def test_04_perfect_overlap_5mm_margin(self):
_, tumor_np = niftireader.load_image(
_get_file_name('T01', '04_perfect_overlap_5mm_margin', 'Tumor'))
_, ablation_np = niftireader.load_image(
_get_file_name('T01', '04_perfect_overlap_5mm_margin', 'Ablation'))
distances = margin.compute_distances(tumor_np, ablation_np, None, 1)
df = margin.summarize_surface_dists('T01', 'L01', distances)
print(df)
record = df.loc[(df['Patient'] == "T01")
& (df['Lesion'] == 'L01')].iloc[0]
self.assertEqual(record["nr_voxels"], 222.0)
self.assertAlmostEqual(record["min_distance"], 4.12, delta=0.01)
self.assertAlmostEqual(record["max_distance"], 5.0, delta=0.01)
def test_03_n5mm_margin_subcapsular(self):
_, tumor_np = niftireader.load_image(
_get_file_name('T02', '03_-5mm_margin_subcapsular', 'Tumor'))
_, ablation_np = niftireader.load_image(
_get_file_name('T02', '03_-5mm_margin_subcapsular', 'Ablation'))
_, liver_np = niftireader.load_image(
_get_file_name('T02', '03_-5mm_margin_subcapsular', 'Liver'))
distances = margin.compute_distances(tumor_np, ablation_np, liver_np,
1)
df = margin.summarize_surface_dists('T02', 'L03', distances)
print(df)
record = df.loc[(df['Patient'] == "T02")
& (df['Lesion'] == 'L03')].iloc[0]
self.assertEqual(record["nr_voxels"], 317.0)
self.assertAlmostEqual(record["min_distance"], -5.09, delta=0.01)
self.assertAlmostEqual(record["max_distance"], -2.23, delta=0.01)
if liver_file is not None:
# load the image file
liver, liver_np = load_image(liver_file)
# check whether there is a liver segmentation or not.
has_liver_segmented = np.sum(liver_np.astype(np.uint8)) > 0
else:
has_liver_segmented = False
# extract the spacing from the ablation file
pixdim = ablation.header['pixdim']
spacing = (pixdim[1], pixdim[2], pixdim[3])
# compute the surface distances based on tumor and ablation segmentations
surface_distance = compute_distances(
mask_gt=tumor_np,
mask_pred=ablation_np,
exclusion_zone=liver_np if has_liver_segmented else None,
spacing_mm=spacing,
connectivity=1,
crop=False)
# call the surface distance extraction function
if surface_distance['distances_gt_to_pred'].size > 0:
# if surface distances returned are not empty
non_ablated, insufficiently_ablated, completely_ablated = \
pm.plot_histogram_surface_distances(pat_name=patient_id, lesion_id=lesion_id,
output_file=output_file_histogram,
distance_map=surface_distance['distances_gt_to_pred'],
title='Quantitative Ablation Margin',
print_case_details=True)
df = pd.DataFrame(
data={
'Patient': [patient_id] *
