def __getitem__(self, idx):
temp = load_fmri(self.samples[idx]).get_data()
# temp[temp < scoreatpercentile(temp.flatten(),2)] = 0
max_weight = temp.max()
temp = temp / max_weight
min_weight = np.abs(temp.min())
temp = temp + min_weight
return temp, max_weight, min_weight
def __getitem__(self, idx):# for each item in the sample
"""
Because we will use binary cross entropy loss function,
it is crutial that we make sure the data range between 0 and 1
"""
# load the nii.gz format data
temp = load_fmri(self.samples[idx]).get_data()
# get the maximum of the volume
max_weight = temp.max()
# ge the minmum of the volume
min_weight = temp.min()
# standardize
temp_std = (temp - min_weight) / (max_weight - min_weight)
temp_scaled = temp_std * (1 - 0) + 0
return temp_scaled,max_weight,min_weight,self.samples[idx]
def __getitem__(self, idx):
label, name = self.samples[idx]
print('label is %s' % label)
print('name is %s' % name)
load = load_fmri(name).get_data()
npimg = np.array(load, dtype='int32')
npimg_fit = (npimg + 1) * 127.5
npimg_fit = npimg_fit.astype(np.uint8)
transform = transforms.Compose([
transforms.ToPILImage(),
transforms.ToTensor(),
])
img = torch.tensor(transform(npimg_fit))
nplabel = np.asarray(label, dtype='int32')
print('nplabel is %s' % nplabel)
return img, nplabel
# -*- coding: utf-8 -*-
"""
Created on Wed Aug 21 10:53:56 2019
@author: nmei
"""
import os
from glob import glob
from nipype.interfaces import afni
from nilearn.image import resample_img
from nibabel import load as load_fmri
data_dir = '../data/converted'
filtered = glob(os.path.join(data_dir, "*/*/*/*/*", "filtered.nii.gz"))
target_func = load_fmri(
os.path.abspath(os.path.join(data_dir, 'target_func.nii.gz')))
for idx in range(len(filtered)):
picked_data = os.path.abspath(filtered[idx])
resample3d = afni.utils.Resample(voxel_size=(2.386364, 2.386364, 2.4))
resample3d.inputs.in_file = picked_data
resample3d.inputs.outputtype = 'NIFTI_GZ'
resample3d.inputs.out_file = picked_data.replace(
'filtered.nii.gz', 'filtered_resample.nii.gz')
print(resample3d.cmdline)
resample3d.run()
resampled = resample_img(resample3d.inputs.out_file,
target_affine=target_func.affine,
target_shape=(88, 88, 66))
resampled.to_filename(
def __getitem__(self, idx):
return load_fmri(self.samples[idx]).get_data() / load_fmri(
self.samples[idx]).get_data().max()