GrayNet-Keras

• A versatile model for Deep learning application for CT images
• Keras implementation

Publications

(Stone AI) Urinary Stone Detection on CT Images Using Deep Convolutional Neural Networks: Evaluation of Model Performance and Generalization, Radiology AI, 2019, link
(Organ segmentation) C-MIMI 2019 presentation link

How to install

## go to your project folder
$ cd <your project path>
$ git clone https://github.com/LMIC-MGH/graynet_keras

How to use

from keras.layers import Input, GlobalAveragePooling2D, Dense
from keras.models import Model
from graynet_keras import DenseNet121_GrayNet

# Set your input
input_shape = (256,256,1)
input_tensor = Input(input_shape, name='input')

# Add densnet archtecture with pretrained weight of graynet
# last layer of graynet is global average pooling layer
model = Densenet121_GrayNet(input_tensor=input_tensor, weights='graynet', w_reg=None)

# Set a fully connected layer for your model 
output = model.output
output = Dense(units=1, activation='sigmoid', name='fc')(output) ## Your label

model = Model(inputs=input_tensor, outputs=output, name='main_model')
model.summary()

# ... compile your model and run!
# Please see example.ipynb for more example (Jupyter notebook)
# Model archtecture 

How to excute Jupyter example codes

cp GrayNet_example.ipynb ../GrayNet_example.ipynb

Then, see & run the codes!