OCT Classification

Dataset Context

Retinal optical coherence tomography (OCT) is an imaging technique used to capture high-resolution cross sections of the retinas of living patients. Approximately 30 million OCT scans are performed each year, and the analysis and interpretation of these images takes up a significant amount of time (Swanson and Fujimoto, 2017)

(A) (Far left) choroidal neovascularization (CNV) with neovascular membrane (white arrowheads) and associated subretinal fluid (arrows). (Middle left) Diabetic macular edema (DME) with retinal-thickening-associated intraretinal fluid (arrows). (Middle right) Multiple drusen (arrowheads) present in early AMD. (Far right) Normal retina with preserved foveal contour and absence of any retinal fluid/edema.

Dataset Content

The dataset is organized into 2 folders (train, test) and contains subfolders for each image category (NORMAL,CNV,DME,DRUSEN). There are 84,495 X-Ray images (JPEG) and 4 categories (NORMAL,CNV,DME,DRUSEN). You can load OCT Dataset by click https://data.mendeley.com/datasets/rscbjbr9sj/2

Images are labeled as (disease)-(randomized patient ID)-(image number by this patient) and split into 4 directories: CNV, DME, DRUSEN, and NORMAL.

Structure

../OCT_Classification
├── checkpoint
│   ├── oct2017_classification_finetune_resnet_10epochs.pth
│   └── PDBL_with_finetune_on_resnet.pkl
├── models
│   ├── pdbl.py
│   ├── resnet.py
│   ├── utils.py
│   └── __init__.py
├── OCT_Dataset
│   ├── OCT2017
│   │   ├── test
│   │   └── train
│   └── Processed_OCT2017
│       ├── test
│       └── train
├── utils
│   ├── dataset.py
│   ├── process.py
│   └── __init__.py
└── trainer.py

The role of each file or folder is as follows:

• checkpoint : save the parameters of models.
• models : architecture of the classifier models which is restnet50 with PDBL.
• OCT_Dataset: OCT2017 is raw data and Processed_OCT2017 is processed data.
• utils: make Dataset and preprocess the data.
• trainer.py: train and test model.

Requirements

• matplotlib==3.3.4
• numpy==1.20.1
• opencv_contrib_python==4.5.4.60
• pandas==1.2.4
• scikit_learn==1.1.3
• torch==1.9.0
• torchvision==0.13.0
• tqdm==4.59.0

Usage

Installation

• Download the repository.

git clone https://github.com/aishangcengloua/OCT_Classification.git

• Install python dependencies.

pip install -r requirements.txt

Training and Inference

python trainer.py --save_dir checkpoint --train_dir OCT_Dataset/Processed_OCT2017/train --val_dir OCT_Dataset/Processed_OCT2017/test --n_classes 4 --train_model True

Reference

• https://github.com/hhyx/OCT-classification
• https://ieeexplore.ieee.org/document/9740140
• https://data.mendeley.com/datasets/rscbjbr9sj/2