ENH test folder structure refactoring and further independent testing in progress

Author: SebastianoF

README.md

@@ -73,7 +73,6 @@ in a [virtualenv](http://docs.python-guide.org/en/latest/dev/virtualenvs/).
     + Install [ITK-snap](http://www.itksnap.org/pmwiki/pmwiki.php?n=Downloads.SNAP3)
 
 
-
 ### Documentations
 
 + [Wiki-pages documentation](https://github.com/SebastianoF/nilabels/wiki)

tests/agents/test_agents_controller.py

@@ -0,0 +1 @@
+# TODO

tests/test_auxiliary_methods.py renamed to tests/tools/test_auxiliary_methods.py

@@ -84,7 +84,7 @@ def test_set_new_data_simple_modifications():
     # see if the infos are the same as in the modified header
     assert_array_equal(im_1.get_data()[:], np.ones([3, 3, 3]))
     assert_equals(im_1_header['intent_code'], 5)
-    assert_array_equal(im_1.get_affine(), aff)
+    assert_array_equal(im_1.affine, aff)
 
 
 def test_compare_two_nib_equals():
@@ -181,7 +181,7 @@ def test_from_permutation_to_disjoints_cycles_single_cycle():
     cauchy_perm = [[1, 2, 3, 4, 5, 6, 7],
                    [3, 4, 5, 1, 2, 7, 6]]
     cycles_perm = permutation_from_cauchy_to_disjoints_cycles(cauchy_perm)
-    expected_ans = [[1, 3, 5, 2, 4], [6,7]]
+    expected_ans = [[1, 3, 5, 2, 4], [6, 7]]
 
     print expected_ans
     print cycles_perm
@@ -202,11 +202,28 @@ def test_from_disjoint_cycles_to_permutation_single_cycle():
         assert_array_equal(c1, c2)
 
 
-test_from_permutation_to_disjoints_cycles_single_cycle()
-test_from_disjoint_cycles_to_permutation_single_cycle()
+if __name__ == '__main__':
+    test_get_morpological_patch()
+    test_get_patch_values_simple()
+    test_get_shell_for_given_radius()
+    test_check_pfi_io()
 
-test_from_disjoint_cycles_to_permutation()
-test_from_permutation_to_disjoints_cycles()
+    test_set_new_data_simple_modifications()
+    test_compare_two_nib_equals()
+    test_compare_two_nib_different_nifti_version()
+    test_compare_two_nib_different_affine()
 
-#
-# permutation_from_cauchy_to_disjoints_cycles,
+    test_eliminates_consecutive_duplicates()
+    test_lift_list_1()
+    test_lift_list_2()
+    test_lift_list_3()
+
+    test_labels_query_int_input()
+    test_labels_query_list_input1()
+    test_labels_query_list_input2()
+    test_labels_query_all_or_tot_input()
+
+    test_from_permutation_to_disjoints_cycles()
+    test_from_disjoint_cycles_to_permutation()
+    test_from_permutation_to_disjoints_cycles_single_cycle()
+    test_from_disjoint_cycles_to_permutation_single_cycle()

tests/test_caliber_distances.py renamed to tests/tools/test_caliber_distances.py

@@ -4,9 +4,8 @@
 from scipy import ndimage as nd
 
 from numpy.testing import assert_array_equal, assert_equal, assert_almost_equal
-from nilabels.tools.aux_methods.utils_nib import set_new_data
-from nilabels.tools.detections.contours import contour_from_segmentation
 
+from nilabels.tools.detections.contours import contour_from_segmentation
 
 from nilabels.tools.caliber.distances import centroid_array, centroid, dice_score, global_dice_score, \
     global_outline_error, covariance_matrices, covariance_distance, hausdorff_distance, \
@@ -17,7 +16,7 @@
 # --- Auxiliaries
 
 
-def test_centroid_array():
+def test_centroid_array_1():
     test_arr = np.array([[0, 0, 0, 0, 0, 0, 0, 0, 0],
                          [0, 0, 0, 0, 0, 0, 0, 0, 0],
                          [0, 1, 1, 1, 0, 0, 0, 0, 0],
@@ -716,96 +715,9 @@ def test_hausdorff_distance():
     assert_almost_equal(hd_1_void['label2'], np.nan)
 
 
-
-# test symmetric_contour_distance
-
-
-def test_normalised_symetric_contour_distance(save_data_path='/Users/aaabbbccc/Desktop', verbose=False):
-    o1 = o_shape(omega=(19, 19, 19), radius=7, background_intensity=0, foreground_intensity=1, dtype=np.uint8)
-    o2 = 2 * o1
-    arr1 = np.concatenate([o1, o2], axis=2)
-
-    half_o1 = np.zeros_like(o1)
-    half_o1[:, :10, :] = o1[:, :10, :]
-    half_o2 = 2 * half_o1
-    arr2 = np.concatenate([half_o1, half_o2], axis=2)
-
-    c1 = cube_shape(omega=(19, 19, 19), center=(9, 9, 9), side_length=13, background_intensity=0, foreground_intensity=1, dtype=np.uint8)
-    c2 = 2 * c1
-    arr3 = np.concatenate([c1, c2], axis=2)
-
-    arr_void = np.zeros_like(arr1)
-
-    im1 = nib.Nifti1Image(arr1, np.eye(4))
-    im2 = nib.Nifti1Image(arr2, np.eye(4))
-    im3 = nib.Nifti1Image(arr3, np.eye(4))
-
-    im_void = nib.Nifti1Image(arr_void, np.eye(4))
-
-    if os.path.exists(save_data_path):
-
-        nib.save(im1, os.path.join(save_data_path, 'zzz1.nii.gz'))
-        nib.save(im2, os.path.join(save_data_path, 'zzz2.nii.gz'))
-        nib.save(im3, os.path.join(save_data_path, 'zzz3.nii.gz'))
-
-        im1_cont = contour_from_segmentation(im1)
-        im2_cont = contour_from_segmentation(im2)
-        im3_cont = contour_from_segmentation(im3)
-
-        nib.save(im1_cont, os.path.join(save_data_path, 'zzz1_contour.nii.gz'))
-        nib.save(im2_cont, os.path.join(save_data_path, 'zzz2_contour.nii.gz'))
-        nib.save(im3_cont, os.path.join(save_data_path, 'zzz3_contour.nii.gz'))
-
-        im_dtb1_l1 = set_new_data(im1_cont, nd.distance_transform_edt(1 - (im1_cont.get_data() == 1)))
-        im_dtb2_l1 = set_new_data(im2_cont, nd.distance_transform_edt(1 - (im2_cont.get_data() == 1)))
-        im_dtb3_l1 = set_new_data(im3_cont, nd.distance_transform_edt(1 - (im3_cont.get_data() == 1)))
-
-        im_dtb1_l2 = set_new_data(im1_cont, nd.distance_transform_edt(1 - (im1_cont.get_data() == 2)))
-        im_dtb2_l2 = set_new_data(im2_cont, nd.distance_transform_edt(1 - (im2_cont.get_data() == 2)))
-        im_dtb3_l2 = set_new_data(im3_cont, nd.distance_transform_edt(1 - (im3_cont.get_data() == 2)))
-
-        nib.save(im_dtb1_l1, os.path.join(save_data_path, 'zzz1_contour_dist_label1.nii.gz'))
-        nib.save(im_dtb2_l1, os.path.join(save_data_path, 'zzz2_contour_dist_label1.nii.gz'))
-        nib.save(im_dtb3_l1, os.path.join(save_data_path, 'zzz3_contour_dist_label1.nii.gz'))
-
-        nib.save(im_dtb1_l2, os.path.join(save_data_path, 'zzz1_contour_dist_label2.nii.gz'))
-        nib.save(im_dtb2_l2, os.path.join(save_data_path, 'zzz2_contour_dist_label2.nii.gz'))
-        nib.save(im_dtb3_l2, os.path.join(save_data_path, 'zzz3_contour_dist_label2.nii.gz'))
-
-        print('Images testing saved in {}'.format(save_data_path))
-
-    ascd_1_1       = normalised_symmetric_contour_distance(im1, im1, [1, 2], ['label1', 'label2'])
-    ascd_1_2       = normalised_symmetric_contour_distance(im1, im2, [1, 2], ['label1', 'label2'])
-    ascd_1_3       = normalised_symmetric_contour_distance(im1, im3, [1, 2], ['label1', 'label2'])
-    ascd_1_2_extra = normalised_symmetric_contour_distance(im1, im2, [1, 2, 3], ['label1', 'label2', 'label3'])
-    ascd_1_void    = normalised_symmetric_contour_distance(im1, im_void, [1, 2], ['label1', 'label2'])
-
-    if verbose:
-        print('\n 1 1')
-        print(ascd_1_1)
-        print('\n 1 2')
-        print(ascd_1_2)
-        print('\n 1 3')
-        print(ascd_1_3)
-        print('\n 1 2 extra')
-        print(ascd_1_2_extra)
-        print('\n 1 void')
-        print(ascd_1_void)
-
-    assert_almost_equal(ascd_1_1['label1'], 0)
-    assert_almost_equal(ascd_1_1['label2'], 0)
-
-    assert_almost_equal(ascd_1_2['label1'], 1.01226261459, decimal=4)
-    assert_almost_equal(ascd_1_2['label2'], 1.01226261459, decimal=4)
-
-    assert_almost_equal(ascd_1_3['label1'], 0.866793279546 , decimal=4)
-    assert_almost_equal(ascd_1_3['label2'], 0.866793279546, decimal=4)
-
-    assert_almost_equal(ascd_1_2_extra['label3'], np.nan)
-
-    assert_almost_equal(ascd_1_void['label1'], np.nan)
-    assert_almost_equal(ascd_1_void['label2'], np.nan)
-
+def test_normalised_symmetric_contour_distance():
+    # TODO
+    pass
 
 # --- extra:
 
@@ -840,3 +752,25 @@ def test_box_side_lenght():
     assert_array_equal(se_answer['2'], [1., 2., 4.])
     assert_equal(se_answer['3'], np.nan)
 
+
+if __name__ == '__main__':
+    test_centroid_array_1()
+    test_centroid_array_2()
+    test_centroid()
+    test_covariance_matrices()
+    test_covariance_distance_between_matrices_simple_case()
+
+    test_dice_score()
+    test_global_dice_score()
+    test_global_outline_error()
+
+    test_dice_score_one_label()
+    test_asymmetric_component_Hausdorff_distance_H_d_and_Hausdorff_distance()
+    test_dice_score_multiple_labels()
+    test_covariance_distance()
+    test_covariance_distance_range()
+    test_hausdorff_distance()
+
+    test_normalised_symmetric_contour_distance()
+
+    test_box_side_lenght()

tests/tools/test_caliber_volumes_and_values.py

@@ -0,0 +1,162 @@
+import os
+from os.path import join as jph
+
+import numpy as np
+import nibabel as nib
+from nose.tools import assert_raises, assert_almost_equal, assert_equal, assert_equals
+from numpy.testing import assert_array_equal
+
+from nilabels.tools.caliber.volumes_and_values import get_total_num_nonzero_voxels, get_num_voxels_from_labels_list, \
+    get_values_below_labels_list
+
+
+def cube_shape(omega, center, side_length, background_intensity=0, foreground_intensity=100, dtype=np.uint8):
+    sky = background_intensity * np.ones(omega, dtype=dtype)
+    half_side_length = int(np.ceil(int(side_length / 2)))
+
+    for lx in range(-half_side_length, half_side_length + 1):
+        for ly in range(-half_side_length, half_side_length + 1):
+            for lz in range(-half_side_length, half_side_length + 1):
+                sky[center[0] + lx, center[1] + ly, center[2] + lz] = foreground_intensity
+    return sky
+
+
+def test_volumes_and_values_total_num_voxels():
+
+    omega = [80, 80, 80]
+    cube_a = [[10, 60, 55], 11, 1]
+    cube_b = [[50, 55, 42], 17, 2]
+    cube_c = [[25, 20, 20], 19, 3]
+    cube_d = [[55, 16, 9], 9, 4]
+
+    sky = cube_shape(omega, center=cube_a[0], side_length=cube_a[1], foreground_intensity=cube_a[2]) + \
+           cube_shape(omega, center=cube_b[0], side_length=cube_b[1], foreground_intensity=cube_b[2]) + \
+           cube_shape(omega, center=cube_c[0], side_length=cube_c[1], foreground_intensity=cube_c[2]) + \
+           cube_shape(omega, center=cube_d[0], side_length=cube_d[1], foreground_intensity=cube_d[2])
+    im_segm = nib.Nifti1Image(sky, affine=np.eye(4))
+
+    num_voxels = get_total_num_nonzero_voxels(im_segm)
+    assert_equal(num_voxels, 11 ** 3 + 17 ** 3 + 19 ** 3 + 9 **3)
+
+    num_voxels = get_total_num_nonzero_voxels(im_segm, list_labels_to_exclude=[2, 4])
+    assert_equal(num_voxels, 11 ** 3 + 19 ** 3)
+
+
+def test_volumes_and_values_total_num_voxels_empthy():
+
+    omega = [80, 80, 80]
+    im_segm = nib.Nifti1Image(np.zeros(omega), affine=np.eye(4))
+
+    num_voxels = get_total_num_nonzero_voxels(im_segm)
+    print(num_voxels)
+    assert_equal(num_voxels, 0)
+
+
+def test_volumes_and_values_total_num_voxels_full():
+
+    omega = [80, 80, 80]
+    im_segm = nib.Nifti1Image(np.ones(omega), affine=np.eye(4))
+
+    num_voxels = get_total_num_nonzero_voxels(im_segm)
+    print(num_voxels)
+    assert_equal(num_voxels, 80 ** 3)
+
+
+def test_get_num_voxels_from_labels_list():
+
+    omega = [80, 80, 80]
+    cube_a = [[10, 60, 55], 11, 1]
+    cube_b = [[50, 55, 42], 15, 2]
+    cube_c = [[25, 20, 20], 13, 3]
+    cube_d = [[55, 16, 9], 7, 4]
+
+    sky = cube_shape(omega, center=cube_a[0], side_length=cube_a[1], foreground_intensity=cube_a[2]) + \
+          cube_shape(omega, center=cube_b[0], side_length=cube_b[1], foreground_intensity=cube_b[2]) + \
+          cube_shape(omega, center=cube_c[0], side_length=cube_c[1], foreground_intensity=cube_c[2]) + \
+          cube_shape(omega, center=cube_d[0], side_length=cube_d[1], foreground_intensity=cube_d[2])
+    im_segm = nib.Nifti1Image(sky, affine=np.eye(4))
+
+    num_voxels = get_num_voxels_from_labels_list(im_segm, labels_list=[1, 2, 3, 4])
+    print(num_voxels, [11 **3, 15 **3, 13 **3, 7 ** 3])
+    assert_array_equal(num_voxels, [11 **3, 15 **3, 13 **3, 7 ** 3])
+
+    num_voxels = get_num_voxels_from_labels_list(im_segm, labels_list=[1, [2, 3], 4])
+    print(num_voxels, [11 ** 3, 15 ** 3 + 13 ** 3, 7 ** 3])
+    assert_array_equal(num_voxels, [11 ** 3, 15 ** 3 + 13 ** 3, 7 ** 3])
+
+    num_voxels = get_num_voxels_from_labels_list(im_segm, labels_list=[[1, 2, 3], 4])
+    print(num_voxels, [11 ** 3, 15 ** 3 + 13 ** 3, 7 ** 3])
+    assert_array_equal(num_voxels, [11 ** 3 + 15 ** 3 + 13 ** 3, 7 ** 3])
+
+    num_voxels = get_num_voxels_from_labels_list(im_segm, labels_list=[[1, 2, 3, 4]])
+    print(num_voxels, [11 ** 3, 15 ** 3 + 13 ** 3, 7 ** 3])
+    assert_array_equal(num_voxels, [11 ** 3 + 15 ** 3 + 13 ** 3 + 7 ** 3])
+
+
+def test_get_num_voxels_from_labels_list_unexisting_labels():
+
+    omega = [80, 80, 80]
+    cube_a = [[10, 60, 55], 11, 1]
+    cube_b = [[50, 55, 42], 15, 2]
+    cube_c = [[25, 20, 20], 13, 3]
+    cube_d = [[55, 16, 9], 7, 4]
+
+    sky = cube_shape(omega, center=cube_a[0], side_length=cube_a[1], foreground_intensity=cube_a[2]) + \
+          cube_shape(omega, center=cube_b[0], side_length=cube_b[1], foreground_intensity=cube_b[2]) + \
+          cube_shape(omega, center=cube_c[0], side_length=cube_c[1], foreground_intensity=cube_c[2]) + \
+          cube_shape(omega, center=cube_d[0], side_length=cube_d[1], foreground_intensity=cube_d[2])
+    im_segm = nib.Nifti1Image(sky, affine=np.eye(4))
+
+    num_voxels = get_num_voxels_from_labels_list(im_segm, labels_list=[1, 2, 3, 5])
+    print(num_voxels, [11 ** 3, 15 ** 3, 13 ** 3, 0])
+    assert_array_equal(num_voxels, [11 ** 3, 15 ** 3, 13 ** 3, 0])
+
+    num_voxels = get_num_voxels_from_labels_list(im_segm, labels_list=[1, 2, [3, 5]])
+    print(num_voxels, [11 ** 3, 15 ** 3, 13 ** 3 + 0])
+    assert_array_equal(num_voxels, [11 ** 3, 15 ** 3, 13 ** 3 + 0])
+
+    num_voxels = get_num_voxels_from_labels_list(im_segm, labels_list=[[1, 2], [7, 8]])
+    print(num_voxels, [11 ** 3 + 15 ** 3,  0])
+    assert_array_equal(num_voxels, [11 ** 3 + 15 ** 3, 0])
+
+    num_voxels = get_num_voxels_from_labels_list(im_segm, labels_list=[[1, 2], [7, -8]])
+    print(num_voxels, [11 ** 3 + 15 ** 3, 0])
+    assert_array_equal(num_voxels, [11 ** 3 + 15 ** 3, 0])
+
+
+def test_get_values_below_labels_list():
+    omega = [80, 80, 80]
+
+    cube_a_seg = [[10, 60, 55], 11, 1]
+    cube_b_seg = [[50, 55, 42], 15, 2]
+    cube_c_seg = [[25, 20, 20], 13, 3]
+    cube_d_seg = [[55, 16, 9], 7, 4]
+
+    cube_a_anat = [[10, 60, 55], 11, 1.5]
+    cube_b_anat = [[50, 55, 42], 15, 2.5]
+    cube_c_anat = [[25, 20, 20], 13, 3.5]
+    cube_d_anat = [[55, 16, 9], 7, 4.5]
+
+    sky_s = cube_shape(omega, center=cube_a_seg[0], side_length=cube_a_seg[1], foreground_intensity=cube_a_seg[2])
+    sky_s += cube_shape(omega, center=cube_b_seg[0], side_length=cube_b_seg[1], foreground_intensity=cube_b_seg[2])
+    sky_s += cube_shape(omega, center=cube_c_seg[0], side_length=cube_c_seg[1], foreground_intensity=cube_c_seg[2])
+    sky_s += cube_shape(omega, center=cube_d_seg[0], side_length=cube_d_seg[1], foreground_intensity=cube_d_seg[2])
+
+    sky_a = cube_shape(omega, center=cube_a_anat[0], side_length=cube_a_anat[1], foreground_intensity=cube_a_anat[2], dtype=np.float32)
+    sky_a += cube_shape(omega, center=cube_b_anat[0], side_length=cube_b_anat[1], foreground_intensity=cube_b_anat[2], dtype=np.float32)
+    sky_a += cube_shape(omega, center=cube_c_anat[0], side_length=cube_c_anat[1], foreground_intensity=cube_c_anat[2], dtype=np.float32)
+    sky_a += cube_shape(omega, center=cube_d_anat[0], side_length=cube_d_anat[1], foreground_intensity=cube_d_anat[2], dtype=np.float32)
+
+    im_segm = nib.Nifti1Image(sky_s, affine=np.eye(4))
+    im_anat = nib.Nifti1Image(sky_a, affine=np.eye(4))
+
+    assert im_segm.shape == im_anat.shape
+
+    labels_list = [[1, 2], [3, 4], 4]
+
+    vals_below = get_values_below_labels_list(im_segm, im_anat, labels_list)
+
+    assert_array_equal(vals_below[0], np.array([1.5, ] * (11**3) + [2.5] * (15**3)) )
+    assert_array_equal(vals_below[1], np.array([3.5, ] * (13**3) + [4.5] * (7**3)) )
+    assert_array_equal(vals_below[2], np.array([4.5] * (7 ** 3)))
+