def test_deprecation_warning_compute_gray_matter_mask():
img = Nifti1Image(np.ones((9, 9, 9)), np.eye(4))
if distutils.version.LooseVersion(sklearn.__version__) < '0.22':
with pytest.deprecated_call():
masking.compute_gray_matter_mask(img)
else:
with pytest.warns(FutureWarning,
match="renamed to 'compute_brain_mask'"):
masking.compute_gray_matter_mask(img)
def extract_gm_signals(self, subj_scan):
if self.gm_mask is None:
self.gm_mask = compute_gray_matter_mask(subj_scan,
smoothing=self.smoothing,
verbose=self.verbosity)
gm_signals = apply_mask(subj_scan,
self.gm_mask,
smoothing=self.smoothing,
verbose=self.verbosity)
return gm_signals
def test_compute_gray_matter_mask():
image = Nifti1Image(np.ones((9, 9, 9)), np.eye(4))
mask = compute_gray_matter_mask(image, threshold=-1)
mask1 = np.zeros((9, 9, 9))
mask1[2:-2, 2:-2, 2:-2] = 1
np.testing.assert_array_equal(mask1, mask.get_data())
# Check that we get a useful warning for empty masks
assert_warns(masking.MaskWarning, compute_gray_matter_mask, image, threshold=1)
# Check that masks obtained from same FOV are the same
img1 = Nifti1Image(np.full((9, 9, 9), np.random.rand()), np.eye(4))
img2 = Nifti1Image(np.full((9, 9, 9), np.random.rand()), np.eye(4))
mask_img1 = compute_gray_matter_mask(img1)
mask_img2 = compute_gray_matter_mask(img2)
np.testing.assert_array_equal(mask_img1.get_data(),
mask_img2.get_data())
def test_compute_gray_matter_mask():
image = Nifti1Image(np.ones((9, 9, 9)), np.eye(4))
mask = compute_gray_matter_mask(image, threshold=-1)
mask1 = np.zeros((9, 9, 9))
mask1[2:-2, 2:-2, 2:-2] = 1
np.testing.assert_array_equal(mask1, get_data(mask))
# Check that we get a useful warning for empty masks
with pytest.warns(masking.MaskWarning):
compute_gray_matter_mask(image, threshold=1)
# Check that masks obtained from same FOV are the same
img1 = Nifti1Image(np.full((9, 9, 9), np.random.rand()), np.eye(4))
img2 = Nifti1Image(np.full((9, 9, 9), np.random.rand()), np.eye(4))
mask_img1 = compute_gray_matter_mask(img1)
mask_img2 = compute_gray_matter_mask(img2)
np.testing.assert_array_equal(get_data(mask_img1), get_data(mask_img2))
def Add_MRI(self, selectors):
if selectors == "brain":
masker = masking.compute_gray_matter_mask(
self.gm_imgs[0]
) #computes full brain mask... misnamed function
else:
Atlas = ATLAS()
masker = Atlas.Mask(selectors)
gm = masking.apply_mask(self.gm_imgs, masker)
wm = masking.apply_mask(self.wm_imgs, masker)
brain = gm + wm
self.masker = masker
self.features = np.hstack((self.features, brain))
def nii2png(niiPath=None,
imgPath=None,
axis='z',
cut=(0, ),
blackBg='auto',
mask=True,
maskThresh=0.8,
maskConnected=True,
maskOpening=3,
verbose=0):
if (os.path.isfile(niiPath) and os.path.splitext[1] == '.nii'):
if (os.path.isdir(os.path.dirname(imgPath))
and not os.path.isdir(imgPath)
and os.path.splitext[1] == '.png'):
imgMask = None
brainImg = image.load_img(niiPath)
if (mask):
imgMask = masking.compute_gray_matter_mask(
brainImg,
threshold=maskThresh,
connected=maskConnected,
opening=maskOpening,
verbose=verbose)
brainImg = image.math_img('img1 * img2',
img1=brainImg,
img2=imgMask)
if (verbose > 0):
print('Outputing image at {}...'.format(imgPath))
plotting.plot_anat(anat_img=niiPath,
display_mode=axis,
output_file=imgPath.format(dataDir),
cut_coords=cut,
annotate=False,
draw_cross=False,
black_bg=blackBg)
elif (not os.path.isdir(os.path.dirname(imgPath))):
print('{} is an invalid directory!!!'.format(
os.path.dirname(imgPath)))
elif (os.path.isdir(imgPath)):
print(
'{} is a directory, not a valid file path!!!'.format(imgPath))
elif (not os.path.splitext[1] == '.png'):
print('{} is not a .png file!!!'.format(os.path.split(imgPath)[1]))
elif (not os.path.isfile(niiPath)):
print('{} is not a valid path!!!'.format(niiPath))
elif (not os.path.splitext(niiPath)[1] == '.nii'):
print('{} is not a .nii file!!!'.format(os.path.split(niiPath)[1]))
def get2Dskeleton(img=None,
axis='z',
sliceIndex=0,
lowThresh=0.21,
highThresh=0.27,
mask=True,
verbose=0):
"""
Gets the 2D skeleton of an image
imgPath -> relative or absolute path to a .png image
lowThresh (optional, default=0.21) -> float between 0 and 1 - low threshold for the hysteresis filter
highThresh (optional, default=0.27) -> float between 0 and 1 - high threshold for the hysteresis filter
crop (optional, default=(15, 143, 30, 158)) -> (x1, x2, y1, y2) Tuple containing the crop coordinates
!!! lowThresh must be lower than highThresh
verbose (optional, default=0) -> int - the higher it is, the more messages will be outputted to the console
"""
img = image.load_img(img)
if (mask):
imgMask = masking.compute_gray_matter_mask(img, verbose=verbose)
img = image.math_img('img1 * img2', img1=img, img2=imgMask)
img = img.get_fdata()
if (axis == 'y'):
img = img[sliceIndex]
elif (axis == 'x'):
img = img[:, sliceIndex, :]
elif (axis == 'z'):
img = img[:, :, sliceIndex]
if (verbose > 0):
print('Applying hysterisis filter...\n')
# Binarize the image with a hysteresis filter
res = filters.apply_hysteresis_threshold(img, lowThresh, highThresh)
# Get the largest connected component of the image
if (verbose > 2):
plotComparison(img, res, 'Original',
'L = ' + str(lowThresh) + " R = " + str(highThresh))
# Skeletonize the image
res = morphology.skeletonize(res)
res = getLargestCC(res)
res = res.astype(int)
#print("Minkowski–Bouligand dimension: ", dimension)
if (verbose > 1):
print('Filtered image:')
plotComparison(img, res, 'Original', 'Skeletonized')
return res
def get3Dskeleton(imgPath=None, thresh='30%', mask=True, verbose=0):
"""
Gets the 3D skeleton of an image
imgPath -> Relative or absolute path to a .nii image or nii variable
thresh (optional, default='30%') -> Threshold value. Either float or string like 'x.xx%'
mask (optional, default=True) -> Applies a gray matter mask if true
verbose (optional, default=0) -> int - the higher it is, the more messages will be outputted to the console
Returns a Nifti1Image object containing the skeletonized image
"""
brain = image.load_img(imgPath)
if (mask):
imgMask = masking.compute_gray_matter_mask(target_img=brain,
threshold=0.3,
connected=True,
opening=1)
brain = image.math_img('img1 * img2', img1=brain, img2=imgMask)
brain = image.threshold_img(brain, thresh)
affine = brain.affine
data = brain.get_data()
data = data / np.amax(data)
#data = filters.threshold_adaptive(data, 35, 10)
skeleton = morphology.skeletonize_3d(data)
skeletonBrain = nifti.Nifti1Image(skeleton, affine)
return skeletonBrain
"""Extract coordinates from images."""
import nilearn
from nilearn import image, masking
from nipy.labs.statistical_mapping import get_3d_peaks
from data_utils import iterator_imgs, retrieve_imgs
template = nilearn.datasets.load_mni152_template()
gray_mask = masking.compute_gray_matter_mask(template)
def get_activations_one_img(img, threshold, verbose=False):
"""
Retrieve the xyz activation coordinates from an image.
Args:
img (Nifti1Image): Nifti1Image from which to extract coordinates.
threshold (float): value below threshold are ignored. Used for
peak detection.
Returns:
(tuple): Size 3 tuple of lists storing respectively the X, Y and
Z coordinates.
"""
if verbose:
print('Extracting')
X, Y, Z = [], [], []
img = image.resample_to_img(img, template)
def nii2csv(niiPath=None,
csvPath=None,
axis='z',
cut=(0, ),
blackBg='auto',
mask=1,
maskThresh=0.5,
maskConnected=True,
maskOpening=3,
verbose=0):
"""
Takes a .nii file and converts it to a .csv file containing a 3D grayscale array
Arguments:
niiPath -> the .nii filepath (relative or absolute)
csvPath -> the filepath of the output image (must end in .png)
axis (optional, default='z') -> 'xyz' 'xy' 'xz' 'yz' 'x' 'y' 'z' - determines the axis along which the cuts are made
cut (optional, default=(0,)) -> list of tuples of 1, 2 or 3 integers - determines the x, y, z position along which the cuts are made
blackBg (optional, default='auto')
mask (optional, default=1) -> int between 0 and 3 - determines the ammount of times a gray matter mask will be applied
maskThresh (optional, default=0.8) -> float between 0 and 1 - determines the threshold used for the gray matter mask
maskConnected (optional, default=True) -> bool - determines if the mask will keep only the largest connected components
maskOpening (optional, default=3) -> int - number of erosions for the morphological opening
verbose (optional, default=0) -> int - the higher it is, the more messages will be outputted to the console
"""
# Checks if the input filepath exists and has a .nii extension
if (os.path.isfile(niiPath) and os.path.splitext(niiPath)[1] == '.nii'):
# Checks if the output filepath's directory is valid, if it is not a directory itself and if it has a .png extension
if (os.path.isdir(os.path.dirname(csvPath))
and not os.path.isdir(csvPath)
and os.path.splitext(csvPath)[1] == '.png'):
imgMask = None
brainImg = image.load_img(niiPath)
# Clamps mask between 0 and 3
mask = min(3, max(0, mask))
# Applies the mask
for i in range(0, mask):
imgMask = masking.compute_gray_matter_mask(
brainImg,
threshold=maskThresh,
connected=maskConnected,
opening=maskOpening,
verbose=verbose)
brainImg = image.math_img('img1 * img2',
img1=brainImg,
img2=imgMask)
if (verbose > 0):
print('Outputing image at {}...'.format(csvPath))
plotting.plot_anat(anat_img=brainImg,
display_mode=axis,
output_file=csvPath,
cut_coords=cut,
annotate=False,
draw_cross=False,
black_bg=blackBg)
elif (not os.path.isdir(os.path.dirname(csvPath))):
print('{} is an invalid directory!!!'.format(
os.path.dirname(csvPath)))
elif (os.path.isdir(csvPath)):
print(
'{} is a directory, not a valid file path!!!'.format(csvPath))
elif (not os.path.splitext(csvPath)[1] == '.png'):
print('{} is not a .png file!!!'.format(os.path.split(csvPath)[1]))
elif (not os.path.isfile(niiPath)):
print('{} is not a valid path!!!'.format(niiPath))
elif (not os.path.splitext(niiPath)[1] == '.nii'):
print('{} is not a .nii file!!!'.format(os.path.split(niiPath)[1]))