%%bash

pip install -q -U mdai

import mdai
mdai.__version__

۰.۷.۸

mdai_client = mdai.Client(domain='public.md.ai', access_token="")

Successfully authenticated to public.md.ai.

p = mdai_client.project('PVq9raBJ', path='./lesson1-data')

Using path './lesson1-data' for data.
Preparing annotations export for project PVq9raBJ...                                                
Using cached annotations data for project PVq9raBJ.
Preparing images export for project PVq9raBJ...                                                     
Using cached images data for project PVq9raBJ.

p.show_label_groups()

Label Group, Id: G_BdW4M5, Name: Default group
	Labels:
	Id: L_38Y7Jl, Name: Abdomen
	Id: L_z8xEkB, Name: Chest

# this maps label ids to class ids as a dict obj
labels_dict = {
    'L_38Y7Jl': 0, # Abdomen 
    'L_z8xEkB': 1, # Chest  
}

print(labels_dict)
p.set_labels_dict(labels_dict)

{'L_38Y7Jl': 0, 'L_z8xEkB': 1}

p.show_datasets() 

Datasets:
Id: D_8ogmzN, Name: TRAIN
Id: D_OoJ98E, Name: VAL
Id: D_8oAvmQ, Name: TEST

# create training dataset 
train_dataset = p.get_dataset_by_name('TRAIN')
train_dataset.prepare() 
train_image_ids = train_dataset.get_image_ids()
print(len(train_image_ids))

# create the validation dataset 
val_dataset = p.get_dataset_by_name('VAL')
val_dataset.prepare()
val_image_ids = val_dataset.get_image_ids()
print(len(val_image_ids))

۶۵
۱۰

# visualize a few train images 
mdai.visualize.display_images(train_image_ids[:2], cols=2)
mdai.visualize.display_images(val_image_ids[:2], cols=2)

%tensorflow_version 2.x

import tensorflow as tf

print(tf.__version__)
tf.keras.backend.clear_session()

۲.۱.۰

# Define model parameters 
img_width = 224
img_height = 224
epochs = 100

params = {
    'dim': (img_width, img_height),
    'batch_size': 5,
    'n_classes': 2,
    'n_channels': 3,
    'shuffle': True,
}

base_model = tf.keras.applications.MobileNet(weights='imagenet', include_top=False, 
                                             input_shape=(img_width, img_height, 3))

output = tf.keras.layers.GlobalAveragePooling2D()(base_model.output)
output = tf.keras.layers.Dense(2, activation='softmax')(output)

model = tf.keras.Model(inputs=base_model.input, outputs=output)

model.compile(optimizer=tf.keras.optimizers.Adam(lr=0.0001), 
              loss='categorical_crossentropy', metrics=['accuracy'])

model.summary()

Model: "model"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
input_1 (InputLayer)         [(None, 224, 224, 3)]     0         
_________________________________________________________________
conv1_pad (ZeroPadding2D)    (None, 225, 225, 3)       0         
_________________________________________________________________
conv1 (Conv2D)               (None, 112, 112, 32)      864       
_________________________________________________________________
conv1_bn (BatchNormalization (None, 112, 112, 32)      128       
_________________________________________________________________
conv1_relu (ReLU)            (None, 112, 112, 32)      0         
_________________________________________________________________
conv_dw_1 (DepthwiseConv2D)  (None, 112, 112, 32)      288       
_________________________________________________________________
conv_dw_1_bn (BatchNormaliza (None, 112, 112, 32)      128       
_________________________________________________________________
conv_dw_1_relu (ReLU)        (None, 112, 112, 32)      0         
_________________________________________________________________
conv_pw_1 (Conv2D)           (None, 112, 112, 64)      2048      
_________________________________________________________________
conv_pw_1_bn (BatchNormaliza (None, 112, 112, 64)      256       
_________________________________________________________________
conv_pw_1_relu (ReLU)        (None, 112, 112, 64)      0         
_________________________________________________________________
conv_pad_2 (ZeroPadding2D)   (None, 113, 113, 64)      0         
_________________________________________________________________
conv_dw_2 (DepthwiseConv2D)  (None, 56, 56, 64)        576       
_________________________________________________________________
conv_dw_2_bn (BatchNormaliza (None, 56, 56, 64)        256       
_________________________________________________________________
conv_dw_2_relu (ReLU)        (None, 56, 56, 64)        0         
_________________________________________________________________
conv_pw_2 (Conv2D)           (None, 56, 56, 128)       8192      
_________________________________________________________________
conv_pw_2_bn (BatchNormaliza (None, 56, 56, 128)       512       
_________________________________________________________________
conv_pw_2_relu (ReLU)        (None, 56, 56, 128)       0         
_________________________________________________________________
conv_dw_3 (DepthwiseConv2D)  (None, 56, 56, 128)       1152      
_________________________________________________________________
conv_dw_3_bn (BatchNormaliza (None, 56, 56, 128)       512       
_________________________________________________________________
conv_dw_3_relu (ReLU)        (None, 56, 56, 128)       0         
_________________________________________________________________
conv_pw_3 (Conv2D)           (None, 56, 56, 128)       16384     
_________________________________________________________________
conv_pw_3_bn (BatchNormaliza (None, 56, 56, 128)       512       
_________________________________________________________________
conv_pw_3_relu (ReLU)        (None, 56, 56, 128)       0         
_________________________________________________________________
conv_pad_4 (ZeroPadding2D)   (None, 57, 57, 128)       0         
_________________________________________________________________
conv_dw_4 (DepthwiseConv2D)  (None, 28, 28, 128)       1152      
_________________________________________________________________
conv_dw_4_bn (BatchNormaliza (None, 28, 28, 128)       512       
_________________________________________________________________
conv_dw_4_relu (ReLU)        (None, 28, 28, 128)       0         
_________________________________________________________________
conv_pw_4 (Conv2D)           (None, 28, 28, 256)       32768     
_________________________________________________________________
conv_pw_4_bn (BatchNormaliza (None, 28, 28, 256)       1024      
_________________________________________________________________
conv_pw_4_relu (ReLU)        (None, 28, 28, 256)       0         
_________________________________________________________________
conv_dw_5 (DepthwiseConv2D)  (None, 28, 28, 256)       2304      
_________________________________________________________________
conv_dw_5_bn (BatchNormaliza (None, 28, 28, 256)       1024      
_________________________________________________________________
conv_dw_5_relu (ReLU)        (None, 28, 28, 256)       0         
_________________________________________________________________
conv_pw_5 (Conv2D)           (None, 28, 28, 256)       65536     
_________________________________________________________________
conv_pw_5_bn (BatchNormaliza (None, 28, 28, 256)       1024      
_________________________________________________________________
conv_pw_5_relu (ReLU)        (None, 28, 28, 256)       0         
_________________________________________________________________
conv_pad_6 (ZeroPadding2D)   (None, 29, 29, 256)       0         
_________________________________________________________________
conv_dw_6 (DepthwiseConv2D)  (None, 14, 14, 256)       2304      
_________________________________________________________________
conv_dw_6_bn (BatchNormaliza (None, 14, 14, 256)       1024      
_________________________________________________________________
conv_dw_6_relu (ReLU)        (None, 14, 14, 256)       0         
_________________________________________________________________
conv_pw_6 (Conv2D)           (None, 14, 14, 512)       131072    
_________________________________________________________________
conv_pw_6_bn (BatchNormaliza (None, 14, 14, 512)       2048      
_________________________________________________________________
conv_pw_6_relu (ReLU)        (None, 14, 14, 512)       0         
_________________________________________________________________
conv_dw_7 (DepthwiseConv2D)  (None, 14, 14, 512)       4608      
_________________________________________________________________
conv_dw_7_bn (BatchNormaliza (None, 14, 14, 512)       2048      
_________________________________________________________________
conv_dw_7_relu (ReLU)        (None, 14, 14, 512)       0         
_________________________________________________________________
conv_pw_7 (Conv2D)           (None, 14, 14, 512)       262144    
_________________________________________________________________
conv_pw_7_bn (BatchNormaliza (None, 14, 14, 512)       2048      
_________________________________________________________________
conv_pw_7_relu (ReLU)        (None, 14, 14, 512)       0         
_________________________________________________________________
conv_dw_8 (DepthwiseConv2D)  (None, 14, 14, 512)       4608      
_________________________________________________________________
conv_dw_8_bn (BatchNormaliza (None, 14, 14, 512)       2048      
_________________________________________________________________
conv_dw_8_relu (ReLU)        (None, 14, 14, 512)       0         
_________________________________________________________________
conv_pw_8 (Conv2D)           (None, 14, 14, 512)       262144    
_________________________________________________________________
conv_pw_8_bn (BatchNormaliza (None, 14, 14, 512)       2048      
_________________________________________________________________
conv_pw_8_relu (ReLU)        (None, 14, 14, 512)       0         
_________________________________________________________________
conv_dw_9 (DepthwiseConv2D)  (None, 14, 14, 512)       4608      
_________________________________________________________________
conv_dw_9_bn (BatchNormaliza (None, 14, 14, 512)       2048      
_________________________________________________________________
conv_dw_9_relu (ReLU)        (None, 14, 14, 512)       0         
_________________________________________________________________
conv_pw_9 (Conv2D)           (None, 14, 14, 512)       262144    
_________________________________________________________________
conv_pw_9_bn (BatchNormaliza (None, 14, 14, 512)       2048      
_________________________________________________________________
conv_pw_9_relu (ReLU)        (None, 14, 14, 512)       0         
_________________________________________________________________
conv_dw_10 (DepthwiseConv2D) (None, 14, 14, 512)       4608      
_________________________________________________________________
conv_dw_10_bn (BatchNormaliz (None, 14, 14, 512)       2048      
_________________________________________________________________
conv_dw_10_relu (ReLU)       (None, 14, 14, 512)       0         
_________________________________________________________________
conv_pw_10 (Conv2D)          (None, 14, 14, 512)       262144    
_________________________________________________________________
conv_pw_10_bn (BatchNormaliz (None, 14, 14, 512)       2048      
_________________________________________________________________
conv_pw_10_relu (ReLU)       (None, 14, 14, 512)       0         
_________________________________________________________________
conv_dw_11 (DepthwiseConv2D) (None, 14, 14, 512)       4608      
_________________________________________________________________
conv_dw_11_bn (BatchNormaliz (None, 14, 14, 512)       2048      
_________________________________________________________________
conv_dw_11_relu (ReLU)       (None, 14, 14, 512)       0         
_________________________________________________________________
conv_pw_11 (Conv2D)          (None, 14, 14, 512)       262144    
_________________________________________________________________
conv_pw_11_bn (BatchNormaliz (None, 14, 14, 512)       2048      
_________________________________________________________________
conv_pw_11_relu (ReLU)       (None, 14, 14, 512)       0         
_________________________________________________________________
conv_pad_12 (ZeroPadding2D)  (None, 15, 15, 512)       0         
_________________________________________________________________
conv_dw_12 (DepthwiseConv2D) (None, 7, 7, 512)         4608      
_________________________________________________________________
conv_dw_12_bn (BatchNormaliz (None, 7, 7, 512)         2048      
_________________________________________________________________
conv_dw_12_relu (ReLU)       (None, 7, 7, 512)         0         
_________________________________________________________________
conv_pw_12 (Conv2D)          (None, 7, 7, 1024)        524288    
_________________________________________________________________
conv_pw_12_bn (BatchNormaliz (None, 7, 7, 1024)        4096      
_________________________________________________________________
conv_pw_12_relu (ReLU)       (None, 7, 7, 1024)        0         
_________________________________________________________________
conv_dw_13 (DepthwiseConv2D) (None, 7, 7, 1024)        9216      
_________________________________________________________________
conv_dw_13_bn (BatchNormaliz (None, 7, 7, 1024)        4096      
_________________________________________________________________
conv_dw_13_relu (ReLU)       (None, 7, 7, 1024)        0         
_________________________________________________________________
conv_pw_13 (Conv2D)          (None, 7, 7, 1024)        1048576   
_________________________________________________________________
conv_pw_13_bn (BatchNormaliz (None, 7, 7, 1024)        4096      
_________________________________________________________________
conv_pw_13_relu (ReLU)       (None, 7, 7, 1024)        0         
_________________________________________________________________
global_average_pooling2d (Gl (None, 1024)              0         
_________________________________________________________________
dense (Dense)                (None, 2)                 2050      
=================================================================
Total params: 3,230,914
Trainable params: 3,209,026
Non-trainable params: 21,888

from mdai.visualize import load_dicom_image
import numpy as np
from PIL import Image


class DataGenerator(tf.keras.utils.Sequence):
    def __init__(
        self,
        dataset,
        batch_size=32,
        dim=(32, 32),
        n_channels=1,
        n_classes=10,
        shuffle=True,
        to_RGB=True,
        rescale=False,
    ):
        """Generates data for Keras fit_generator() function.
        """

        # Initialization
        self.dim = dim
        self.batch_size = batch_size

        self.img_ids = dataset.image_ids
        self.imgs_anns_dict = dataset.imgs_anns_dict
        self.dataset = dataset

        self.n_channels = n_channels
        self.n_classes = n_classes
        self.shuffle = shuffle
        self.to_RGB = to_RGB
        self.rescale = rescale
        self.on_epoch_end()

    def __len__(self):
        "Denotes the number of batches per epoch"
        return int(np.floor(len(self.img_ids) / self.batch_size))

    def __getitem__(self, index):
        "Generate one batch of data"

        # Generate indexes of the batch
        indexes = self.indexes[index * self.batch_size : (index + 1) * self.batch_size]

        # Find list of IDs
        img_ids_temp = [self.img_ids[k] for k in indexes]

        # Generate data
        X, y = self.__data_generation(img_ids_temp)

        return X, y

    def on_epoch_end(self):
        "Updates indexes after each epoch"
        self.indexes = np.arange(len(self.img_ids))
        if self.shuffle:
            np.random.shuffle(self.indexes)

    def __data_generation(self, img_ids_temp):
        "Generates data containing batch_size samples"

        # Initialization
        X = np.empty((self.batch_size, *self.dim, self.n_channels))
        y = np.empty((self.batch_size), dtype=int)

        # Generate data
        for i, ID in enumerate(img_ids_temp):
            image = load_dicom_image(ID, to_RGB=self.to_RGB, rescale=self.rescale)
            try:
                image = Image.fromarray(image)
            except Exception:
                print(
                    "Pil.Image can't read image. Possible 12 or 16 bit image. Try rescale=True to "
                    + "scale to 8 bit."
                )

            image = image.resize((self.dim[0], self.dim[1]))

            X[i,] = image

            ann = self.imgs_anns_dict[ID][0]
            y[i] = self.dataset.classes_dict[ann["labelId"]]["class_id"]
        return X, tf.keras.utils.to_categorical(y, num_classes=self.n_classes)

train_generator = DataGenerator(train_dataset, **params)
val_generator = DataGenerator(val_dataset, **params)

# Set callback functions to early stop training and save the best model so far
callbacks = [
    tf.keras.callbacks.EarlyStopping(monitor='val_loss', patience=4, verbose=2, min_delta=1e-3),
    tf.keras.callbacks.ModelCheckpoint(filepath='./model.h5', monitor='val_loss', 
                                       save_best_only=True, verbose=2)
]

history = model.fit(
    train_generator,
    epochs=epochs,
    verbose=1,
    callbacks=callbacks,
    validation_data=val_generator,
)

WARNING:tensorflow:sample_weight modes were coerced from
  ...
    to  
  ['...']
WARNING:tensorflow:sample_weight modes were coerced from
  ...
    to  
  ['...']
Train for 13 steps, validate for 2 steps
Epoch 1/100
۱۱/۱۳ [========================>.....] - ETA: 0s - loss: 0.4748 - accuracy: 0.7818
Epoch 00001: val_loss improved from inf to 0.87730, saving model to ./model.h5
۱۳/۱۳ [==============================] - ۳s 203ms/step - loss: 0.4067 - accuracy: 0.8154 - val_loss: 0.8773 - val_accuracy: 0.7000
Epoch 2/100
۱۱/۱۳ [========================>.....] - ETA: 0s - loss: 0.0112 - accuracy: 1.0000
Epoch 00002: val_loss improved from 0.87730 to 0.18718, saving model to ./model.h5
۱۳/۱۳ [==============================] - ۱s 47ms/step - loss: 0.0104 - accuracy: 1.0000 - val_loss: 0.1872 - val_accuracy: 0.8000
Epoch 3/100
۱۱/۱۳ [========================>.....] - ETA: 0s - loss: 0.0034 - accuracy: 1.0000
Epoch 00003: val_loss improved from 0.18718 to 0.01628, saving model to ./model.h5
۱۳/۱۳ [==============================] - ۱s 47ms/step - loss: 0.0029 - accuracy: 1.0000 - val_loss: 0.0163 - val_accuracy: 1.0000
Epoch 4/100
۱۱/۱۳ [========================>.....] - ETA: 0s - loss: 0.0012 - accuracy: 1.0000
Epoch 00004: val_loss improved from 0.01628 to 0.00735, saving model to ./model.h5
۱۳/۱۳ [==============================] - ۱s 47ms/step - loss: 0.0015 - accuracy: 1.0000 - val_loss: 0.0074 - val_accuracy: 1.0000
Epoch 5/100
۱۲/۱۳ [==========================>...] - ETA: 0s - loss: 9.7492e-04 - accuracy: 1.0000
Epoch 00005: val_loss improved from 0.00735 to 0.00498, saving model to ./model.h5
۱۳/۱۳ [==============================] - ۱s 47ms/step - loss: 0.0010 - accuracy: 1.0000 - val_loss: 0.0050 - val_accuracy: 1.0000
Epoch 6/100
۱۲/۱۳ [==========================>...] - ETA: 0s - loss: 8.0281e-04 - accuracy: 1.0000
Epoch 00006: val_loss improved from 0.00498 to 0.00331, saving model to ./model.h5
۱۳/۱۳ [==============================] - ۱s 46ms/step - loss: 7.4724e-04 - accuracy: 1.0000 - val_loss: 0.0033 - val_accuracy: 1.0000
Epoch 7/100
۱۲/۱۳ [==========================>...] - ETA: 0s - loss: 6.2463e-04 - accuracy: 1.0000
Epoch 00007: val_loss improved from 0.00331 to 0.00251, saving model to ./model.h5
۱۳/۱۳ [==============================] - ۱s 46ms/step - loss: 5.9762e-04 - accuracy: 1.0000 - val_loss: 0.0025 - val_accuracy: 1.0000
Epoch 8/100
۱۱/۱۳ [========================>.....] - ETA: 0s - loss: 3.6382e-04 - accuracy: 1.0000
Epoch 00008: val_loss improved from 0.00251 to 0.00230, saving model to ./model.h5
۱۳/۱۳ [==============================] - ۱s 46ms/step - loss: 5.1158e-04 - accuracy: 1.0000 - val_loss: 0.0023 - val_accuracy: 1.0000
Epoch 9/100
۱۱/۱۳ [========================>.....] - ETA: 0s - loss: 4.4580e-04 - accuracy: 1.0000
Epoch 00009: val_loss did not improve from 0.00230
۱۳/۱۳ [==============================] - 0s 33ms/step - loss: 4.7193e-04 - accuracy: 1.0000 - val_loss: 0.0025 - val_accuracy: 1.0000
Epoch 10/100
۱۱/۱۳ [========================>.....] - ETA: 0s - loss: 3.7999e-04 - accuracy: 1.0000
Epoch 00010: val_loss did not improve from 0.00230
۱۳/۱۳ [==============================] - 0s 35ms/step - loss: 4.1578e-04 - accuracy: 1.0000 - val_loss: 0.0027 - val_accuracy: 1.0000
Epoch 11/100
۱۱/۱۳ [========================>.....] - ETA: 0s - loss: 3.3985e-04 - accuracy: 1.0000
Epoch 00011: val_loss did not improve from 0.00230
۱۳/۱۳ [==============================] - 0s 35ms/step - loss: 3.7973e-04 - accuracy: 1.0000 - val_loss: 0.0027 - val_accuracy: 1.0000
Epoch 12/100
۱۱/۱۳ [========================>.....] - ETA: 0s - loss: 3.8943e-04 - accuracy: 1.0000
Epoch 00012: val_loss improved from 0.00230 to 0.00210, saving model to ./model.h5
۱۳/۱۳ [==============================] - ۱s 47ms/step - loss: 3.4834e-04 - accuracy: 1.0000 - val_loss: 0.0021 - val_accuracy: 1.0000
Epoch 00012: early stopping

import matplotlib.pyplot as plt

print(history.history.keys())

plt.figure()
plt.plot(history.history['accuracy'], 'orange', label='Training accuracy')
plt.plot(history.history['val_accuracy'], 'blue', label='Validation accuracy')
plt.plot(history.history['loss'], 'red', label='Training loss')
plt.plot(history.history['val_loss'], 'green', label='Validation loss')
plt.legend()
plt.show()

%tensorflow_version 2.x

%%bash

pip install -q -U tf-explain

import tensorflow as tf

print(tf.__version__)
tf.keras.backend.clear_session()

۲.۱.۰

model = tf.keras.models.load_model('./model.h5')

test_dataset = p.get_dataset_by_name('TEST')
test_dataset.prepare()

import numpy as np
from PIL import Image
from tf_explain.core.grad_cam import GradCAM
from tf_explain.core.smoothgrad import SmoothGrad
import cv2
import matplotlib.pyplot as plt

img_width = 224
img_height = 224

plt.figure(figsize=(20, 10))

for i, image_id in enumerate(test_dataset.image_ids): 
    
    image = mdai.visualize.load_dicom_image(image_id, to_RGB=True)
    image = Image.fromarray(image)
    image = image.resize((img_width, img_height))
    
    X = np.empty((1, img_width, img_height, 3))
    X[0,:] = image
    y_prob = model.predict(X)
    y_classes = y_prob.argmax(axis=-1)
    
    pred_class = test_dataset.class_id_to_class_text(y_classes[0])
    pred_prob = y_prob[0][y_classes[0]]
    title = f'Pred: {pred_class}, Prob: {pred_prob:.3}'
    
    plt.subplot(len(test_dataset.image_ids), 3, i*3+1)
    plt.title(title)
    plt.imshow(image)
    plt.axis('off')
    
    explainer = GradCAM()
    output = explainer.explain(
        validation_data=(X, None),
        model=model,
        layer_name="conv_pw_13",
        class_index=y_classes[0],
        colormap=cv2.COLORMAP_TURBO,
    )
    
    plt.subplot(len(test_dataset.image_ids), 3, i*3+2)
    plt.imshow(output)
    plt.axis('off')
    
    explainer = SmoothGrad()
    output = explainer.explain(
        validation_data=(X, None),
        model=model,
        class_index=y_classes[0],
    )
    
    plt.subplot(len(test_dataset.image_ids), 3, i*3+3)
    plt.imshow(output, cmap='gray')
    plt.axis('off')
    
plt.show()

WARNING:tensorflow:From /home/leon/.pyenv/versions/3.7.5/lib/python3.7/site-packages/tensorflow_core/python/ops/image_ops_impl.py:1556: div (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.
Instructions for updating:
Deprecated in favor of operator or tf.math.divide.