-
Notifications
You must be signed in to change notification settings - Fork 2
/
plot_manifold_adni.py
127 lines (94 loc) · 5.07 KB
/
plot_manifold_adni.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
import pandas as pd
import seaborn as sns
import numpy as np
import os
import matplotlib.pyplot as plt
def load_yaml_as_dict(yaml_path):
import yaml
with open(yaml_path, "r") as stream:
config_dict = yaml.safe_load(stream)
return config_dict
def get_ids(filename_split, split='train'):
split = load_yaml_as_dict(filename_split)[split]
return split
def read_data(split, filename_datasource, attributes):
df_data = pd.read_csv(filename_datasource, header=0)
df_data_split = df_data.loc[df_data['id'].astype('str').isin(split)]
# read covariates
list_attributes = []
for ith_attribute in attributes:
arr_current_attribute = np.array(df_data_split[ith_attribute])
list_attributes.append(arr_current_attribute)
features = np.array(list_attributes).T
return features
def get_airway_data_for_id(test_idx, filename_datasource, train_split, attributes_names):
# train_split = get_ids(filename_split, split='train')
train_covariates = read_data(train_split, filename_datasource, attributes_names)
case_covariates = read_data(test_idx, filename_datasource, attributes_names)
attributes = {}
ori_attributes = {}
for ith_attri in range(len(attributes_names)):
ori_attributes[attributes_names[ith_attri]] = case_covariates[:, ith_attri]
attributes[attributes_names[ith_attri]] = (case_covariates[:, ith_attri] - train_covariates[:,
ith_attri].min()) / (
train_covariates[:, ith_attri].max() - train_covariates[:,
ith_attri].min())
attributes[attributes_names[ith_attri]] = attributes[attributes_names[ith_attri]] * 2 - 1
attributes[attributes_names[ith_attri]] = attributes[attributes_names[ith_attri]][None, :]
return attributes
def get_adni_data_for_id(test_idx, df_data, train_split, attributes_names):
train_covariates = read_data(train_split, df_data, attributes_names)
case_covariates = read_data(test_idx, df_data, attributes_names)
attributes = {}
ori_attributes = {}
mean_attributes = {}
std_attributes = {}
for ith_attri in range(len(attributes_names)):
ori_attributes[attributes_names[ith_attri]] = case_covariates[:, ith_attri]
attributes[attributes_names[ith_attri]] = (case_covariates[:, ith_attri] - train_covariates[:,ith_attri].mean()) / train_covariates[:, ith_attri].std()
attributes[attributes_names[ith_attri]] = attributes[attributes_names[ith_attri]][None, :]
mean_attributes[attributes_names[ith_attri]] = train_covariates[:,ith_attri].mean()
std_attributes[attributes_names[ith_attri]] = train_covariates[:,ith_attri].std()
return attributes, mean_attributes, std_attributes
def get_data_for_id(shapetype):
if shapetype == 'ADNI':
return get_adni_data_for_id
elif shapetype == 'Airway':
return get_airway_data_for_id
if __name__ == "__main__":
path_dataset = 'examples/hippocampus/3dshape.csv'
filename_split = 'examples/hippocampus/newsplit.yaml'
pd_cov = pd.read_csv(path_dataset)
all_ids = get_ids(filename_split, split='train') + get_ids(filename_split, split='test')
train_ids = get_ids(filename_split, split='train')
attributes, mean_attributes, std_attributes = get_data_for_id("ADNI")(all_ids, path_dataset, train_ids, ['age', 'AD'])
pd_cov = pd_cov.loc[pd_cov['id'].astype('str').isin(all_ids)]
pd_cov_train = pd_cov.loc[pd_cov['id'].astype('str').isin(train_ids)]
train_ages = pd_cov_train['age'].values
train_weight = pd_cov_train['AD'].values
pd_cov['age'] = attributes['age'][0]
pd_cov['AD'] = attributes['AD'][0]
scatters = sns.jointplot(data=pd_cov, x="age", y="AD", hue="sex", palette={
1: "#00FFFF",
0: "#FF00FF"
}, alpha=0.35)
palette = {
1: "#00FFFF",
0: "#FF00FF"
}
# for i,gr in pd_cov.groupby('sex'):
# sns.regplot(x="age", y="weight", data=gr, scatter=False, ax=scatters.ax_joint, truncate=False,
# scatter_kws={"color": palette[i]}, line_kws={"color": palette[i]})
scatters.ax_marg_x.set_xlim(-3.2, 3.2)
scatters.ax_marg_y.set_ylim(-1., 2.)
# set the labels
scatters.ax_marg_x.set_xticks(np.linspace(-3.0, 3.0, 7))# * std_attributes['age'] + mean_attributes['age'])
scatters.ax_marg_x.set_xticklabels(np.round(np.linspace(-3., 3., 7) * std_attributes['age'] + mean_attributes['age'], 1))
scatters.ax_marg_y.set_yticks(np.linspace(-1., 2., 7))# * std_attributes['weight'] + mean_attributes['weight'])
scatters.ax_marg_y.set_yticklabels(np.round(np.linspace(-1., 2., 7) * std_attributes['AD'] + mean_attributes['AD'], 1))
scatters.set_axis_labels('age / yrs', 'AD', fontsize=16)
# boxes.set(title=title)
# fig = scatters.get_figure()
# sns.move_legend(scatters, "lower right")
scatters.ax_joint.legend_.remove()
scatters.savefig('./data_generation/b.svg', transparent=True)