"""
R2
"""
"""
PCA
"""
"""
R1
"""
data_t1, y_t1, subs_t1 = load_cort(measure = 'r1', cort='midgray')
dataset = 'stanford_ms_run1'
cortical_mat_ms1_t1, age_ms1, subs, _ = load_dataset(dataset, cortical_parc='midgray',
measure_type='r1_les')
dataset = 'stanford_ms_run2'
cortical_mat_ms2_t1, age_ms2, subs, _ = load_dataset(dataset, cortical_parc='midgray',
measure_type='r1_les')
NUM_REPS = 1000
X = data_t1.T
youngsters = np.where(y_t1<=AGE)
y_old = np.delete(y_t1, youngsters)
data_old = np.delete(X, youngsters, axis=0)
data_old = np.delete(data_old, np.where(areas_notdeg2==1), axis=1)
#df_pca, X_pca, pca_evr, df_corr, df_n, df_na, df_nabv, df_rank = pca_full_report(X=data_old, features_=np.delete(areas, np.where(areas_notdeg2==1)), save_plot=False, fig_dpi=50)
adults = np.where(y_t1>AGE)
y_young = np.delete(y_t1, adults)
from src.data.preprocess import run_pca
from src.data.make_dataset import load_dataset, load_subjects
from sklearn import svm
from sklearn.metrics import confusion_matrix, accuracy_score
from sklearn.ensemble import RandomForestClassifier
import numpy as np
import pickle
def transform_data(data, channels):
"""Select a number of channels from each data record and concatenate them into a single vector"""
return [np.ravel(d[0][:channels]) for d in data['data']]
dataset = load_dataset('data/processed/hcp', 'data/hcp-train.csv')
testdataset = load_dataset('data/processed/hcp', 'data/hcp-eval-dist.csv')
channels = 10
dd = transform_data(dataset, channels)
td = transform_data(testdataset, channels)
model = svm.SVC(gamma=0.01, C=10.)
model.fit(dd, dataset['age'])
with open('models/svm-age.dat', 'wb') as out:
pickle.dump(model, out)
predicted = model.predict(td)
print("id,age")
areas = load_cort_areas()
from scipy.stats import pearsonr
#"""
#CT
#"""
#dataset = 'huji'
#cortical_parc = 'volume' ## 'volume'
#data_matrix_imputed, age_list, subs, area_names = load_dataset(dataset, cortical_parc=cortical_parc, measure_type='t1')
"""
T1
"""
dataset = 'huji'
cortical_parc = 'midgray' ## 'volume'
data_t1, y_t1, subs_t1, areas = load_dataset(dataset, cortical_parc=cortical_parc, measure_type='r1')
subjects_to_delete = [1, 3, 4, 5, 17, 24]
data_t1 = np.delete(data_t1, subjects_to_delete, axis=1)
y_t1 = np.delete(y_t1, subjects_to_delete)
#subs_t1 = np.delete(subs_t1, subjects_to_delete)
#sex= [0, 1, 1, 1, 1, 1,
# 1, 0, 1, 1, 1, 1,
# 0, 1, 1, 0, 1, 1,
# 1, 1, 1, 1, 0, 0,
# 0, 0, 1, 0, 1, 1,
# 0, 1, 1, 0] # 1-male. 23 Male. Young:10 Male
dataset = 'huji'
cortical_parc = 'volume'
file_data = nib.load(file).get_data()
except:
print('File {} does not exist'.format(file))
x, y, z = file_data.shape
file_axial = file_data[:, z // 2, :]
return file_axial
#dataset = 'stanford_2'
#data_stan, age_stan, subs_stan, _ = load_dataset(dataset)
#dataset = 'kalanit'
#data_kalanit, age_kalanit, subs_kalanit, _ = load_dataset(dataset)
dataset = 'gotlib'
data_matrix_gotlib, age_gotlib, subs_gotlib, _ = load_dataset(dataset)
figures = []
for sub in subs_gotlib:
sub_fs = sub + '_' + dataset
aparc_file = '/ems/elsc-labs/mezer-a/Mezer-Lab/analysis/freesurfer_subjects/' + sub_fs + '/mri/aparc.a2009s+aseg.mgz'
orig_file = '/ems/elsc-labs/mezer-a/Mezer-Lab/analysis/freesurfer_subjects/' + sub_fs + '/mri/orig.mgz'
if os.path.exists(aparc_file) and os.path.exists(orig_file):
axial_slice = get_axial(aparc_file)
axial_slice2 = get_axial(orig_file)
# make figure
plt.imshow(axial_slice2, cmap='gray')
plt.imshow(axial_slice, cmap='gray', alpha=0.2)
ax.set_rasterized(True)
ax = plt.title('Subject {}'.format(sub))
areas_notdeg2[areas_notdeg2_idx] = 1
"""
PCA
"""
"""
R1
"""
data_t1, y_t1, subs_t1 = load_cort(measure = 'r1', cort='midgray')
dataset = 'reading'
data_matrix_reading, age_reading, subs_reading, _ = load_dataset(dataset,
cortical_parc='midgray',
measure_type = 'r1')
no_preterm =['s016', 's088', 's114', 's027', 's041',
's121', 's132', 's143', 's057',
's061', 's079', 's070', 's067',
's080', 's148', 's112', 's117', 's160', 's113', ]
#
# # ringing
# 's007','s034','s037','s045','s046','s064','s078','s082',
# 's083','s083_2','s092_dti30','s097_2','s101','s109','s111','s116',
# 's120','s123','s126','s126_2','s127','s129','s133','s137',
# 's141','s142','s144','s145','s146','s147','s153','s154',
# 's157','s168','s169','s171','s175',
# 's006','s038','AOK07_run1','s008_2','s040','s055',
# 's062','s081','s086','s096','s102_2','s125','s134','s138']
no_preterm_idx = [index for index, elem in enumerate(subs_reading) if elem in no_preterm]
def prepare_experiment_data(download_new=False, test_seed:int=123, valid_seed:int=456) -> ExperimentData:
raw_data = load_dataset(download_new)
raw_data = get_dummy_values(fill_missing_values(raw_data))
data_min_n_ratings = get_users_with_min_n_ratings(raw_data, 3)
experiment_data = train_validation_test_split(data_min_n_ratings, test_seed=test_seed, valid_seed=valid_seed)
return experiment_data