Skip to content

Anuttan/Retinal-Defect-Classification

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 

Repository files navigation

Retinal Defect Classifier powered by Deep Learning

Automated way to detect and classify Retinal Diseases using Deep Learning

Objective

Given a new OCT image, determine whether the images belong to which 4 classes: CNV, DRUSEN, DME and NORMAL. We are using CNN for classification model.

Dataset Description

The dataset is organized into 3 folders (train, test, val) 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).

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

Optical coherence tomography (OCT) images (Spectralis OCT, Heidelberg Engineering, Germany) were selected from retrospective cohorts of adult patients from the Shiley Eye Institute of the University of California San Diego, the California Retinal Research Foundation, Medical Center Ophthalmology Associates, the Shanghai First People’s Hospital, and Beijing Tongren Eye Center between July 1, 2013 and March 1, 2017.

Results

The model is tested using the test data and it returns an Accuracy of 96.88%. This project discusses about the most accurate and fastest method to detect the retinal defects and sort using CNN. This method will be very useful when incorporated in OCT machine so that the time taken to diagnose the defect can be decreased and the Doctors can do the required procedure. The accuracy of this method can be increased by adding more training data and also increasing the number of epochs.

Acknowledgements

Dataset : https://www.kaggle.com/paultimothymooney/kermany2018
Citation : http://www.cell.com/cell/fulltext/S0092-8674(18)30154-5

About

Retinal Defect Classifier powered by Deep Learning.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published