Here you can find the implementation of a new strategy for the task of simultaneously locating the optic disc and the fovea in eye fundus images. This method was presented at MICCAI 2018 in Granada (oral presentation). If you find this code useful for your research, please consider citing our paper:
Meyer M.I., Galdran A., Mendonça A.M., Campilho A.. A Pixel-Wise Distance Regression Approach for Joint Retinal Optical Disc and Fovea Detection. In: Medical Image Computing and Computer Assisted Intervention – MICCAI 2018, LNCS, vol 11071, pp 39-47, 2018. doi:10.1007/978-3-030-00934-2_5
I have added an environment.yml file with the updated requirements such that the demo works as expected. Note that the training and prediction scripts were not tested with the new updated libraries, since I have stopped working on this project. Please take this into account when running the code!
In contrast with previous techniques, the proposed method does not attempt to directly detect only OD and fovea centers. Instead, the distance to both locations is regressed for every pixel in a retinal image. This regression problem can be solved by means of a Fully-Convolutional Neural Network. This strategy poses a multi-task-like problem, on which information of every pixel contributes to generate a globally consistent prediction map where likelihood of OD and fovea locations are maximized.
In particular, we make use of the a U-net architecture, while using a loss function suitable to perform pixel-wise distance regression (L2 Loss).
To install all the requirements to run this code, first clone this repository to your local machine:
git clone https://github.com/minesmeyer/od-fovea-regression.git
If you have an Anaconda installation, you can use the conda package manager as follows:
conda env create -f environment.yml
where env_name is the name you want to use for creating a conda environment.
The method was trained and validated on the Messidor dataset. If you wish to replicate the training, the first step is to exclude the Messidor images that do not contain OD and Fovea location information, as provided by [1] (http://www.uhu.es/retinopathy/eng/bd.php).
Split the remaining 1136 images in two (half_1 and half_2) and train your model in two different splits:
- Split 1 will use half_1 for training and half_2 for testing.
- Split 2 will use half_2 for training and half_1 for testing.
To start training, run the script-train-messidor.py file, specifying which split is being used and the directory where you wish
to save the model's weights.
To obtain the resulting predictions on Messidor, use the predict_od_fov.py file, specifying which split is being used
and the directory holding the weights of the model:
The pre-trained weights of the model are in the file best_weights.h5. Please be aware that this model was trained on
a particular random split of the data as specified above. If you use it on a Messidor subset without re-training, some of
the images may belong to the original training set. The example image in the images/ folder is from the test set of the
particular split this model was trained on.
You can run the demo.py script, which takes a retinal image and returns the location of the OD and Fovea:
python demo.py --img_dir images/messidor_test.tif --mask_dir images/messidor_test_mask.tif
This script saves an image to a results/ folder, which will be created automatically if it does not exist.
When possible, input images should be cropped around the FOV, or a mask of the FOV provided as specified above. If no FOV mask is provided, this implementation will still return optic disc and fovea locations, although results may be suboptimal.
Alternatively, you can call the above script with the flag --estimate_fov, which will perform a FOV
segmentation based on a simple thresholding:
python demo.py --img_dir images/messidor_test.tif --estimate_fov
.
- Gegundez-Arias, M.E., Marin, D., Bravo, J.M., Suero, A.: Locating the fovea center position in digital fundus images using thresholding and feature extraction techniques. Computerized Medical Imaging and Graphics 37, 386–393 (2013)