f_vali = theano.function([input_var, input2_var, input3_var, target_var], [costV])
from confusionmatrix import ConfusionMatrix
batch_size = 100
num_epochs = 25
train_acc= []
valid_acc = []
cur_loss = 0
loss = []
valid_loss = []
Train = DP.get_paths("/home/xvt131/Biomediq/Data/adni/trainMat")
Test = DP.get_paths("/home/xvt131/Biomediq/Data/adni/valiMat")
import gc
for epoch in range(num_epochs):
cur_loss = 0
val_loss = 0
confusion_valid = ConfusionMatrix(2)
confusion_train = ConfusionMatrix(2)
print epoch
for im in Train:
XY, XZ, YZ, Y_train = DP.Patch_triplanar_para(im, PS)
num_samples_train = Y_train.shape[0]
num_batches_train = num_samples_train // batch_size
for i in range(num_batches_train):
paths = np.empty([0])
for i in listing:
paths = np.hstack((paths, str(path + '/' + i)))
return paths
def get_masks(path, path_L, path_R):
lPath = path_L + "/" + path[-24:]
rPath = path_R + "/" + path[-24:]
leftMask = nib.load(lPath).get_data()
rightMask = nib.load(rPath).get_data()
mask = (leftMask) + rightMask
return mask
Test = TD.get_paths("/home/xvt131/Biomediq/Data/adni/vali_mri")
Left = "/home/xvt131/Biomediq/Data/adni/vali_leftH"
Right = "/home/xvt131/Biomediq/Data/adni/vali_rightH"
for img in Test:
Label = get_masks(img, Left, Right)
Scan = nib.load(img).get_data()
io.savemat("/home/xvt131/Biomediq/Data/adni/valiMat/%s" % (img[-24:-4]),
mdict={
"Scan": Scan,
"Label": Label
})
f_train = theano.function([input_var,input2_var, target_var], [cost], updates=updates)
from confusionmatrix import ConfusionMatrix
batch_size = 250
num_epochs = 25
train_acc = []
valid_loss = []
valid_acc = []
train_loss = []
cur_loss = 0
loss = []
Left = "/home/xvt131/Biomediq/Data/adni/train_leftH"
Right = "/home/xvt131/Biomediq/Data/adni/train_rightH"
Train = DP.get_paths("/home/xvt131/Biomediq/Data/adni/train_mri")
Test = DP.get_paths("/home/xvt131/Biomediq/Data/adni/test_mri")
testLeft = "/home/xvt131/Biomediq/Data/adni/test_leftH"
testRight = "/home/xvt131/Biomediq/Data/adni/test_rightH"
for epoch in range(num_epochs):
cur_loss = 0
val_loss = 0
confusion_valid = ConfusionMatrix(3)
confusion_train = ConfusionMatrix(3)
# Train = np.random.choice(Train_all, 5)
for im in Train:
X_train, XX, Y_train = DP.Patch_3D_para(im,Left, Right, PS, PS2)
num_samples_train = Y_train.shape[0]
num_batches_train = num_samples_train // batch_size