Glaucoma Recognition, Macular Fovea Detection, and Segmentation of Optic Disc and Cup

This is the code of Glaucoma Grading from Multi-Modality imAges. Task 1 is glaucoma grading, task 2 is macular fovea localization, and task 3 is optic disc/cup segmentation.
Author: Marco Yangjun Liu, [email protected]
Framework: PaddlePaddle (It is a Torch-like DL framework built by Baidu)
GPU: Tesla V100
Mainly achieved by Jupter Notebook in Python, transforms.py and functional.py are written for data augmentation and image processing.

Task 1 Glaucoma Recognition

Glaucoma Grading (Classification, 3 classed): non, early, mid-advanced
Backbone: ResNet101
Cross Entropy Loss
Metric: kappa

Task 2 Macular Fovea Detection

Locate the macular fovea by (x,y)
Loss = 0.5MSE Loss + 0.5Euclidean Distance

Task 3 Segment optic disc/cup from fundus images.

UNet, G Residual Network.
Cross Entropy Loss
Metric: Dice

Citation

If you are using any of these code, please add the following citation to your publication:

@misc{liu2021grm,
  author = {Yangjun Liu},
  title = {Glaucoma Recognition, Macular Fovea Detection, and Segmentation of Optic Disc and Cup},
  year = {2021},
  publisher = {GitHub},
  journal = {GitHub Repository},
  howpublished = {\url{https://github.com/RealMarco/GlaucomaRecognition}}
}