def distance_matrices_ext(samples, metric, metric_params=None, ordn=2):
pdist = PairwiseDistance(metric=metric,
metric_params=metric_params,
order=None,
n_jobs=-1)
D = pdist.fit_transform(samples)
scale = (D[:, :, 0].max() - D[:, :, 0].min()) / (D[:, :, 1].max() -
D[:, :, 1].min())
Dscaled = np.stack((D[:, :, 0], D[:, :, 1] * scale), axis=2)
Dn = np.linalg.norm(D, ord=ordn, axis=2)
Dscaledn = np.linalg.norm(Dscaled, ord=ordn, axis=2)
plt.figure(figsize=(8, 8))
#plot H0 and H1 distances separated
plt.subplot(221)
plt.imshow(D[:, :, 0])
plt.axis('off')
plt.colorbar(shrink=0.8)
plt.title('H0')
plt.subplot(222)
plt.imshow(D[:, :, 1])
plt.axis('off')
plt.colorbar(shrink=0.8)
plt.title('H1')
#plot complessive distance (whith or whithout scaling)
plt.subplot(223)
plt.imshow(Dn)
plt.axis('off')
plt.colorbar(shrink=0.8)
plt.title('Not scaled, norm=' + str(ordn) + '-norm')
plt.subplot(224)
plt.imshow(Dscaledn)
plt.axis('off')
plt.colorbar(shrink=0.8)
plt.title('Scaled, norm=' + str(ordn) + '-norm')
plt.suptitle(metric)
def test_not_fitted():
dd = PairwiseDistance()
da = Amplitude()
with pytest.raises(NotFittedError):
dd.transform(X_1)
with pytest.raises(NotFittedError):
da.transform(X_1)
def test_dd_transform(metric, metric_params, order, n_jobs):
# X_fit == X_transform
dd = PairwiseDistance(metric=metric, metric_params=metric_params,
order=order, n_jobs=n_jobs)
X_res = dd.fit_transform(X1)
assert (X_res.shape[0], X_res.shape[1]) == (X1.shape[0], X1.shape[0])
if order is None:
assert X_res.shape[2] == n_homology_dimensions
# X_fit != X_transform
dd = PairwiseDistance(metric=metric, metric_params=metric_params,
order=order, n_jobs=n_jobs)
X_res = dd.fit(X1).transform(X2)
assert (X_res.shape[0], X_res.shape[1]) == (X2.shape[0], X1.shape[0])
if order is None:
assert X_res.shape[2] == n_homology_dimensions
# X_fit != X_transform, default metric_params
dd = PairwiseDistance(metric=metric, order=order, n_jobs=n_jobs)
X_res = dd.fit(X1).transform(X2)
assert (X_res.shape[0], X_res.shape[1]) == (X2.shape[0], X1.shape[0])
if order is None:
assert X_res.shape[2] == n_homology_dimensions
def compute_distance_matrix(
diagrams,
metric,
metric_params,
plot_distance_matrix=False,
title=None,
file_prefix=None,
):
PD = PairwiseDistance(metric=metric,
metric_params=metric_params,
order=None,
n_jobs=N_JOBS)
X_distance = PD.fit_transform(diagrams)
return X_distance
def test_dd_transform(metric, metric_params, order, n_jobs):
# X_fit == X_transform
dd = PairwiseDistance(metric=metric,
metric_params=metric_params,
order=order,
n_jobs=n_jobs)
X_res = dd.fit_transform(X_1)
assert (X_res.shape[0], X_res.shape[1]) == (X_1.shape[0], X_1.shape[0])
# X_fit != X_transform
dd = PairwiseDistance(metric=metric,
metric_params=metric_params,
order=order,
n_jobs=n_jobs)
X_res = dd.fit(X_1).transform(X_2)
assert (X_res.shape[0], X_res.shape[1]) == (X_1.shape[0], X_2.shape[0])
if order is None:
assert X_res.shape[2] == len(np.unique(X_2[:, :, 2]))
# X_fit != X_transform, default metric_params
dd = PairwiseDistance(metric=metric, order=order, n_jobs=n_jobs)
X_res = dd.fit(X_1).transform(X_2)
assert (X_res.shape[0], X_res.shape[1]) == (X_1.shape[0], X_2.shape[0])
def get_pairwise_distance_metrics():
metrics = {
"bottleneck":
PairwiseDistance(metric="bottleneck"),
"wasserstein":
PairwiseDistance(metric="wasserstein", metric_params={"p": 1}),
"betti":
PairwiseDistance(metric="betti"),
"landscape":
PairwiseDistance(metric="landscape"),
"silhouette":
PairwiseDistance(metric="silhouette"),
"heat":
PairwiseDistance(metric="heat"),
"persistence_image":
PairwiseDistance(metric="persistence_image"),
}
return metrics
def distance_matrices(samples, metric, metric_params=None, ordn=2):
pdist = PairwiseDistance(metric=metric,
metric_params=metric_params,
order=None,
n_jobs=2)
D = pdist.fit_transform(samples)
scale = (D[:, :, 0].max() - D[:, :, 0].min()) / (D[:, :, 1].max() -
D[:, :, 1].min())
Dscaled = np.stack((D[:, :, 0], D[:, :, 1] * scale), axis=2)
Dn = np.linalg.norm(D, ord=ordn, axis=2)
Dscaledn = np.linalg.norm(Dscaled, ord=ordn, axis=2)
plt.figure(figsize=(10, 5))
plt.subplot(121)
plt.imshow(Dn)
plt.axis('off')
plt.colorbar()
plt.title('Not scaled, norm=' + str(ordn) + '-norm')
plt.subplot(122)
plt.imshow(Dscaledn)
plt.axis('off')
plt.colorbar()
plt.title('Scaled, norm=' + str(ordn) + '-norm')
plt.suptitle(metric)
def main():
path_to_diags = "../data/collected_diagnoses_complete.json"
with open(path_to_diags) as f:
diagnoses = json.load(f)
patients = list(diagnoses.keys())
# Sort diagnoses key
diagnoses = collections.OrderedDict(sorted(diagnoses.items()))
# Where the data comes from
data_dir = DOTENV_KEY2VAL["DATA_DIR"] + "/patch_91/"
# Where the resulting distance matrices are saved.
distance_dir = "/temporal_evolution/"
utils.make_dir(DOTENV_KEY2VAL["GEN_DATA_DIR"] + distance_dir)
distances_to_evaluate = [
# "bottleneck",
# "wasserstein",
# "betti",
"landscape",
# "silhouette",
# "heat",
"persistence_image",
]
# patients = ["sub-ADNI011S0023", "sub-ADNI029S0878"]
# If we want to process multiple patients, we just throw them in a loop.
for i, patient in tqdm(enumerate(patients), total=len(patients)):
for distance in distances_to_evaluate:
patches = []
for mri in diagnoses[patient]:
try:
patches.append(
np.load(
data_dir
+ patient
+ mri.replace("ses", "")
+ "-MNI.npy"
)
)
except FileNotFoundError:
pass
# print(
# data_dir
# + patient
# + mri.replace("ses", "")
# + "-MNI.npy"
# + " not found"
# )
# Stacking enables multiprocessing
patches = np.stack(patches)
cp = CubicalPersistence(
homology_dimensions=HOMOLOGY_DIMENSIONS,
coeff=2,
periodic_dimensions=None,
infinity_values=None,
reduced_homology=True,
n_jobs=-1,
)
diagrams_cubical_persistence = cp.fit_transform(patches)
pl_dist = PairwiseDistance(
metric=distance,
metric_params=None,
order=None,
n_jobs=-1,
)
X_distance = pl_dist.fit_transform(
diagrams_cubical_persistence
)
with open(
DOTENV_KEY2VAL["GEN_DATA_DIR"]
+ distance_dir
+ f"patient_evolution_distance_data_patient_{patient}"
f"_{distance}.npy",
"wb",
) as f:
np.save(f, X_distance)
def compute_all_distances(X, metric):
if metric == "euclidean":
return squareform(pdist(X, metric))
else:
return PairwiseDistance(metric=metric,
n_jobs=self.n_jobs).fit_transform(X)
def compute_distance(emb2diagram, metric: str = "bottleneck"):
dis = PairwiseDistance(metric=metric)
dis.fit(emb2diagram["SBert"])
dis.transform(emb2diagram["LF"])
return dis
[0, 1, 0.],
[0, 0, 1.]], # Expected bottleneck ampl: [1, 0]
[[3, 3.5, 0.],
[0, 0, 0.],
[5, 9, 1.]] # Expected bottleneck ampl: [1/4, 2]
])
X_bottleneck_res_exp = np.array([
[1/2, 2],
[1, 0],
[1/4, 2]
])
@pytest.mark.parametrize('transformer', [PairwiseDistance(), Amplitude()])
def test_not_fitted(transformer):
with pytest.raises(NotFittedError):
transformer.transform(X1)
parameters_distance = [
('bottleneck', None),
('wasserstein', {'p': 2, 'delta': 0.1}),
('betti', {'p': 2.1, 'n_bins': 10}),
('landscape', {'p': 2.1, 'n_bins': 10, 'n_layers': 2}),
('silhouette', {'p': 2.1, 'power': 1.2, 'n_bins': 10}),
('heat', {'p': 2.1, 'sigma': 0.5, 'n_bins': 10}),
('persistence_image',
{'p': 2.1, 'sigma': 0.5, 'n_bins': 10}),
('persistence_image',
def main():
path_to_diags = "../data/collected_diagnoses_complete.json"
patients = ["sub-ADNI005S0223"]
progr = ["cn_mci_ad"]
with open(path_to_diags) as f:
diagnoses = json.load(f)
# Sort diagnoses key
diagnoses = collections.OrderedDict(sorted(diagnoses.items()))
# Where the data comes from
data_dir = DOTENV_KEY2VAL["DATA_DIR"] + "/patch_91/"
# Where the figures are saved
temporal_progression_dir = "/temporal_evolution/"
utils.make_dir(
DOTENV_KEY2VAL["GEN_FIGURES_DIR"] + temporal_progression_dir
)
# Where the resulting distance matrices are saved.
time_series_dir = "/temporal_evolution/"
utils.make_dir(DOTENV_KEY2VAL["GEN_DATA_DIR"] + temporal_progression_dir)
# If we want to process multiple patients, we just throw them in a loop.
for i, patient in enumerate(patients):
print(
"Processing longitudinal data for "
+ patient
+ " with progression pattern "
+ progr[i]
)
patches = []
for mri in diagnoses[patient]:
try:
patches.append(
np.load(
data_dir
+ patient
+ mri.replace("ses", "")
+ "-MNI.npy"
)
)
except FileNotFoundError:
print(
data_dir
+ patient
+ mri.replace("ses", "")
+ "-MNI.npy"
+ " not found"
)
# Stacking enables multiprocessing
patches = np.stack(patches)
cp = CubicalPersistence(
homology_dimensions=HOMOLOGY_DIMENSIONS,
coeff=2,
periodic_dimensions=None,
infinity_values=None,
reduced_homology=True,
n_jobs=-1,
)
diagrams_cubical_persistence = cp.fit_transform(patches)
pl_dist = PairwiseDistance(
metric="landscape", metric_params=None, order=None, n_jobs=-1
)
X_distance = pl_dist.fit_transform(
diagrams_cubical_persistence
)
with open(
DOTENV_KEY2VAL["GEN_DATA_DIR"]
+ time_series_dir
+ f"distance_data_patient_{patient}_{progr[i]}_landscape.npy",
"wb",
) as f:
np.save(f, X_distance)