def test_binary_deletion_vector():
mr_directory = get_data_directory()
# We will generate data with the following overlap percentages
overlap_percents = [0.0,0.25,0.5,0.75,1.0]
for overlap in overlap_percents:
vector_length = 10000
image_vector1 = numpy.zeros((vector_length))
image_vector2 = numpy.zeros((vector_length))
number_overlap_voxels = int(numpy.floor(overlap*vector_length))
remaining_voxels = int(vector_length - number_overlap_voxels)
idx = range(0,vector_length)
# We break the remaining voxels into 4 groups:
# - nans that will overlap
# - zeros that will overlap (no change to images here, already zeros)
# - nans in image1, random sample of values in image2
# - zeros in image2, random sample of values in image1
group_size = remaining_voxels/4
if overlap != 0.0:
# Here are the overlapping voxels for each image
overlap_idx = range(0,number_overlap_voxels)
image_vector1[overlap_idx] = 1
image_vector2[overlap_idx] = 1
if overlap != 1.0:
# Nans that will overlap
nans_overlap_idx = range(number_overlap_voxels,(number_overlap_voxels+group_size))
image_vector1[nans_overlap_idx] = numpy.nan
image_vector2[nans_overlap_idx] = numpy.nan
# Nans in image1, random sample of values in image 2
start = number_overlap_voxels+group_size
end = number_overlap_voxels+2*group_size
nans_image1 = idx[start:end]
values_image2 = range(nans_image1[-1],(nans_image1[-1] + int(group_size/2)))
image_vector1[nans_image1] = numpy.nan
image_vector2[values_image2] = 0.5
# Zeros in image2, random sample of values in image 1
start = number_overlap_voxels+2*group_size
end = number_overlap_voxels+3*group_size
zeros_image2 = idx[start:end]
values_image1 = range(zeros_image2[-1],(zeros_image2[-1] + int(group_size/2)))
image_vector1[values_image1] = 0.75
# Create nifti images and pdmask
pdmask = make_binary_deletion_vector([image_vector1,image_vector2])
actual_overlap = len(numpy.where(pdmask!=0)[0])
print "Overlap %s percent: should have %s, actual %s" %(overlap,number_overlap_voxels,actual_overlap)
assert_equal(actual_overlap,number_overlap_voxels)
# Also check that is binary
if overlap != 0 and overlap != 1:
assert_equal(numpy.unique(pdmask)[0],0)
assert_equal(numpy.unique(pdmask)[1],1)
if overlap == 0:
assert_equal(numpy.unique(pdmask)[0],0)
if overlap == 1:
assert_equal(numpy.unique(pdmask)[0],1)
def test_binary_deletion_vector():
mr_directory = get_data_directory()
# We will generate data with the following overlap percentages
overlap_percents = [0.0,0.25,0.5,0.75,1.0]
for overlap in overlap_percents:
vector_length = 10000
image_vector1 = numpy.zeros((vector_length))
image_vector2 = numpy.zeros((vector_length))
number_overlap_voxels = int(numpy.floor(overlap*vector_length))
remaining_voxels = int(vector_length - number_overlap_voxels)
idx = list(range(0,vector_length))
# We break the remaining voxels into 4 groups:
# - nans that will overlap
# - zeros that will overlap (no change to images here, already zeros)
# - nans in image1, random sample of values in image2
# - zeros in image2, random sample of values in image1
group_size = old_div(remaining_voxels,4)
if overlap != 0.0:
# Here are the overlapping voxels for each image
overlap_idx = list(range(0,number_overlap_voxels))
image_vector1[overlap_idx] = 1
image_vector2[overlap_idx] = 1
if overlap != 1.0:
# Nans that will overlap
nans_overlap_idx = list(range(number_overlap_voxels,(number_overlap_voxels+group_size)))
image_vector1[nans_overlap_idx] = numpy.nan
image_vector2[nans_overlap_idx] = numpy.nan
# Nans in image1, random sample of values in image 2
start = number_overlap_voxels+group_size
end = number_overlap_voxels+2*group_size
nans_image1 = idx[start:end]
values_image2 = list(range(nans_image1[-1],(nans_image1[-1] + int(old_div(group_size,2)))))
image_vector1[nans_image1] = numpy.nan
image_vector2[values_image2] = 0.5
# Zeros in image2, random sample of values in image 1
start = number_overlap_voxels+2*group_size
end = number_overlap_voxels+3*group_size
zeros_image2 = idx[start:end]
values_image1 = list(range(zeros_image2[-1],(zeros_image2[-1] + int(old_div(group_size,2)))))
image_vector1[values_image1] = 0.75
# Create nifti images and pdmask
pdmask = make_binary_deletion_vector([image_vector1,image_vector2])
actual_overlap = len(numpy.where(pdmask!=0)[0])
print("Overlap %s percent: should have %s, actual %s" %(overlap,number_overlap_voxels,actual_overlap))
assert_equal(actual_overlap,number_overlap_voxels)
# Also check that is binary
if overlap != 0 and overlap != 1:
assert_equal(numpy.unique(pdmask)[0],0)
assert_equal(numpy.unique(pdmask)[1],1)
if overlap == 0:
assert_equal(numpy.unique(pdmask)[0],0)
if overlap == 1:
assert_equal(numpy.unique(pdmask)[0],1)
def save_voxelwise_pearson_similarity_reduced_representation(pk1, pk2):
from neurovault.apps.statmaps.models import Similarity, Comparison
import numpy as np
# We will always calculate Comparison 1 vs 2, never 2 vs 1
if pk1 != pk2:
try:
sorted_images = get_images_by_ordered_id(pk1, pk2)
except Http404:
# files have been deleted in the meantime
return
image1 = sorted_images[0]
image2 = sorted_images[1]
pearson_metric = Similarity.objects.get(
similarity_metric="pearson product-moment correlation coefficient",
transformation="voxelwise")
# Make sure we have a transforms for pks in question
if not image1.reduced_representation or not os.path.exists(
image1.reduced_representation.path):
image1 = save_resampled_transformation_single(
pk1) # cannot run this async
if not image2.reduced_representation or not os.path.exists(
image1.reduced_representation.path):
image2 = save_resampled_transformation_single(
pk2) # cannot run this async
# Load image pickles
image_vector1 = np.load(image1.reduced_representation.file)
image_vector2 = np.load(image2.reduced_representation.file)
# Calculate binary deletion vector mask (find 0s and nans)
mask = make_binary_deletion_vector([image_vector1, image_vector2])
# Calculate pearson
pearson_score = calculate_pairwise_correlation(
image_vector1[mask == 1],
image_vector2[mask == 1],
corr_type="pearson")
# Only save comparison if is not nan
if not numpy.isnan(pearson_score):
Comparison.objects.update_or_create(image1=image1,
image2=image2,
defaults={
'similarity_metric':
pearson_metric,
'similarity_score':
pearson_score
})
return image1.pk, image2.pk, pearson_score
else:
print "Comparison returned NaN."
else:
raise Exception("You are trying to compare an image with itself!")
def save_voxelwise_pearson_similarity_reduced_representation(pk1, pk2):
from neurovault.apps.statmaps.models import Similarity, Comparison
import numpy as np
# We will always calculate Comparison 1 vs 2, never 2 vs 1
if pk1 != pk2:
try:
sorted_images = get_images_by_ordered_id(pk1, pk2)
except Http404:
# files have been deleted in the meantime
return
image1 = sorted_images[0]
image2 = sorted_images[1]
pearson_metric = Similarity.objects.get(similarity_metric="pearson product-moment correlation coefficient",
transformation="voxelwise")
# Make sure we have a transforms for pks in question
if not image1.reduced_representation or not os.path.exists(image1.reduced_representation.path):
image1 = save_resampled_transformation_single(pk1) # cannot run this async
if not image2.reduced_representation or not os.path.exists(image1.reduced_representation.path):
image2 = save_resampled_transformation_single(pk2) # cannot run this async
# Load image pickles
image_vector1 = np.load(image1.reduced_representation.file)
image_vector2 = np.load(image2.reduced_representation.file)
# Calculate binary deletion vector mask (find 0s and nans)
mask = make_binary_deletion_vector([image_vector1,image_vector2])
# Calculate pearson
pearson_score = calculate_pairwise_correlation(image_vector1[mask==1],
image_vector2[mask==1],
corr_type="pearson")
# Only save comparison if is not nan
if not numpy.isnan(pearson_score):
Comparison.objects.update_or_create(image1=image1, image2=image2,
similarity_metric=pearson_metric,
similarity_score=pearson_score)
return image1.pk,image2.pk,pearson_score
else:
print "Comparison returned NaN."
else:
raise Exception("You are trying to compare an image with itself!")