def test_truncate_range():
"""Test range truncation."""
# Given
data = np.random.random(100)
data.ravel()[np.random.choice(data.size, 10, replace=False)] = 0
data.ravel()[np.random.choice(data.size, 5, replace=False)] = np.nan
p_min, p_max = 2.5, 97.5
expected = np.nanpercentile(data, [p_min, p_max])
# When
output, _, _ = truncate_range(data, percMin=p_min, percMax=p_max,
discard_zeros=False)
# Then
assert all(np.nanpercentile(output, [0, 100]) == expected)
def test_truncate_range():
"""Test range truncation."""
# Given
data = np.random.random(100)
data.ravel()[np.random.choice(data.size, 10, replace=False)] = 0
data.ravel()[np.random.choice(data.size, 5, replace=False)] = np.nan
p_min, p_max = 2.5, 97.5
expected = np.nanpercentile(data, [p_min, p_max])
# When
output, _, _ = truncate_range(data,
percMin=p_min,
percMax=p_max,
discard_zeros=False)
# Then
assert all(np.nanpercentile(output, [0, 100]) == expected)
tmp2 = np.zeros((tmp.shape))
for ind2, val in enumerate(uniqueVals):
tmp2[tmp == val] = newVals[ind2]
counter = counter + nrUniqueVals
ncut_labels[:, :, ind] = tmp2
lMax = np.max(ncut_labels)
orig_ncut_labels = ncut_labels.copy()
ima_ncut_labels = ncut_labels.copy()
#
"""Data Processing"""
orig, dims = check_data(nii.get_data(), cfg.force_original_precision)
# Save min and max truncation thresholds to be used in axis labels
orig, pMin, pMax = truncate_range(orig, percMin=cfg.perc_min,
percMax=cfg.perc_max)
# Continue with scaling the original truncated image and recomputing gradient
orig = scale_range(orig, scale_factor=cfg.scale, delta=0.0001)
gra = set_gradient_magnitude(orig, cfg.gramag)
if cfg.export_gramag:
export_gradient_magnitude_image(gra, nii.get_filename(), nii.affine)
# Reshape ima (more intuitive for voxel-wise operations)
ima = np.ndarray.flatten(orig)
gra = np.ndarray.flatten(gra)
#
"""Plots"""
print("Preparing GUI...")
# Plot 2D histogram
fig = plt.figure(facecolor='0.775')
from nibabel import load
from segmentator.utils import map_ima_to_2D_hist, prep_2D_hist
from segmentator.utils import truncate_range, scale_range, check_data
from segmentator.utils import set_gradient_magnitude
from segmentator.utils import export_gradient_magnitude_image
from segmentator.gui_utils import sector_mask, responsiveObj
from segmentator.config_gui import palette, axcolor, hovcolor
#
"""Data Processing"""
nii = load(cfg.filename)
orig, dims = check_data(nii.get_data(), cfg.force_original_precision)
# Save min and max truncation thresholds to be used in axis labels
if np.isnan(cfg.valmin) or np.isnan(cfg.valmax):
orig, pMin, pMax = truncate_range(orig,
percMin=cfg.perc_min,
percMax=cfg.perc_max)
else: # TODO: integrate this into truncate range function
orig[orig < cfg.valmin] = cfg.valmin
orig[orig > cfg.valmax] = cfg.valmax
pMin, pMax = cfg.valmin, cfg.valmax
# Continue with scaling the original truncated image and recomputing gradient
orig = scale_range(orig, scale_factor=cfg.scale, delta=0.0001)
gra = set_gradient_magnitude(orig, cfg.gramag)
if cfg.export_gramag:
export_gradient_magnitude_image(gra, nii.get_filename(), cfg.gramag,
nii.affine)
# Reshape for voxel-wise operations
ima = np.copy(orig.flatten())
gra = gra.flatten()