A Data-Efficient Approach for Automated Classification of OCT Images using Generative Adversarial Network
Keras = 2.2.4
tensorflow-gpu = 1.9.0
Code should be compatible with Python versions 2.7-3.5. (tested in python2.7)
The database used to obtain the results is taken from the paper, R. Rasti, et al. "Macular OCT classification using a multi-scale convolutional neural network ensemble." IEEE transactions on medical imaging 37.4 (2017): 1024-1034.
For training purposes, the data should be divided into the train and validation set. A portion of the trained data (500 samples per class) is used as labeled data and should be placed in the following path: data/labeled
The remaining trained data should be placed in: data/unlabeled
The model can be trained with the command: python train.py
To reproduce the results of the paper, we have povided the learned model along with the validation set. Use the following command to test the model: python testModel.py
If you use this code for research, please cite
V. Das, S. Dandapat and P. K. Bora, "A Data-Efficient Approach for Automated Classification of OCT Images using Generative Adversarial Network," in IEEE Sensors Letters.
doi: 10.1109/LSENS.2019.2963712
URL: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8949716&isnumber=7862766