = vtk.vtkPolyDataMapper()
mapper.SetInputData(area)
# mapper.ScalarVisibilityOn()
actor = vtk.vtkActor()
actor.SetMapper(mapper)
# actor.GetProperty().SetOpacity(0.7)
renderer = vtk.vtkRenderer()
renderer.AddActor(actor)
window = vtk.vtkRenderWindow()
window.AddRenderer(renderer)
interactor = vtk.vtkRenderWindowInteractor()
window.SetInteractor(interactor)
window.Render()
interactor.Start()
def mesh_2_mask(inputMesh,
outputImage,
inputImage=None,
superRes=False,
spacing=(1.0, 1.0, 1.0)):
"""
This program takes in a closed 3D surface, vtkPolyData, and converts it into volume
representation (vtkImageData) where the foreground voxels are 1 and the background voxels
are 0. Internally vtkPolyDataToImageStencil is utilized. The resultant image is saved to disk
in NIFTIimage file format. SimpleITK is used to convert images to standard orientation used
for 3D medical images.
:param inputMesh: a vtkPolyData file of a 3D surface
:param outputImage: output file path for NIFTI image
:param inputImage: reference image to get desired spacing, origin, and direction.
:param superRes:
:param spacing:
:return:
"""
VTK_MAJOR_VERSION = str(vtk.vtkVersion().GetVTKVersion()).split(".")[0]
outputImage = str(outputImage)
if inputImage:
inputImage = str(inputImage)
# check output image extension
out_ext = outputImage.split(".", 1)[-1]
if "nii" in out_ext:
writer = vtk.vtkNIFTIImageWriter()
else:
print(
"ERROR: Output must be NIFTI image file. \n\tUnrecognized extension: {0}"
.format(out_ext))
return sys.exit(os.EX_IOERR)
if inputImage:
image = read_vtk_image(inputImage)
else:
image = None
# read the mesh in
pd = read_poly_data(inputMesh)
pd = preprocess_mesh(pd)
# allocate whiteImage
whiteImage = vtk.vtkImageData()
# polygonal data -. image stencil:
pol2stenc = vtk.vtkPolyDataToImageStencil()
if VTK_MAJOR_VERSION <= 5:
pol2stenc.SetInput(pd)
else:
pol2stenc.SetInputData(pd)
# get the bounds
bounds = pd.GetBounds()
if image:
spacing = image.GetSpacing()
dim = image.GetDimensions()
# set VTK direction to RAS
vtk_direction = (-1.0, 0.0, 0.0, 0.0, -1.0, 0.0, 0.0, 0.0, 1.0)
# vtk does not get the correct origin
# origin = image.GetOrigin ()
# use SimpleITK instead
image_sitk = sitk.ReadImage(inputImage)
origin = image_sitk.GetOrigin()
direction = image_sitk.GetDirection()
print(direction)
print(origin)
print(spacing)
# superRes slows down the script quite a bit
if superRes:
"""Creates an image with pixels a fourth the size of the original
This helps allivaite some of the partial voluming effect that
can take place."""
denom = 2
spacing = (
spacing[0] / float(denom),
spacing[1] / float(denom),
spacing[2] / float(denom),
)
dim = (dim[0] * denom, dim[1] * denom, dim[2] * denom)
# VTKImages seem to always have the same direction
origin = (
origin[0] * vtk_direction[0],
origin[1] * vtk_direction[4],
origin[2] * vtk_direction[8],
)
if direction != vtk_direction:
if direction == (1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0):
origin = (
origin[0] - spacing[0] * (dim[0] - 1),
origin[1] - spacing[1] * (dim[1] - 1),
origin[2],
)
else:
print("ERROR: Not sure how to handle input image direction")
sys.exit()
print(origin)
else:
if superRes:
spacing = (
spacing[0] / float(2),
spacing[1] / float(2),
spacing[2] / float(2),
)
# compute dimensions
dim = [0, 0, 0]
for i in range(3):
dim[i] = int(
math.ceil((bounds[i * 2 + 1] - bounds[i * 2]) / spacing[i]))
dim = tuple(dim)
# get origin
origin = [0, 0, 0]
origin[0] = bounds[0] + spacing[0] / float(2)
origin[1] = bounds[2] + spacing[1] / float(2)
origin[2] = bounds[4] + spacing[2] / float(2)
origin = tuple(origin)
whiteImage.SetSpacing(spacing)
whiteImage.SetOrigin(origin)
pol2stenc.SetOutputOrigin(origin)
pol2stenc.SetOutputSpacing(spacing)
# set dimensions
whiteImage.SetDimensions(dim)
whiteImage.SetExtent(0, dim[0] - 1, 0, dim[1] - 1, 0, dim[2] - 1)
if VTK_MAJOR_VERSION <= 5:
whiteImage.SetScalarTypeToUnsignedChar()
whiteImage.AllocateScalars()
else:
whiteImage.AllocateScalars(vtk.VTK_UNSIGNED_CHAR, 1)
# fill the image with foreground voxels:
inval = 1
outval = 0
count = whiteImage.GetNumberOfPoints()
for i in range(count):
whiteImage.GetPointData().GetScalars().SetTuple1(i, inval)
# update pol2stenc
pol2stenc.SetOutputWholeExtent(whiteImage.GetExtent())
pol2stenc.Update()
# cut the corresponding white image and set the background:
imgstenc = vtk.vtkImageStencil()
if VTK_MAJOR_VERSION <= 5:
imgstenc.SetInput(whiteImage)
imgstenc.SetStencil(pol2stenc.GetOutput())
else:
imgstenc.SetInputData(whiteImage)
imgstenc.SetStencilConnection(pol2stenc.GetOutputPort())
imgstenc.ReverseStencilOff()
imgstenc.SetBackgroundValue(outval)
imgstenc.Update()
# write image to file
writer.SetFileName(outputImage)
if inputImage != None and direction != vtk_direction:
flipFilter = vtk.vtkImageFlip()
flipFilter.SetFilteredAxis(1) # flip y axis
flipFilter.SetInputData(imgstenc.GetOutput())
flipFilter.SetFlipAboutOrigin(1)
flipFilter.Update()
flipFilter2 = vtk.vtkImageFlip()
flipFilter2.SetFilteredAxis(0) # flip x axis
flipFilter2.SetInputData(flipFilter.GetOutput())
flipFilter2.SetFlipAboutOrigin(1)
flipFilter2.Update()
if VTK_MAJOR_VERSION <= 5:
writer.SetInput(flipFilter2.GetOutput())
else:
writer.SetInputData(flipFilter2.GetOutput())
writer.Write()
itk_image = sitk.ReadImage(inputImage)
out_image = sitk.ReadImage(outputImage)
out_image.SetDirection(itk_image.GetDirection())
out_image.SetOrigin(itk_image.GetOrigin())
sitk.WriteImage(out_image, outputImage)
else:
if VTK_MAJOR_VERSION <= 5:
writer.SetInput(imgstenc.GetOutput())
else:
writer.SetInputData(imgstenc.GetOutput())
writer.Write()
return os.path.abspath(outputImage)
def write(objct, fileoutput, binary=True):
"""
Write 3D object to file. (same as `save()`).
Possile extensions are:
- vtk, vti, npy, ply, obj, stl, byu, vtp, vti, mhd, xyz, tif, png, bmp.
"""
obj = objct
if isinstance(obj, Actor): # picks transformation
obj = objct.polydata(True)
elif isinstance(obj, (vtk.vtkActor, vtk.vtkVolume)):
obj = objct.GetMapper().GetInput()
elif isinstance(obj, (vtk.vtkPolyData, vtk.vtkImageData)):
obj = objct
fr = fileoutput.lower()
if ".vtk" in fr:
writer = vtk.vtkPolyDataWriter()
elif ".ply" in fr:
writer = vtk.vtkPLYWriter()
pscal = obj.GetPointData().GetScalars()
if not pscal:
pscal = obj.GetCellData().GetScalars()
if pscal and pscal.GetName():
writer.SetArrayName(pscal.GetName())
#writer.SetColorMode(0)
lut = objct.GetMapper().GetLookupTable()
if lut:
writer.SetLookupTable(lut)
elif ".stl" in fr:
writer = vtk.vtkSTLWriter()
elif ".obj" in fr:
writer = vtk.vtkOBJWriter()
elif ".vtp" in fr:
writer = vtk.vtkXMLPolyDataWriter()
elif ".vtm" in fr:
g = vtk.vtkMultiBlockDataGroupFilter()
for ob in objct:
g.AddInputData(ob)
g.Update()
mb = g.GetOutputDataObject(0)
wri = vtk.vtkXMLMultiBlockDataWriter()
wri.SetInputData(mb)
wri.SetFileName(fileoutput)
wri.Write()
return mb
elif ".xyz" in fr:
writer = vtk.vtkSimplePointsWriter()
elif ".facet" in fr:
writer = vtk.vtkFacetWriter()
elif ".tif" in fr:
writer = vtk.vtkTIFFWriter()
writer.SetFileDimensionality(len(obj.GetDimensions()))
elif ".vti" in fr:
writer = vtk.vtkXMLImageDataWriter()
elif ".mhd" in fr:
writer = vtk.vtkMetaImageWriter()
elif ".nii" in fr:
writer = vtk.vtkNIFTIImageWriter()
elif ".png" in fr:
writer = vtk.vtkPNGWriter()
elif ".jpg" in fr:
writer = vtk.vtkJPEGWriter()
elif ".bmp" in fr:
writer = vtk.vtkBMPWriter()
elif ".npy" in fr:
if utils.isSequence(objct):
objslist = objct
else:
objslist = [objct]
dicts2save = []
for obj in objslist:
dicts2save.append( _np_dump(obj) )
np.save(fileoutput, dicts2save)
return dicts2save
elif ".xml" in fr: # write tetrahedral dolfin xml
vertices = objct.coordinates().astype(str)
faces = np.array(objct.faces()).astype(str)
ncoords = vertices.shape[0]
outF = open(fileoutput, "w")
outF.write('<?xml version="1.0" encoding="UTF-8"?>\n')
outF.write('<dolfin xmlns:dolfin="http://www.fenicsproject.org">\n')
if len(faces[0]) == 4:# write tetrahedral mesh
ntets = faces.shape[0]
outF.write(' <mesh celltype="tetrahedron" dim="3">\n')
outF.write(' <vertices size="' + str(ncoords) + '">\n')
for i in range(ncoords):
x, y, z = vertices[i]
outF.write(' <vertex index="'+str(i)+'" x="'+x+'" y="'+y+'" z="'+z+'"/>\n')
outF.write(' </vertices>\n')
outF.write(' <cells size="' + str(ntets) + '">\n')
for i in range(ntets):
v0, v1, v2, v3 = faces[i]
outF.write(' <tetrahedron index="'+str(i)
+ '" v0="'+v0+'" v1="'+v1+'" v2="'+v2+'" v3="'+v3+'"/>\n')
elif len(faces[0]) == 3:# write triangle mesh
ntri = faces.shape[0]
outF.write(' <mesh celltype="triangle" dim="2">\n')
outF.write(' <vertices size="' + str(ncoords) + '">\n')
for i in range(ncoords):
x, y, dummy_z = vertices[i]
outF.write(' <vertex index="'+str(i)+'" x="'+x+'" y="'+y+'"/>\n')
outF.write(' </vertices>\n')
outF.write(' <cells size="' + str(ntri) + '">\n')
for i in range(ntri):
v0, v1, v2 = faces[i]
outF.write(' <triangle index="'+str(i)+'" v0="'+v0+'" v1="'+v1+'" v2="'+v2+'"/>\n')
outF.write(' </cells>\n')
outF.write(" </mesh>\n")
outF.write("</dolfin>\n")
outF.close()
return objct
else:
colors.printc("~noentry Unknown format", fileoutput, "file not saved.", c="r")
return objct
try:
if hasattr(writer, 'SetFileTypeToBinary'):
if binary:
writer.SetFileTypeToBinary()
else:
writer.SetFileTypeToASCII()
writer.SetInputData(obj)
writer.SetFileName(fileoutput)
writer.Write()
colors.printc("~save Saved file: " + fileoutput, c="g")
except Exception as e:
colors.printc("~noentry Error saving: " + fileoutput, "\n", e, c="r")
return objct
def ImageFilter(input_filename, output_filename, range, overwrite=False):
# Python 2/3 compatible input
from six.moves import input
# Check if output exists and should overwrite
if os.path.isfile(output_filename) and not overwrite:
result = input('File \"{}\" already exists. Overwrite? [y/n]: '.format(
output_filename))
if result.lower() not in ['y', 'yes']:
print('Not overwriting. Exiting...')
os.sys.exit()
# Check valid range
if (range[0] > range[1] or range[0] < 0):
os.sys.exit('[ERROR] Invalid range: {:d} {:d}'.format(
range[0], range[1]))
# Read input
if not os.path.isfile(input_filename):
os.sys.exit('[ERROR] Cannot find file \"{}\"'.format(input_filename))
if input_filename.lower().endswith('.nii'):
reader = vtk.vtkNIFTIImageReader()
elif input_filename.lower().endswith('.nii.gz'):
reader = vtk.vtkNIFTIImageReader()
else:
os.sys.exit('[ERROR] Cannot find reader for file \"{}\"'.format(
input_filename))
print('Reading input image ' + input_filename)
reader.SetFileName(input_filename)
reader.Update()
scalarType = reader.GetOutput().GetScalarType()
print('Input image scalar type: {:s}'.format(
reader.GetOutput().GetScalarTypeAsString()))
print('Input image labels:')
histogram(reader.GetOutput())
thres = vtk.vtkImageThreshold()
thres.SetInputConnection(reader.GetOutputPort())
thres.SetOutputScalarType(scalarType)
thres.ThresholdBetween(1, 10)
thres.ReplaceOutOn()
thres.SetOutValue(0)
thres.Update()
print('Output image labels:')
histogram(thres.GetOutput())
# Create writer
if output_filename.lower().endswith('.nii'):
writer = vtk.vtkNIFTIImageWriter()
elif output_filename.lower().endswith('.nii.gz'):
writer = vtk.vtkNIFTIImageWriter()
else:
os.sys.exit('[ERROR] Cannot find writer for file \"{}\"'.format(
output_filename))
writer.SetInputConnection(thres.GetOutputPort())
writer.SetFileName(output_filename)
writer.SetTimeDimension(reader.GetTimeDimension())
writer.SetTimeSpacing(reader.GetTimeSpacing())
writer.SetRescaleSlope(reader.GetRescaleSlope())
writer.SetRescaleIntercept(reader.GetRescaleIntercept())
writer.SetQFac(reader.GetQFac())
writer.SetQFormMatrix(reader.GetQFormMatrix())
writer.SetNIFTIHeader(reader.GetNIFTIHeader())
print('Saving image ' + output_filename)
writer.Update()
