def generate_lists(data_params):
list_data = []
file_path = data_params['adni_data_des'] + tls.get_convention_name(
data_params) + '/List_data.pkl'
adni_out = generate_lists_from_adni_dataset(data_params)
list_data.append(adni_out)
daf.save_lists_to_file(path_file=file_path, data_list=list_data)
def save_data_params(data_params):
path_file = data_params['adni_data_des'] + tls.get_convention_name(data_params) + '/Data_params.pkl'
try:
os.makedirs(os.path.dirname(path_file))
except OSError as e:
if e.errno != errno.EEXIST:
raise
with open(path_file, 'wb') as f:
pickle.dump(data_params, f)
def generate_data_from_lists(data_params, selected_label=None):
file_path = data_params['adni_data_des'] + tls.get_convention_name(
data_params) + '/List_data.pkl'
data_list = daf.read_lists_from_file(file_path)
adni_1_in = data_list[0]
lists_with_names = zip(
[adni_1_in[0], adni_1_in[1], adni_1_in[2]],
['alz_ADNI_1_train', 'alz_ADNI_1_valid', 'alz_ADNI_1_test'])
time.sleep(1)
generate_data_from_selected_dataset(data_params, lists_with_names,
selected_label)
def main():
binaries_classes = ['AD-NC', 'AD-MCI', 'MCI-NC']
data_params = rsd.get_all_data_params()
root_path = data_params['adni_data_des']
name_cnv = root_path + tls.get_convention_name(data_params) + '/' + str(data_params['ROI_list'][data_params['ROI_selection']] + '/' + data_params['3D_or_2D'])
line = name_cnv + '/' + binaries_classes[0] + '/test/'
list_files = get_pickle_from_folder(line)
for i in list_files:
model = daf.read_model(i)
print(" HIPP_L : {} - HIPP_R: {} - Vector: {} - Label: {}".format(model.hippLeft.shape, model.hippRight.shape, model.hippMetaDataVector, model.hippLabel))
# print(model)
left_dims, right_dims = [[13,16],[13,16],[13,16]], [[13,16],[13,16],[13,16]]
plot_ROI_all(model.hippLeft, model.hippRight, left_dims, right_dims)
def save_desc_table(data_params, text_data):
classes = ['AD ', 'MCI', 'NC ']
path_file = data_params['adni_data_des'] + tls.get_convention_name(data_params) + '/Desciption_ADNI_demography.txt'
try:
os.makedirs(os.path.dirname(path_file))
except OSError as e:
if e.errno != errno.EEXIST:
raise
with open(path_file, 'w') as f:
f.write("----------------------------------------------------------------------------------------------------------\n")
f.write("| ADNI-1 description |\n")
f.write("----------------------------------------------------------------------------------------------------------\n")
f.write("| #Subject | Sex (F/M) | Age [min, max]/mean(std) | MMSE [min, max]mean/std |\n")
f.write("----------------------------------------------------------------------------------------------------------\n")
for i in range(3):
f.write("| {} | {} | {} | {} | {} |\n".format(classes[i], text_data[i][1], text_data[i][2], text_data[i][3], text_data[i][4]))
f.write("----------------------------------------------------------------------------------------------------------\n")
f.close()
def process_extracting_3D_data(data_params, lst, data_name, label_code,
indice_ROI):
if ("HIPP" in indice_ROI):
l, r = tls.get_dimensions_cubes_HIPP(
data_params) # exctract only Hippocampus ROI
elif ("PPC" in indice_ROI):
l, r = tls.get_dimensions_cubes_PPC(
data_params) # exctract only Posterior PC ROI
else:
# compute both ROIs (in future)
pass
# get dimensions from the selected ROI (max - min)
names = ['sag', 'cor', 'axi']
list_cord_l = [int(l[i + 1] - l[i]) for i in range(0, 6, 2)]
list_cord_r = [int(r[i + 1] - r[i]) for i in range(0, 6, 2)]
# compute the indexes for selected slices
neighbors = int(
data_params['neighbors']) # used for how many of slices we will select
sag_l, cor_l, axi_l = [[(int(i / 2) - neighbors),
(int(i / 2) + neighbors + 1)] for i in {
"l_" + str(names[j]): list_cord_l[j]
for j in range(len(list_cord_l))
}.values()]
sag_r, cor_r, axi_r = [[(int(i / 2) - neighbors),
(int(i / 2) + neighbors + 1)] for i in {
"r_" + str(names[j]): list_cord_r[j]
for j in range(len(list_cord_r))
}.values()]
data_selection = str(str(data_name).split('_')[1]).upper() + '_' + str(
str(data_name).split('_')[2]).upper()
data_set = str(data_name).split('_')[3]
binary_label = str(data_name).split('_')[4]
target_path = data_params['adni_data_des'] + tls.get_convention_name(
data_params) + '/' + indice_ROI + "/3D/"
data_size = 0
key = 0
for input_line in lst:
#-----------------------------------------------------------------------------------------------------------------------
# Mean ROI (L & R)
# data_roi_mean = prc.process_mean_hippocampus(input_line, data_params) # mean cube
# cross mean between cubes (in future)
# return computed cubes ROIs Left and Right
#-----------------------------------------------------------------------------------------------------------------------
data_roi_left, data_roi_right = prc.process_cube_HIPP(
input_line, data_params) # left, right cube
# [ID, Date, Class, Age, Sex, MMSE, GDS, CDR] # meta-data
subject_ID = str(input_line[1]).split('/')[7]
meta_data = tls.get_meta_data_xml(data_params, subject_ID)
# print(meta_data, binary_label, data_set, label_code[input_line[0]])
model_object_normal = HippModel(data_roi_left, data_roi_right,
meta_data,
int(label_code[input_line[0]]))
data_size += getsizeof(model_object_normal)
model_abs_normal_path = target_path + binary_label + '/' + str(
data_set) + '/' + str(input_line[0]) + '/' + str(key) + str(
'_' + indice_ROI +
'_').upper() + data_name + '_' + subject_ID + '_[' + str(
input_line[0]) + ']' + str('_normal') + '.pkl'
# store model data
daf.save_model(model_object_normal, model_abs_normal_path)
if data_params['flip']:
# Fliped Felt & Right ROI
data_roi_left_flip = prc.flip_3d(data_roi_left)
data_roi_right_flip = prc.flip_3d(data_roi_right)
#cross fliped model
model_object_fliped = HippModel(data_roi_right_flip,
data_roi_left_flip, meta_data,
int(label_code[input_line[0]]))
data_size += getsizeof(model_object_fliped)
model_abs_fliped_path = target_path + binary_label + '/' + str(
data_set) + '/' + str(input_line[0]) + '/' + str(key) + str(
'_' + indice_ROI +
'_').upper() + data_name + '_' + subject_ID + '_[' + str(
input_line[0]) + ']' + str('_fliped') + '.pkl'
# store data model
daf.save_model(model_object_fliped, model_abs_fliped_path)
key += 1
# Progress of computation
print(CP.bg.RED + CP.style.BRIGHT +
" {} % percent complete of 100% ".format(
round(key / len(lst) * 100, 2)) + " " + CP.style.RESET_ALL +
CP.bg.RESET,
end='\r')
#==========================================================================================================================
print("\n", end='\r')
print(CP.style.BRIGHT + "\n>> Data Size is: {} Mb -> {} Gb\n".format(
round((data_size / 1024) *
6.43, 2), round(((data_size / 1024) * 6.43) / 1024, 2)) +
CP.style.RESET_ALL)