oanhdoan0601/tensorflow

Using tensorflow CNN to Predict IDC in Breast Cancer Histology Images

Tensorflow

Using tensorflow CNN to Predict IDC in Breast Cancer Histology Images Dataset : Breast Histology Images (Classify IDC vs non IDC images) https://www.kaggle.com/simjeg/lymphoma-subtype-classification-fl-vs-cll CNN Model (Lenet) Input layer: [., 50, 50, 3] layer: Conv1 -> ReLu -> MaxPool: [., 25, 25, 36] layer: Conv2 -> ReLu -> MaxPool: [., 13, 13, 36] layer: Conv3 -> ReLu -> MaxPool: [., 7, 7, 36] layer: FC -> ReLu: [., 576] output layer: FC -> ReLu: [., 2] Multilayer Perceptron Input layer: (., 50, 50, 3) hidden layer 1 = 512, hidden layer 2 = 256, hidden layer 3 = 128, Output: 2 We compare two results of different neural network architectures. The neural network is implemented as a python class and the complete TensorFlow session can be saved to or restored from a file. We also implement tensor summaries, which can be visualized with TensorBoard. The output layer gives for each image a probability for IDC=0 and IDC=1.