def main():
'''
construct = True
if construct == True:
with open('./UHG_T2_data_mean.pkl', 'rb') as filep:
image_dic = pickle.load(filep)
construct_image_from_dic(image_dic)
rotate_image()
translate_image('/usr/share/fsl/5.0/data/standard/MNI152_T1_2mm_brain.nii.gz')
sys.exit(0)
'''
config = Config()
parser = argparse.ArgumentParser(description="Visualize features")
parser.add_argument("paramfile", type=str,
help='Path to the parameter file')
args = parser.parse_args()
config.parse(path.abspath(args.paramfile))
params = config.config.get('parameters')
file_path = params['file_path']
features_file = params['features_path']
features = pickle.load(open(features_file, 'rb'))
out_path = params['out_path']
print(len(features))
IMG_SIZE = params['height'] * params['width'] * params['depth']
filter = np.zeros(IMG_SIZE)
for index in features:
filter[index] = 1
rec_image = reconstruct_image(file_path, filter)
nb.save(rec_image, out_path)
def main():
config = Config()
parser = argparse.ArgumentParser(description="Run SVM")
parser.add_argument("paramfile",
type=str,
help='Path to the parameter file')
args = parser.parse_args()
config.parse(path.abspath(args.paramfile))
params = config.config.get('parameters')
brain_mask_path = params['brain_mask']
fisher_scores_all = params['fisher_score_all']
print("Input fisher scores: ", fisher_scores_all)
for percent in [0.1, 0.3, 0.5, 0.8, 1]:
print("Percentage of features chosen: " + str(percent * 100))
features, scores = get_masked_features(brain_mask_path,
fisher_scores_all, percent)
print("Length of selected features: ", len(features))
mask_features_out = params['out']+'Top_'+\
str(percent*100)+'_features.pkl'
mask_fisher_out = params['out']+'Top_'+\
str(percent*100)+'_scores.pkl'
with open(mask_features_out, 'wb') as p_filep:
pickle.dump(features, p_filep)
with open(mask_fisher_out, 'wb') as p_filep:
pickle.dump(scores, p_filep)
svm = SVM(params)
svm.train_and_predict(False, features)
import os
import pickle
import re
import argparse
import sys
import nibabel as nb
import subprocess
import math
from pathos.multiprocessing import ProcessPool
from dementia_prediction.config_wrapper import Config
from dementia_prediction.data_input import DataInput
from dementia_prediction.cnn_baseline.ensemble_models import CNNEnsembleModels
# Parse the parameter file
config = Config()
parser = argparse.ArgumentParser(description="Run the ADNI Ensemble model")
parser.add_argument("paramfile", type=str, help='Path to the parameter file')
args = parser.parse_args()
config.parse(path.abspath(args.paramfile))
params = config.config.get('parameters')
paths = config.config.get('data_paths')
patients_dict = pickle.load(open(paths['class_labels'], 'rb'))
train_patients = pickle.load(open(paths['train_data'], 'rb'))
valid_patients = pickle.load(open(paths['valid_data'], 'rb'))
print("Valid Patients:", len(valid_patients), "Train:", len(train_patients))
cad_patients = pickle.load(open(paths['cad_dict'], 'rb'))
print("CAD:", len(cad_patients))
classes = params['cnn']['classes']