Mostitch is a program written in C++ that was created during my master’s thesis at Poznań University of Technology titled “Spatial coregistration algorithms for angiographic imaging with optical coherence tomography”.

What mostitch does

Mostitch implements an algorithm which as an input takes set of angiography images of eye’s retina. Such images show blood flow in the back of the eye. Additionally it takes an information about relative position of the images. The purpose of mostitch is to automatically stitch those images into uniform mosaic:

Retinal eye images

Input images (shown above on the left) where acquired using Optical Coherence Tomography (OCT).

Why mostitch was developed

The use of Optical Coherence Tomography to capture structure of blood vessels in eye’s retina is done to replace Fluorescent Angiography (FA). FA involves injection of sodium fluorescein into the patient’s blood system, which can cause allergic reaction or headaches. Due to non-invasive characteristic of OCT and no side effects to human health, OCT is getting more and more popular. Unfortunately the quality of the images produced by OCT is lower than those produced by FA. Research facilities around the globe try to compensate this difference by improving OCT equipment and by using image processing algorithms.

Mostitch by automatically stitching OCT angiography images tries to improve overall overview and quality of retina’s blood vessels which can help researchers and doctors to diagnose eye damage.

Mostitch installation and usage

Mostitch is using OpenCV for image processing, libconfig for loading and managing configuration file, and Doxygen for documentation. The program is written in C++.

The project is available on GitHub. Installation can be performed in 3 steps:

mkdir build
make
sudo make install

To run the program:

mostitch path_to_config_file

Where path_to_config_file is pointing to configuration file that consists all mandatory information for program to work properly (e.g. path to input images, information about relative position).