Initial working CNN model

main.py

@@ -2,8 +2,9 @@ import torch
 import torchvision
 
 # FOR DATA
-from utils.preprocess import prepare_datasets
+from utils.preprocess import prepare_datasets, prepare_predict
 from utils.show_image import show_image
+from utils.newCNN import CNN_Net
 from torch.utils.data import DataLoader
 from torchvision import datasets
 
@@ -42,63 +43,115 @@ os.environ["CUDA_VISIBLE_DEVICES"] = "0"  # use id from $ nvidia-smi
 # data & training properties:
 val_split = 0.2     # % of val and test, rest will be train
 seed = 12       # TODO Randomize seed
-'''
-target_rows = 91
-target_cols = 109
-depth = 91
-axis = 1
-num_clinical = 2
-CNN_drop_rate = 0.3
-RNN_drop_rate = 0.1
-CNN_w_regularizer = regularizers.l2(2e-2)
-RNN_w_regularizer = regularizers.l2(1e-6)
-CNN_batch_size = 10
-RNN_batch_size = 5
-val_split = 0.2
-optimizer = Adam(lr=1e-5)
-final_layer_size = 5
-'''
+
+# params = {
+#     "target_rows": 91,
+#     "target_cols": 109,
+#     "depth": 91,
+#     "axis": 1,
+#     "num_clinical": 2,
+#     "CNN_drop_rate": 0.3,
+#     "RNN_drop_rate": 0.1,
+#     # "CNN_w_regularizer": regularizers.l2(2e-2),
+#     # "RNN_w_regularizer": regularizers.l2(1e-6),
+#     "CNN_batch_size": 10,
+#     "RNN_batch_size": 5,
+#     "val_split": 0.2,
+#     "final_layer_size": 5
+# }
+
+properties = {
+    "batch_size":4,
+    "padding":0,
+    "dilation":1,
+    "groups":1,
+    "bias":True,
+    "padding_mode":"zeros",
+    "drop_rate":0
+}
 
 
 # Might have to replace datapaths or separate between training and testing
-model_filepath = '//data/data_wnx1/rschuurs/Pytorch_CNN-RNN'
-mri_datapath = './ADNI_volumes_customtemplate_float32/'
-annotations_datapath = './LP_ADNIMERGE.csv'
+model_filepath = '/data/data_wnx1/rschuurs/Pytorch_CNN-RNN'
+CNN_filepath = '/data/data_wnx1/rschuurs/Pytorch_CNN-RNN/cnn_net.pth'
+mri_datapath = '/data/data_wnx1/rschuurs/Pytorch_CNN-RNN/MRI_volumes_customtemplate_float32/'
+annotations_datapath = './data/data_wnx1/rschuurs/Pytorch_CNN-RNN/LP_ADNIMERGE.csv'
 
 # annotations_file = pd.read_csv(annotations_datapath)    # DataFrame
-
 # show_image(17508)
 
 # TODO: Datasets include multiple labels, such as medical info
 training_data, val_data, test_data = prepare_datasets(mri_datapath, val_split, seed)
-batch_size = 64
 
 # Create data loaders
-train_dataloader = DataLoader(training_data, batch_size=batch_size, shuffle=True)
-test_dataloader = DataLoader(test_data, batch_size=batch_size, shuffle=True)
-val_dataloader = DataLoader(val_data, batch_size=batch_size, shuffle=True)
-
-for X, y in train_dataloader:
-    print(f"Shape of X [N, C, H, W]: {X.shape}")
-    print(f"Shape of y: {y.shape} {y.dtype}")
-    break
-
-
-# Display 10 images and labels.
-x = 0
-while x < 10:
-    train_features, train_labels = next(iter(train_dataloader))
-    print(f"Feature batch shape: {train_features.size()}")
-    img = train_features[0].squeeze()
-    image = img[:, :, 40]
-    label = train_labels[0]
-    plt.imshow(image, cmap="gray")
-    plt.show()
-    print(f"Label: {label}")
-    x = x+1
+train_dataloader = DataLoader(training_data, batch_size=properties['batch_size'], shuffle=True)
+test_dataloader = DataLoader(test_data, batch_size=properties['batch_size'], shuffle=True)
+val_dataloader = DataLoader(val_data, batch_size=properties['batch_size'], shuffle=True)
+
+# for X, y in train_dataloader:
+#     print(f"Shape of X [Channels (colors), Y, X, Z]: {X.shape}")   # X & Y are from TOP LOOKING DOWN
+#     print(f"Shape of Y (Dataset?): {y.shape} {y.dtype}")
+#     break
+
+
+# Display 4 images and labels.
+# x = 1
+# while x < 1:
+#     train_features, train_labels = next(iter(train_dataloader))
+#     print(f"Feature batch shape: {train_features.size()}")
+#     img = train_features[0].squeeze()
+#     print(f"Feature batch shape: {img.size()}")
+#     image = img[:, :, 40]
+#     print(f"Feature batch shape: {image.size()}")
+#     label = train_labels[0]
+#     print(f"Label: {label}")
+#     plt.imshow(image, cmap="gray")
+#     plt.show()
+#     x = x+1
+
+
+train = False
+predict = True
+CNN = CNN_Net(train_dataloader, prps=properties, final_layer_size=2)
+CNN.cuda()
+
+# RUN CNN
+if(train):
+    CNN.train_model(train_dataloader, CNN_filepath, epochs=10)
+    CNN.evaluate_model(val_dataloader)
+
+else:
+    CNN.load_state_dict(torch.load(CNN_filepath))
+    CNN.evaluate_model(val_dataloader)
+
+
+# PREDICT MODE TO TEST INDIVIDUAL IMAGES
+if(predict):
+    on = True
+    print("---- Predict mode ----")
+    print("Integer for image")
+    print("x or X for exit")
+
+    while(on):
+        inp = input("Next image: ")
+        if(inp == None or inp.lower() == 'x' or not inp.isdigit()): on = False
+        else:
+            dataloader = DataLoader(prepare_predict(mri_datapath, [inp]), batch_size=properties['batch_size'], shuffle=True)
+            prediction = CNN.predict(dataloader)
+
+            features, labels = next(iter(dataloader), )
+            img = features[0].squeeze()
+            image = img[:, :, 40]
+            print(f"Expected class: {labels}")
+            print(f"Prediction: {prediction}")
+            plt.imshow(image, cmap="gray")
+            plt.show()
 
 print("--- END ---")
 
+
+
+
 # EXTRA
 
 # will I need these params?

original_model/mci_train.py

@@ -51,7 +51,7 @@ optimizer = Adam(lr=1e-5)
 final_layer_size = 5
 
 model_filepath = '//data/data_wnx3/data_wnx1/_Data/AlzheimersDL/CNN+RNN-2class-1cnn+data'
-mri_datapath = '//data/data_wnx3/data_wnx1/_Data/AlzheimersDL/CNN+RNN-2class-1cnn+data/ADNI_volumes_customtemplate_float32'
+mri_datapath = '//data/data_wnx3/data_wnx1/_Data/AlzheimersDL/CNN+RNN-2class-1cnn+data/MRI_volumes_customtemplate_float32'
 
 
 params_dict = { 'CNN_w_regularizer': CNN_w_regularizer, 'RNN_w_regularizer': RNN_w_regularizer,

original_model/utils/models.py

@@ -46,9 +46,9 @@ class CNN_Net ():
     def __init__ (self, params):
         self.params = params
 
-        self.xls = Input (shape = (self.params.num_clinical,),name='input_xls')
-        self.mri = Input (shape = (self.params.image_shape),name='input_mri')
-        self.jac = Input (shape = (self.params.image_shape),name='input_jac')
+        self.xls = Input (shape = (self.params.num_clinical,),name='input_xls')     # MEDICAL DATA
+        self.mri = Input (shape = (self.params.image_shape),name='input_mri')       # MRI SCAN
+        self.jac = Input (shape = (self.params.image_shape),name='input_jac')       # JAC
         
         xalex3D = XAlex3D(w_regularizer = self.params.CNN_w_regularizer, drop_rate = self.params.CNN_drop_rate, final_layer_size=self.params.final_layer_size)
     
@@ -323,7 +323,8 @@ def XAlex3D(w_regularizer = None, drop_rate = 0., final_layer_size = 50) :
         #                     padding="same", kernel_regularizer = w_regularizer)(conv6_concat)
     
         #Flatten 3D conv network representations
-        flat_conv_6 = Reshape((np.prod(K.int_shape(conv6_concat)[1:]),))(conv6_concat)    
+        flat_conv_6 = Reshape((np.prod(K.int_shape(conv6_concat)[1:]),))(conv6_concat)
+        print(flat_conv_6.shape)
 
         #2-layer Dense network for clinical features
         vol_fc1 = _fc_bn_relu_drop(64,  w_regularizer = w_regularizer,
@@ -347,6 +348,7 @@ def XAlex3D(w_regularizer = None, drop_rate = 0., final_layer_size = 50) :
         return fc2
     return f
 
+
 ###Define pieces of CNN
 def _fc_bn_relu_drop (units, w_regularizer = None, drop_rate = 0., name = None):
     #Defines Fully connected block (see fig. 3 in paper)
@@ -358,6 +360,7 @@ def _fc_bn_relu_drop (units, w_regularizer = None, drop_rate = 0., name = None):
         return fc
     return f
 
+
 def _conv_bn_relu_pool_drop(filters, height, width, depth, strides=(1, 1, 1), padding = 'same', w_regularizer = None, 
                             drop_rate = None, name = None, pool = False):
    #Defines convolutional block (see fig. 3 in paper)
@@ -372,6 +375,7 @@ def _conv_bn_relu_pool_drop(filters, height, width, depth, strides=(1, 1, 1), pa
        return Dropout(drop_rate) (elu)
    return f
 
+
 def _sepconv_bn_relu_pool_drop (filters, height, width, depth, strides = (1, 1, 1), padding = 'same', depth_multiplier = 1, w_regularizer = None, 
                             drop_rate = None, name = None, pool = False):
     #Defines separable convolutional block (see fig. 3 in paper)

original_model/utils/preprocess.py

@@ -12,7 +12,7 @@ from utils.patientsort import PatientSorter
 ##for 2 class model CNN + RNN ##
 
 class DataLoader:
-    """The DataLoader class is intended to be used on images placed in folder ../ADNI_volumes_customtemplate_float32
+    """The DataLoader class is intended to be used on images placed in folder ../MRI_volumes_customtemplate_float32
         
         naming convention is: class_subjectID_imageType.nii.gz
         masked_brain denotes structural MRI, JD_masked_brain denotes Jacobian Determinant 
@@ -26,7 +26,7 @@ class DataLoader:
     
     
     def __init__(self, target_shape, seed = None):
-        self.mri_datapath = '//data/data_wnx3/data_wnx1/_Data/AlzheimersDL/CNN+RNN-2class-1cnn+data/ADNI_volumes_customtemplate_float32'
+        self.mri_datapath = '//data/data_wnx3/data_wnx1/_Data/AlzheimersDL/CNN+RNN-2class-1cnn+data/MRI_volumes_customtemplate_float32'
         self.xls_datapath = '//data/data_wnx3/data_wnx1/_Data/AlzheimersDL/CNN+RNN-2class-1cnn+data'
         self.target_shape = target_shape
         self.seed = seed
@@ -39,7 +39,7 @@ class DataLoader:
 
     
     def get_filenames (self,mri_datapath):
-        '''Puts filenames in ../ADNI_volumes_customtemplate_float32 in
+        '''Puts filenames in ../MRI_volumes_customtemplate_float32 in
         dictionaries according to class (stableMCI, MCItoAD, stableNL and stableAD)
         with keys corresponding to image modality (mri and JD)
         '''

utils/CNN_methods.py

@@ -0,0 +1,78 @@
+from torch import add
+import torch.nn as nn
+import torch.nn.functional as F
+import torch.optim as optim
+
+"""
+Returns a function that convolutes or separable convolutes, normalizes, activates (ELU), pools and dropouts input.
+ 
+Kernel_size = (height, width, depth)
+
+CAN DO SEPARABLE CONVOLUTION IF GROUP = 2!!!! :))))
+"""
+
+def conv_elu_maxpool_drop(in_channel, filters, kernel_size, stride=(1,1,1), padding=0, dilation=1,
+                      groups=1, bias=True, padding_mode='zeros', pool=False, drop_rate=0, sep_conv = False):
+    def f(input):
+
+        # SEPARABLE CONVOLUTION
+        if(sep_conv):
+
+            # SepConv depthwise, Normalizes, and ELU activates
+            sepConvDepthwise = nn.Conv3d(in_channel, filters, kernel_size, stride=stride, padding=padding,
+                                         groups=in_channel, bias=bias, padding_mode=padding_mode)(input)
+
+            # SepConv pointwise
+            # Todo, will stride & padding be correct for this?
+            conv = nn.Conv3d(in_channel, filters, kernel_size=1, stride=stride, padding=padding,
+                                         groups=1, bias=bias, padding_mode=padding_mode)(sepConvDepthwise)
+
+        # CONVOLUTES
+        else:
+            # Convolutes, Normalizes, and ELU activates
+            conv = nn.Conv3d(in_channel, filters, kernel_size, stride=stride, padding=padding, dilation=dilation,
+                             groups=groups, bias=bias, padding_mode=padding_mode)(input)
+
+        normalization = nn.BatchNorm3d(filters)(conv)
+        elu = nn.ELU()(normalization)
+
+        # Pools
+        if (pool):
+            elu = nn.MaxPool2d(kernel_size=3, stride=2, padding=0)(elu)
+
+        return nn.Dropout(p=drop_rate)(elu)
+    return f
+
+
+'''
+Mid_flow in CNN. sep_convolutes 3 times, adds residual (initial input) to 3 times convoluted, and activates through ELU()
+'''
+
+def mid_flow(I, drop_rate, filters):
+    in_channel = None   # TODO, IN_CHANNEL
+
+    residual = I        # TODO, DOES THIS ACTUALLY COPY?
+
+    x = conv_elu_maxpool_drop(in_channel, filters, (3,3,3), drop_rate=drop_rate)(I)
+    x = conv_elu_maxpool_drop(in_channel, filters, (3,3,3), drop_rate=drop_rate)(x)
+    x = conv_elu_maxpool_drop(in_channel, filters, (3, 3, 3), drop_rate=drop_rate)(x)
+
+    x = add(x, residual)
+    x = nn.ELU()(x)
+    return x
+
+
+"""
+Returns a function that Fully Connects (FC), normalizes, activates (ELU), and dropouts input.
+"""
+
+def fc_elu_drop(in_features, units, drop_rate=0):
+    def f(input):
+
+        fc = nn.Linear(in_features, out_features=units)(input)
+        fc = nn.BatchNorm3d(units)(fc)          # TODO 3d or 2d???
+        fc = nn.ELU()(fc)
+        fc = nn.Dropout(p=drop_rate)
+        return fc
+
+    return f

utils/models.py

@@ -0,0 +1,222 @@
+from torch import device, cuda
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+import torch.optim as optim
+import utils.CNN_methods as CNN
+
+
+# METHODS: CONV3D, CONV2D, MAXPOOL, LINEAR, ...
+
+
+class CNN_Net(nn.Module):
+
+    # Defines all properties / layers that can be used
+    def __init__(self, mri_volume, params):
+        super().__init__()
+
+        # self.parameters = nn.ParameterList(params)
+        self.model = xalex3D(mri_volume)
+        self.device = device('cuda:0' if cuda.is_available() else 'cpu')
+
+        print("CNN Initialized. Using: " + str(self.device))
+
+
+    # Implements layers with x data, "running an epoch on x"
+    def forward(self, x):
+        x = F.relu(self.model.f(x, []))         # TODO Add Clinical
+        return x
+
+    # Training data
+    def train(self, trainloader, PATH):
+        criterion = nn.CrossEntropyLoss()
+        optimizer = optim.Adam(self.parameters(), lr=1e-5)
+
+        for epoch in range(2):  # loop over the dataset multiple times
+
+            running_loss = 0.0
+            for i, data in enumerate(trainloader, 0):
+                # get the inputs; data is a list of [inputs, labels]
+                inputs, labels = data[0].to(self.device), data[1].to(self.device)
+
+                # zero the parameter gradients
+                optimizer.zero_grad()
+
+                # forward + backward + optimize
+                outputs = self.forward(inputs)
+                loss = criterion(outputs, labels)
+                loss.backward()
+                optimizer.step()
+
+                # print statistics
+                running_loss += loss.item()
+                if i % 2000 == 1999:  # print every 2000 mini-batches
+                    print(f'[{epoch + 1}, {i + 1:5d}] loss: {running_loss / 2000:.3f}')
+                    running_loss = 0.0
+
+        print('Finished Training')
+
+        torch.save(self.state_dict(), PATH)
+
+
+    def test(self, testloader):
+        correct = 0
+        total = 0
+        # since we're not training, we don't need to calculate the gradients for our outputs
+        with torch.no_grad():
+            for data in testloader:
+                images, labels = data[0].to(self.device), data[1].to(self.devie)
+                # calculate outputs by running images through the network
+                outputs = self.forward(images)
+                # the class with the highest energy is what we choose as prediction
+                _, predicted = torch.max(outputs.data, 1)
+                total += labels.size(0)
+                correct += (predicted == labels).sum().item()
+
+        print(f'Accuracy of the network: {100 * correct // total} %')
+
+
+
+'''
+XAlex3D model.
+
+Functions used:
+- conv_elu_maxpool_drop(in_channel, filters, kernel_size, stride=(1,1,1), padding=0, dilation=1,
+                      groups=1, bias=True, padding_mode='zeros', pool=False, drop_rate=0, sep_conv = False)
+'''
+
+# TODO, figure out IN_CHANNEL
+# TODO, in_channel
+class xalex3D(nn.Module):
+    def __init__(self, mri_volume, drop_rate=0, final_layer_size=50):
+        self.drop_rate = drop_rate
+        self.final_layer_size = final_layer_size
+
+        # self.conv1 = CNN.conv_elu_maxpool_drop(len(next(iter(mri_volume))), 192, (11, 13, 11), stride=(4, 4, 4), drop_rate=self.drop_rate, pool=True)(next(iter(mri_volume)))
+        # self.conv2 = CNN.conv_elu_maxpool_drop(self.conv1.shape(), 384, (5, 6, 5), stride=(1, 1, 1), drop_rate=self.drop_rate, pool=True)(self.conv1)
+        # self.conv_mid_3 = CNN.mid_flow(self.conv2.shape(), self.drop_rate, filters=384)
+        # self.groupConv4 = CNN.conv_elu_maxpool_drop(self.conv_mid_3.shape(), 96, (3, 4, 3), stride=(1, 1, 1), drop_rate=self.drop_rate,
+        #                                        pool=True, groups=2)(self.conv_mid_3)
+        # self.groupConv5 = CNN.conv_elu_maxpool_drop(self.groupConv4.shape(), 48, (3, 4, 3), stride=(1, 1, 1), drop_rate=self.drop_rate,
+        #                                        pool=True, groups=2)(self.groupConv4)
+        #
+        # self.fc1 = CNN.fc_elu_drop(self.groupConv5.shape(), 20, drop_rate=self.drop_rate)(self.groupConv5)
+        #
+        # self.fc2 = CNN.fc_elu_drop(self.fc1.shape(), 50, drop_rate=self.drop_rate)(self.fc1)
+
+
+    def f(self, mri_volume, clinical_inputs):
+
+        conv1 = CNN.conv_elu_maxpool_drop(mri_volume.size(), 192, (11, 13, 11), stride=(4, 4, 4), drop_rate=self.drop_rate, pool=True)(mri_volume)
+
+        conv2 = CNN.conv_elu_maxpool_drop(conv1.size(), 384, (5, 6, 5), stride=(1, 1, 1), drop_rate=self.drop_rate, pool=True)(conv1)
+
+        # MIDDLE FLOW, 3 times sepConv & ELU()
+        print(f"Residual: {conv2.shape}")
+        conv_mid_3 = CNN.mid_flow(conv2, self.drop_rate, filters=384)
+
+        # CONV in 2 groups (left & right)
+        groupConv4 = CNN.conv_elu_maxpool_drop(conv_mid_3.size(), 96, (3, 4, 3), stride=(1, 1, 1), drop_rate=self.drop_rate,
+                                               pool=True, groups=2)(conv_mid_3)
+        groupConv5 = CNN.conv_elu_maxpool_drop(groupConv4.size(), 48, (3, 4, 3), stride=(1, 1, 1), drop_rate=self.drop_rate,
+                                               pool=True, groups=2)(groupConv4)
+
+        # FCs
+        fc1 = CNN.fc_elu_drop(groupConv5.size(), 20, drop_rate=self.drop_rate)(groupConv5)
+
+        fc2 = CNN.fc_elu_drop(fc1.size(), 50, drop_rate=self.drop_rate)(fc1)
+
+        return fc2
+
+
+
+
+
+
+"""     LAST PART:
+        
+        # Flatten 3D conv network representations
+        flat_conv_6 = Reshape((np.prod(K.int_shape(conv6_concat)[1:]),))(conv6_concat)
+
+        # 2-layer Dense network for clinical features
+        vol_fc1 = _fc_bn_relu_drop(64, w_regularizer=w_regularizer,
+                                   drop_rate=drop_rate)(clinical_inputs)
+
+        flat_volume = _fc_bn_relu_drop(20, w_regularizer=w_regularizer,
+                                       drop_rate=drop_rate)(vol_fc1)
+
+        # Combine image and clinical features embeddings
+
+        fc1 = _fc_bn_relu_drop(20, w_regularizer, drop_rate=drop_rate, name='final_conv')(flat_conv_6)
+        flat = concatenate([fc1, flat_volume])
+
+        # Final 4D embedding
+
+        fc2 = Dense(units=final_layer_size, activation='linear', kernel_regularizer=w_regularizer, name='features')(
+            flat)  # was linear activation"""
+
+
+''' FULL CODE:
+
+
+    # First layer
+    conv1_left = _conv_bn_relu_pool_drop(192, 11, 13, 11, strides=(4, 4, 4), w_regularizer=w_regularizer,
+                                         drop_rate=drop_rate, pool=True)(mri_volume)
+   
+    # Second layer
+    conv2_left = _conv_bn_relu_pool_drop(384, 5, 6, 5, w_regularizer=w_regularizer, drop_rate=drop_rate, pool=True)(
+        conv1_left)
+
+    # Introduce Middle Flow (separable convolutions with a residual connection)
+    print('residual shape ' + str(conv2_left.shape))
+    conv_mid_1 = mid_flow(conv2_left, drop_rate, w_regularizer,
+                          filters=384)  # changed input to conv2_left from conv2_concat
+    
+    # Split channels for grouped-style convolution
+    conv_mid_1_1 = Lambda(lambda x: x[:, :, :, :, :192])(conv_mid_1)
+    conv_mid_1_2 = Lambda(lambda x: x[:, :, :, :, 192:])(conv_mid_1)
+
+    conv5_left = _conv_bn_relu_pool_drop(96, 3, 4, 3, w_regularizer=w_regularizer, drop_rate=drop_rate, pool=True)(
+        conv_mid_1_1)
+
+    conv5_right = _conv_bn_relu_pool_drop(96, 3, 4, 3, w_regularizer=w_regularizer, drop_rate=drop_rate, pool=True)(
+        conv_mid_1_2)
+
+    conv6_left = _conv_bn_relu_pool_drop(48, 3, 4, 3, w_regularizer=w_regularizer, drop_rate=drop_rate, pool=True)(
+        conv5_left)
+
+    conv6_right = _conv_bn_relu_pool_drop(48, 3, 4, 3, w_regularizer=w_regularizer, drop_rate=drop_rate, pool=True)(
+        conv5_right)
+
+    conv6_concat = concatenate([conv6_left, conv6_right], axis=-1)
+
+    # convExtra = Conv3D(48, (20,30,20),
+    #                     strides = (1,1,1), kernel_initializer="he_normal",
+    #                     padding="same", kernel_regularizer = w_regularizer)(conv6_concat)
+
+    # Flatten 3D conv network representations
+    flat_conv_6 = Reshape((np.prod(K.int_shape(conv6_concat)[1:]),))(conv6_concat)
+
+    # 2-layer Dense network for clinical features
+    vol_fc1 = _fc_bn_relu_drop(64, w_regularizer=w_regularizer,
+                               drop_rate=drop_rate)(clinical_inputs)
+
+    flat_volume = _fc_bn_relu_drop(20, w_regularizer=w_regularizer,
+                                   drop_rate=drop_rate)(vol_fc1)
+
+    # Combine image and clinical features embeddings
+
+    fc1 = _fc_bn_relu_drop(20, w_regularizer, drop_rate=drop_rate, name='final_conv')(flat_conv_6)
+    # fc2 = _fc_bn_relu_drop (40, w_regularizer, drop_rate = drop_rate) (fc1)
+    flat = concatenate([fc1, flat_volume])
+
+    # Final 4D embedding
+
+    fc2 = Dense(units=final_layer_size, activation='linear', kernel_regularizer=w_regularizer, name='features')(
+        flat)  # was linear activation
+    '''
+
+
+
+
+

utils/newCNN.py

@@ -0,0 +1,116 @@
+from torch import device, cuda
+import torch
+from torch import add
+import torch.nn as nn
+import utils.newCNN_Layers as CustomLayers
+import torch.nn.functional as F
+import torch.optim as optim
+import utils.CNN_methods as CNN
+import copy
+
+
+class CNN_Net(nn.Module):
+    def __init__(self, input, prps, final_layer_size=5):
+        super(CNN_Net, self).__init__()
+        self.final_layer_size = final_layer_size
+        self.device = device('cuda:0' if cuda.is_available() else 'cpu')
+        print("CNN Initialized. Using: " + str(self.device))
+
+        # GETS FIRST IMAGE FOR SIZE
+        data_iter = iter(input)
+        first_batch = next(data_iter)
+        first_features = first_batch[0]
+        image = first_features[0]
+
+        # LAYERS
+        print(f"CNN Model Initialization. Input size: {image.size()}")
+        self.conv1 = CustomLayers.Conv_elu_maxpool_drop(1, 192, (11, 13, 11), stride=(4,4,4), pool=True, prps=prps)
+        self.conv2 = CustomLayers.Conv_elu_maxpool_drop(192, 384, (5, 6, 5), stride=(1,1,1), pool=True, prps=prps)
+        self.conv3_mid_flow = CustomLayers.Mid_flow(384, 384, prps=prps)
+        self.conv4_sepConv = CustomLayers.Conv_elu_maxpool_drop(384, 96,(3, 4, 3), stride=(1,1,1), pool=True, prps=prps,
+                                                                sep_conv=True)
+        self.conv5_sepConv = CustomLayers.Conv_elu_maxpool_drop(96, 48, (3, 4, 3), stride=(1, 1, 1), pool=True,
+                                                                prps=prps, sep_conv=True)
+        self.fc1 = CustomLayers.Fc_elu_drop(113568, 20, prps=prps)      # TODO, concatenate clinical data after this
+        self.fc2 = CustomLayers.Fc_elu_drop(20, final_layer_size, prps=prps)
+
+    # FORWARDS
+    def forward(self, x):
+        x = self.conv1(x)
+        x = self.conv2(x)
+        x = self.conv3_mid_flow(x)
+        x = self.conv4_sepConv(x)
+        x = self.conv5_sepConv(x)
+
+
+        # FLATTEN x
+        flatten_size = x.size(1) * x.size(2) * x.size(3) * x.size(4)
+        x = x.view(-1, flatten_size)
+
+        x = self.fc1(x)
+        x = self.fc2(x)
+        return x
+
+    # TRAIN
+    def train_model(self, trainloader, PATH, epochs):
+        self.train()
+        criterion = nn.CrossEntropyLoss()
+        optimizer = optim.Adam(self.parameters(), lr=1e-5)
+
+        for epoch in epochs:  # loop over the dataset multiple times
+            print(f"Training... {epoch}/{epochs}")
+            running_loss = 0.0
+            for i, data in enumerate(trainloader, 0):
+                # get the inputs; data is a list of [inputs, labels]
+                inputs, labels = data[0].to(self.device), data[1].to(self.device)
+
+                # zero the parameter gradients
+                optimizer.zero_grad()
+
+                # forward + backward + optimize
+                outputs = self.forward(inputs)
+                loss = criterion(outputs, labels)
+                loss.backward()
+                optimizer.step()
+
+                # print statistics
+                running_loss += loss.item()
+                if i % 2000 == 1999:  # print every 2000 mini-batches
+                    print(f'[{epoch + 1}, {i + 1:5d}] loss: {running_loss / 2000:.3f}')
+                    running_loss = 0.0
+
+        print('Finished Training')
+        torch.save(self.state_dict(), PATH)
+
+    # TEST
+    def evaluate_model(self, testloader):
+        correct = 0
+        total = 0
+        self.eval()
+        # since we're not training, we don't need to calculate the gradients for our outputs
+        with torch.no_grad():
+            for data in testloader:
+                images, labels = data[0].to(self.device), data[1].to(self.device)
+                # calculate outputs by running images through the network
+                outputs = self.forward(images)
+                # the class with the highest energy is what we choose as prediction
+                _, predicted = torch.max(outputs.data, 1)
+                total += labels.size(0)
+                print(f"Predicted class vals: {predicted}")
+                correct += (predicted == labels).sum().item()
+
+        print(f'Accuracy of the network on {total} scans: {100 * correct // total}%')
+        self.train()
+
+
+    # PREDICT
+    def predict(self, loader):
+        self.eval()
+        with torch.no_grad():
+            for data in loader:
+                images, labels = data[0].to(self.device), data[1].to(self.device)
+                outputs = self.forward(images)
+                # the class with the highest energy is what we choose as prediction
+                _, predicted = torch.max(outputs.data, 1)
+        self.train()
+        return predicted

utils/newCNN_Layers.py

@@ -0,0 +1,107 @@
+from torch import device, cuda
+import torch
+from torch import add
+import torch.nn as nn
+import torch.nn.functional as F
+import torch.optim as optim
+import utils.CNN_methods as CNN
+import copy
+
+class Conv_elu_maxpool_drop(nn.Module):
+    def __init__(self, input_size, output_size, kernel_size, prps, stride=(1,1,1), pool = False, sep_conv = False, padding = 0):
+        super(Conv_elu_maxpool_drop, self).__init__()
+        self.input_size = input_size
+        self.output_size = output_size
+        self.pool_status = pool
+        self.sep_conv_status = sep_conv
+
+        # LAYERS
+        # TODO Check here, how many groups? just 2? or groups=input_size?
+        if(self.sep_conv_status):
+            self.sepConvDepthwise = nn.Conv3d(input_size, output_size, kernel_size=kernel_size, stride=stride,
+                                              padding=padding, dilation=prps['dilation'], groups=2, bias=prps["bias"], padding_mode=prps["padding_mode"])
+
+        self.conv = nn.Conv3d(input_size, output_size, kernel_size=kernel_size, stride=stride,
+                                          padding=padding, groups=1, bias=prps["bias"], padding_mode=prps["padding_mode"])
+        self.normalization = nn.BatchNorm3d(output_size)
+        self.elu = nn.ELU()
+        self.maxpool = nn.MaxPool3d(kernel_size=3, stride=2, padding=0)
+        self.dropout = nn.Dropout(p=prps['drop_rate'])
+
+        self.weight = nn.Parameter(torch.randn(input_size, output_size))
+        self.bias = nn.Parameter(torch.randn(output_size))
+
+
+    def forward(self, x):
+        # print(f"Forward Input: {x.size()}")
+        if(self.sep_conv_status): x = self.sepConvDepthwise(x)
+        else: x = self.conv(x)
+        x = self.normalization(x)
+        x = self.elu(x)
+        if(self.pool_status): self.maxpool(x)
+        x = self.dropout(x)
+
+        # return torch.matmul(x, self.weight) + self.bias
+        return x        # TODO WHAT??? WEIGHT & BIAS YES OR NO?
+
+
+
+class Mid_flow(nn.Module):
+    def __init__(self, input_size, output_size, prps):
+        super(Mid_flow, self).__init__()
+        self.input_size = input_size
+        self.output_size = output_size
+
+        # LAYERS
+        self.conv = Conv_elu_maxpool_drop(input_size, output_size, kernel_size=(3,3,3), stride=(1,1,1), sep_conv=True, padding='same', prps=prps)
+        self.elu = nn.ELU()
+
+        self.weight = nn.Parameter(torch.randn(input_size, output_size))
+        self.bias = nn.Parameter(torch.randn(output_size))
+
+
+    def forward(self, x):
+        # print("AT MIDFLOW!")
+        residual = x.clone()
+
+        # print(f"Input: {x.size()}")
+        x = self.conv(x)
+        x = self.conv(x)
+        x = self.conv(x)
+        # print(f"Output: {x.size()}")
+
+        x = add(x, residual)
+        x = self.elu(x)
+
+        # return torch.matmul(x, self.weight) + self.bias       # TODO WHAT??? WEIGHT & BIAS YES OR NO?
+        return x
+
+
+
+class Fc_elu_drop(nn.Module):
+    def __init__(self, input_size, output_size, prps):
+        super(Fc_elu_drop, self).__init__()
+        self.input_size = input_size
+        self.output_size = output_size
+
+        # LAYERS
+        self.linear = nn.Linear(input_size, output_size)
+        self.normalization = nn.BatchNorm1d(output_size)
+        self.elu = nn.ELU()
+        self.dropout = nn.Dropout(p=prps['drop_rate'])
+
+        self.weight = nn.Parameter(torch.randn(input_size, output_size))
+        self.bias = nn.Parameter(torch.randn(output_size))
+
+
+    def forward(self, x):
+        # print("AT FC")
+        # print(f"Forward Input: {x.size()}")
+        x = self.linear(x)
+        # print(f"After Linear: {x.size()}")
+        x = self.normalization(x)
+        x = self.elu(x)
+        x = self.dropout(x)
+
+        # return torch.matmul(x, self.weight) + self.bias
+        return x        # TODO WHAT??? WEIGHT & BIAS YES OR NO?

utils/preprocess.py

@@ -1,12 +1,12 @@
-# NEEDS TO BE FINISHED
-# TODO CHECK ABOUT IMAGE DIMENSIONS
-# TODO ENSURE ITERATION WORKS
 import glob
 import nibabel as nib
 import numpy as np
 import random
 import torch
 from torch.utils.data import Dataset
+import torchvision.transforms as transforms
+import re
+
 
 
 '''
@@ -17,6 +17,7 @@ def prepare_datasets(mri_dir, val_split=0.2, seed=50):
     rndm = random.Random(seed)
 
     raw_data = glob.glob(mri_dir + "*")
+
     AD_list = []
     NL_list = []
 
@@ -42,14 +43,42 @@ def prepare_datasets(mri_dir, val_split=0.2, seed=50):
     rndm.shuffle(val_list)
     rndm.shuffle(test_list)
 
+    print(f"DATA INITIALIZATION")
+    print(f"Training size: {len(train_list)}")
+    print(f"Validation size: {len(val_list)}")
+    print(f"Test size: {len(test_list)}")
+
+
+    # # TRANSFORM
+    # transform = transforms.Compose([
+    #     transforms.Grayscale(num_output_channels=1)
+    # ])
+
     train_dataset = CustomDataset(train_list)
     val_dataset = CustomDataset(val_list)
     test_dataset = CustomDataset(test_list)
 
     return train_dataset, val_dataset, test_dataset
 
-    # TODO  Normalize data? Later add / Exctract clinical data? Which data?
+    # TODO  Normalize data? Later add / Extract clinical data? Which data?
+
 
+def prepare_predict(mri_dir, IDs):
+
+    raw_data = glob.glob(mri_dir + "*")
+
+    image_list = []
+
+    # Gets all images and prepares them for Dataset
+    for ID in IDs:
+        pattern = re.compile(ID)
+        matches = [item for item in raw_data if pattern.search(item)]
+        if (len(matches) != 1): print("No image found, or more than one")
+        for match in matches:
+            if "NL" in match: image_list.append((match, 0))
+            if "AD" in match: image_list.append((match, 1))
+
+    return CustomDataset(image_list)
 
 
 '''
@@ -98,8 +127,12 @@ class CustomDataset(Dataset):
     def __getitem__(self, idx):     # RETURNS TUPLE WITH IMAGE AND CLASS_ID, BASED ON INDEX IDX
         mri_path, class_id = self.data[idx]
         mri = nib.load(mri_path)
-        mri_data = mri.get_fdata()
+        image = np.asarray(mri.dataobj)
+        mri_data = np.asarray(np.expand_dims(image, axis=0))
+
+        # mri_data = mri.get_fdata()
         # mri_array = np.array(mri)
         # mri_tensor = torch.from_numpy(mri_array)
         # class_id = torch.tensor([class_id]) TODO return tensor or just id (0, 1)??
+
         return mri_data, class_id

utils/show_image.py

@@ -6,7 +6,7 @@ import matplotlib.pyplot as plt
 '''
 Function to load and show image. If the image is NL, control must be true. 
 '''
-def show_image(image_id, show=True, data_path='./ADNI_volumes_customtemplate_float32/', annotations_path='./LP_ADNIMERGE.csv'):
+def show_image(image_id, show=True, data_path='./MRI_volumes_customtemplate_float32/', annotations_path='./LP_ADNIMERGE.csv'):
     print('Image ID: ' + str(image_id))
 
     annotations_file = pd.read_csv(annotations_path)