ICMLA2019 OCT_Classification

A model for classifying different Retinal Diseases using Deep Learning from Optical Coherence Tomography Images. This code is part of the supplementary materials for the IEEE ICMLA 2019 for our paper Optic-net: A Novel Convolutional Neural Network for Diagnosis of Retinal Diseases from Optical Tomography Images. The paper has since been accpeted to IEEE ICMLA 2019 and will be preseneted in December 2019.

IEEE Xplore Digital Library

https://ieeexplore.ieee.org/document/8999264

Arxiv Pre-print

https://arxiv.org/abs/1910.05672

Citation

@inproceedings{kamran2019optic,
  title={Optic-Net: A Novel Convolutional Neural Network for Diagnosis of Retinal Diseases from Optical Tomography Images},
  author={Kamran, Sharif Amit and Saha, Sourajit and Sabbir, Ali Shihab and Tavakkoli, Alireza},
  booktitle={2019 18th IEEE International Conference On Machine Learning And Applications (ICMLA)},
  pages={964--971},
  year={2019}
  organization={IEEE}
}

Pre-requisite

Ubuntu 18.04 / Windows 7 or later
NVIDIA Graphics card

Installation Instruction for Ubuntu

Download and Install Nvidia Drivers
Download and Install via Runfile Nvidia Cuda Toolkit 9.0
Download and Install Nvidia CuDNN 7.1 or later
Install Pip3 and Python3 enviornment
```
sudo apt-get install pip3 python3-dev
```

Install Tensorflow-Gpu version-1.12 and Keras version-2.2.4

sudo pip3 install tensorflow-gpu==1.12.1
sudo pip3 install keras==2.2.4

N.B. If you don’t have GPU, you can also build Tensorflow-CPU

sudo pip3 install tensorflow==1.12.1
sudo pip3 install keras==2.2.4

Install packages from requirements.txt
```
sudo pip3 -r requirements.txt
```
Installation Instruction for Windows
Download and Install Nvidia Drivers
Download and Install Nvidia Cuda Toolkit 9.0
Download and Install Nvidia CuDNN 7.1 or later
Install Pip and Python3 environemnt from Web link

Install Tensorflow-Gpu version-1.12 and Keras version-2.2.4

sudo pip3 install tensorflow-gpu==1.12.1
sudo pip3 install keras==2.2.4

N.B. If you don’t have GPU, you can also build Tensorflow-CPU

sudo pip3 install tensorflow==1.12.1
sudo pip3 install keras==2.2.4

Install packages from requirements.txt
```
sudo pip3 -r requirements.txt
```

Demo

Download the weight file for model trained on Kermany2018 Link
Download the weight file for model trained on Srinivasan2014 Link

Run the inference.py file with the below command

python3 inference.py --imgpath='location of the testing image(single file)' --weights='location to the .h5 file' --dataset='Srinivasan2014 or Kermany2018'

Training on Kermany2018 Dataset

Please cite the paper if you use their data

@article{kermany2018identifying,
title={Identifying medical diagnoses and treatable diseases by image-based deep learning},
author={Kermany, Daniel S and Goldbaum, Michael and Cai, Wenjia and Valentim, Carolina CS and Liang, Huiying and Baxter, Sally L and McKeown, Alex and Yang, Ge and Wu, Xiaokang and Yan, Fangbing and others},
journal={Cell},
volume={172},
number={5},
pages={1122--1131},
year={2018},
publisher={Elsevier}
}

Dataset download link for Kermany2014

https://data.mendeley.com/datasets/rscbjbr9sj/3

Folder structure for training given below. Please make sure it matches with your local repository.

├── data
|   ├──OCT2017
|       ├──train
|           ├──CNV
|           ├──DME
|           ├──DRUSEN
|           └──NORMAL
|       ├──test
|           ├──CNV
|           ├──DME
|           ├──DRUSEN
|           └──NORMAL
├── src
├── LICENSE
├── README.md
├── data_preprocess_sri2014.py
├── inference.py
├── requirements.txt
├── test.py
└── train.py

Type this in terminal to run the train.py file

python3 train.py --dataset=Kermany2018 --datadir=data/OCT2017 --batch=4 --epoch=30 --logdir=optic-net-oct2017-log --snapshot_name=optic-net-oct2017

There are different flags to choose from. Not all of them are mandatory

   '--dataset', type=str, required=True, help='Choosing between 2 OCT datasets', choices=['Srinivasan2014','Kermany2018']
   '--batch', type=int, default=8
   '--input_dim', type=int, default=224
   '--datadir', type=str, required=True, help='path/to/data_directory'
   '--epoch', type=int, default=30
   '--logdir', type=str
   '--weights', type=str,default=None, help='Resuming training from previous weights'
   '--snapshot_name',type=str, default=None, help='Name the saved snapshot'

Training on Srinivasan2014 Dataset

Please cite the paper if you use their data

@article{sri2014,
title={Fully automated detection of diabetic macular edema and dry age-related macular degeneration from optical coherence tomography images},
author={Srinivasan, Pratul P and Kim, Leo A and Mettu, Priyatham S and Cousins, Scott W and Comer, Grant M and Izatt, Joseph A and Farsiu, Sina},
journal={Biomedical optics express},
volume={5},
number={10},
pages={3568--3577},
year={2014},
publisher={Optical Society of America}
}

Dataset download link for Srnivinasan2014

http://people.duke.edu/~sf59/Srinivasan_BOE_2014_dataset.htm

Folder Structure Rearrangement

The folder stucture for this dataset needs to be rearranged. We have provided a .py file to rearrange the structure and also, create the necessary directories and sub-directories. Please run the python file data_preprocess_sri2014.py. In cmd/terminal type the below command. Run it as it is or tinker with it, if you wanna change the directory.

python3 data_preprocess_sri2014.py

The structure will look like this before the rearrangement.

|──2014_BOE_Srinivasan
|  ├──Publication_Dataset
|     ├──AMD1
|     ├──AMD2
|     ├──DME1
|     ├──DME2
|     .....
|     .....
|     └──NORMAL15
|         └──TIFFs
|             └──8bitTIFFs
├── src
├── LICENSE
├── README.md
├── data_preprocess_sri2014.py
├── inference.py
├── requirements.txt
├── test.py
└── train.py

And it will look like this afterwards.

├── data
|   ├──Srinivasan2014
|       ├──Train
|           ├──AMD
|           ├──DME
|           └──NORMAL
|       ├──Test
|           ├──AMD
|           ├──DME
|           └──NORMAL
├── src
├── LICENSE
├── README.md
├── data_preprocess_sri2014.py
├── inference.py
├── requirements.txt
├── test.py
└── train.py