Afib_classification/cnn_model.py at master · dumplingman0403/Afib_classification · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Conv1D, MaxPooling1D, Dense, Activation
from tensorflow.keras.layers import Flatten, Dropout, BatchNormalization
from tensorflow.keras.layers import Conv2D, MaxPooling2D
from tensorflow.keras.utils import to_categorical
from tensorflow.keras.callbacks import EarlyStopping, ModelCheckpoint, Callback, TensorBoard
import numpy as np
import os
from datetime import datetime
logdir = './logs'

def get_1DCNN(x_train, y_train, x_test, y_test, name, epochs = 50):

    model = Sequential()
    model.add(Conv1D(16, 3, activation= 'relu', \
        input_shape = (x_train.shape[1], 1)))
    model.add(BatchNormalization())
    model.add(MaxPooling1D(2))

    model.add(Conv1D(32, 3, activation = 'relu'))
    model.add(BatchNormalization())
    model.add(MaxPooling1D(2))

    model.add(Conv1D(64, 3, activation = 'relu'))
    model.add(BatchNormalization())
    model.add(MaxPooling1D(2))

    model.add(Flatten())
    model.add(Dropout(0.5))

    model.add(Dense(512, activation = 'relu'))
    model.add(Dense(128, activation = 'relu'))

    model.add(Dense(4, activation = 'softmax'))

    model.compile(loss = 'categorical_crossentropy',
                  optimizer = 'adam',
                  metrics = ['accuracy'])
    now = datetime.now()
    now = now.strftime("%H:%M:%S")
    callback = [EarlyStopping(monitor = 'val_loss', patience = 8),
                ModelCheckpoint(filepath ='1DCNN_best_model.h5', monitor = 'val_loss', save_best_only = True),
                TensorBoard(log_dir='logs/{}{}'.format(now, name), histogram_freq=1)]

    history = model.fit(x_train, y_train,
                        batch_size = 32,
                        epochs = epochs,
                        callbacks = callback,
                        validation_data = (x_test, y_test))

    return model, history


def get_2DCNN(x_train, y_train, x_test, y_test, name, epochs =50):
    nClass = 4
    model = Sequential()
    model.add(Conv2D(32, (3,3), activation = 'relu', input_shape = x_train.shape[1:]))
    model.add(Conv2D(64, (3,3), activation = 'relu'))
    model.add(MaxPooling2D(pool_size = (2,2)))
    model.add(Dropout(0.25))
    model.add(Conv2D(64, (3,3), activation = 'relu'))
    model.add(Conv2D(64, (3,3), activation = 'relu'))
    model.add(MaxPooling2D(pool_size = (2,2)))
    model.add(Dropout(0.25))

    model.add(Flatten())
    model.add(Dropout(0.2))
    model.add(Dense(512, activation = 'relu'))
    model.add(Dense(128, activation = 'relu'))
    #model.add(Dropout(0.2))
    model.add(Dense(nClass, activation = 'softmax' ))

    model.compile(optimizer = 'adam', loss = 'categorical_crossentropy',
                  metrics = ['accuracy'])
    now = datetime.now()
    now = now.strftime("%H:%M:%S")
    callbacks = [EarlyStopping(monitor='val_loss', patience = 20),
                 ModelCheckpoint(filepath='best_model.h5', monitor='val_loss', save_best_only=True),
                 TensorBoard(log_dir= 'logs/{}{}'.format(now, name), histogram_freq=1)
                 ]
    history = model.fit(x_train, y_train, batch_size = 32, epochs = epochs,
                        callbacks = callbacks, validation_data = (x_test, y_test) )

    #model.load_weights('best_model.h5')
    model.summary()
    return model, history

def InputPreprocess(x_train, y_train, x_test, y_test, model_type = '1d'):
    if model_type == '1d':

        #reshape x
        x_train = np.reshape(x_train, (x_train.shape[0], x_train.shape[1] , 1))
        x_test  = np.reshape(x_test, (x_test.shape[0], x_test.shape[1], 1))

        y_true = y_test
        #reshape y
        y_test = to_categorical(y_test, 4, dtype = 'int8')
        y_train = to_categorical(y_train, 4, dtype = 'int8')

        return x_train, y_train, x_test, y_test, y_true

    elif model_type == '2d':
        #reshape
        x_train /= 255
        x_test  /= 255

        x_train = np.reshape(x_train, (-1, IMG_SIZE, IMG_SIZE, 1))
        x_test = np.reshape(x_test, (-1, IMG_SIZE, IMG_SIZE, 1))


        y_true = y_test
        #to categorical label

        y_train = to_categorical(y_train, num_classes = 4, dtype = 'int8')
        y_test = to_categorical(y_test, num_classes = 4, dtype = 'int8' )

        return x_train, y_train, x_test, y_test, y_true
    else:
        raise ValueError('Select model type 1d or 2d.')