Anguelova_2024/modelAn_EEG_2024_02_14.py

# Second-level deep neural network
# Version 1 12-05-2022, GV Anguelova
# 05-02-2023: aanpassing splitsing train en test set, zodat het aparte EEG's zijn; tevens wegingsfactor aangepast
# 12-02-2023: complexer maken model met extra hidden layer: model.add(Dense(50, activation='relu'))
# 20-02-2023: complexer maken model met extra Conv2D layer: model.add(Conv2D(25, (3, 3), activation='relu')), met zeropadding, maxpooling en droput

#Import libraries
import numpy
import numpy as np
import tensorflow as tf
import scipy.io as sio
import time
import os

from keras.models import Sequential
from keras.layers import Dense
from keras.layers import Dropout
from keras.layers import Flatten
from keras.constraints import maxnorm
from tensorflow.keras.optimizers import SGD
from keras.layers.convolutional import Conv2D
from keras.layers.convolutional import MaxPooling2D
from keras.utils import np_utils
from keras import backend as K
K.set_image_data_format('channels_last')

from tensorflow.keras.layers import ZeroPadding2D
from tensorflow.keras.optimizers import Adam

from sklearn.metrics import confusion_matrix
from sklearn.metrics import classification_report
from sklearn.utils import class_weight

# Select EEG directories
BaseDir = '/Users/galiaanguelova/Downloads/tryout/'  # directory of the EEG
filename = 'Ambulatory_Epi'  # EEG file
EEGdir = (BaseDir + filename) # + "/Selectie/")
#print(EEGdir)

# fix random seed for reproducibility
seed = 7
numpy.random.seed(seed)

# Load data: selected EEG epochs uit first-level deep neural network
EEGfile = os.path.join(EEGdir, 'train_test_1483_2023_02_05')
EEGdata = sio.loadmat(EEGfile)
EEG_input_train = EEGdata['seizureLocation_selectie_input_train_2023_02_05'] # name variable uit matlab
EEG_input_test = EEGdata['seizureLocation_selectie_input_test_2023_02_05'] # name variable uit matlab

EEG_output_file = os.path.join(EEGdir, 'train_test_1483_2023_02_05')
EEG_output_data = sio.loadmat(EEG_output_file)
EEG_output_train = EEG_output_data['EEG_output_train_2023_02_05']
EEG_output_test = EEG_output_data['EEG_output_test_2023_02_05']

# Reshape input variable
EEG_input_train_reshaped = EEG_input_train.reshape(EEG_input_train.shape[0], EEG_input_train.shape[1], EEG_input_train.shape[2], -1).astype('float32')
EEG_input_test_reshaped = EEG_input_test.reshape(EEG_input_test.shape[0], EEG_input_test.shape[1], EEG_input_test.shape[2], -1).astype('float32')
input_dim = (EEG_input_train.shape[1], EEG_input_train.shape[2], 1) # (250, 18, 1)

# Weight factor
EEG_output_train_cat = np_utils.to_categorical(EEG_output_train)
weight_epil = 2493/1136
class_weight = {0: 1, 1: weight_epil}
EEG_output_test_cat = np_utils.to_categorical(EEG_output_test)

# Create model
def baseline_model():
    model = Sequential()
    model.add(Conv2D(25, (3, 3), activation='relu', input_shape=input_dim, padding='same'))
    model.add(ZeroPadding2D(padding=1))
    model.add(MaxPooling2D(pool_size=(2, 2)))
    model.add(Dropout(0.2))
    ## model.add(Conv2D(25, (3, 3), activation='relu'))
    ## model.add(ZeroPadding2D(padding=1))
    ## model.add(MaxPooling2D(pool_size=(2, 2)))
    ## model.add(Dropout(0.2))
    model.add(Flatten())
    model.add(Dense(100, activation='relu'))
    #model.add(Dense(50, activation='relu'))
    model.add(Dense(2, activation='softmax'))
    # Compile model
    model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])
    print(model.summary())
    return model

# Evaluate model
model = baseline_model()
model.fit(EEG_input_train_reshaped, EEG_output_train_cat, validation_split=0.2, epochs=10, batch_size=50, class_weight=class_weight)

scores = model.evaluate(EEG_input_train_reshaped, EEG_output_train_cat, verbose=0)
print("Accuracy_train: %.2f%%" % (scores[1]*100))

scores = model.evaluate(EEG_input_test_reshaped, EEG_output_test_cat, verbose=0)
print("Accuracy_test: %.2f%%" % (scores[1]*100))

# Save probabilities in mat-file
EEG_pred = model.predict(EEG_input_test_reshaped)
print(EEG_pred) # y_pred
sio.savemat('Ambulatory_Epi/EEG_pred_2023_02_20_v2.mat', {'EEG_pred':EEG_pred}) # ! adjust name and date in file name

# Create classification labels and adjust EEG_pred for calculation confusion matrix
EEG_pred_afgerond = np.argmax(EEG_pred, axis=1)
label = np.argmax(EEG_output_test_cat, axis=1)
print(label)    # y_true/y_test

# Confusion matrix
cm1 = confusion_matrix(label, EEG_pred_afgerond)
print(cm1)

# Confusion matrix, Accuracy, sensitivity and specificity
total1=sum(sum(cm1))
accuracy1=(cm1[0,0]+cm1[1,1])/total1
print('Accuracy : ', accuracy1)

sensitivity1 = cm1[0,0]/(cm1[0,0]+cm1[0,1])
print('Sensitivity : ', sensitivity1)

specificity1 = cm1[1,1]/(cm1[1,0]+cm1[1,1])
print('Specificity : ', specificity1)