About

This project implements algorithm for retinal diseases (Age-related macular degeneration, Diabetic retinopathy) classification based on images of the patient's retina (fundus images).

It uses OpenCV (Open Source Computer Vision) library.

Examples

Below is an example of what you can achieve with this project. On first image you see fundus with some exudates. Image has no special markings, it is just output from fundus camera with no metadata about eye or diseases.

Input fundus image:

Output image (blue border means that yellow stuff is found - probably exudate and red border means that red stuff is found - probably hemorrhage):

Installation

1. CMake, version 3.3 or later:

wget http://www.cmake.org/files/v3.4/cmake-3.4.1.tar.gz
tar -xvzf cmake-3.4.1.tar.gz
cd cmake-3.4.1/
./configure
make
sudo make install
sudo update-alternatives --install /usr/bin/cmake cmake /usr/local/bin/cmake 1 --force

2. Compilation of source code:

mkdir build
cd build
cmake ../
make

Usage manual

Usage: ./retinaDiseaseClasifier [ OPTIONS ... ] IMAGE

 -h       	Prints usage manual to stdout.
 -o PATH  	Save image with marked findings in it ti file.
 -m PATH  	Set path to statistical models for clasificator.
            Mandatory in combination with '-t'.
 -d PATH  	Set path to database files of property vectors of classified findings.
            Mandatory in combination with '-c' and '-t'. 
 -c [0-2] 	Manual classification of findings. 
            Value specifies type of desired specifies: 0 = drusen, 1 = exudates, 2 = hemorrhages.
 -t [0-2] 	Reset classificator by new database. 
            Value specifies type of desired specifies: 0 = drusen, 1 = exudates, 2 = hemorrhages.

License

This content is released under the (http://opensource.org/licenses/MIT) MIT License.