Team Work

AUTOMATIC DETECTION OF DIABETIC RETINOPATHY USING CNN

Abstract:

A Convolutional Neural Networks (CNNs) approach is proposed to automate the method of Diabetic Retinopathy(DR) screening using Color fundus retinal photography as input. Our network uses CNN along with denoising to identify features like micro-aneurysms and haemorrhages on the retina. Our models were developed leveraging Theano, an open source numerical computation library for Python. We trained this network using a high-end GPU on the publicly available Kaggle dataset. On the data set of over 30,000 images our proposed model achieves around 95% accuracy for the two class classification and around 85% accuracy for the five class classification on around 3,000 validation images.

Keywords—Convolutional Neural Network, Retinopathy, Fundus photography, Image Classification, Deep Learning.

Existing System:

Numerous techniques are tested by researchers in the area for DR classification with encouraging results. Recent work for addressing blood vessel segmentation includes the application of CNN (LeNet-5 architecture) as feature extractor [6].Three heads are used in this model at different layers of the convent which are then fed into three random forests. The final classifier achieved an accuracy of 0.97 and 0.98 on the DRIVE [7] and STARE dataset. An automatic segmentation of blood vessels in Color fundus images is implemented by M.Melinscak et al [8] using deep max-pooling convnet to separate the blood vessels. The model contains a deep max-pooling convolutional neural networks to segment blood vessels. It deployed 10- layer architecture for achieving a maximum accuracy of around 0.94. It was carried around 4-convolutional and 4-max pooling layer with 2 additional fully connected layers for vessel segmentation. Automated analysis of DR using images processing techniques are introduced by Adarsh et al [9].

Proposed System:

In this approach, extraction of retinal blood vessels, exudate, micro-aneurysms, haemorrhages and texture features takes place, followed by construction of Multiclass SVM using area of lesions and texture features. Impressive results are reported using the publicly available datasets DIARETDB0 and DIARETDB1 with accuracy of 0.96 and 0.946 respectively. Considering the heterogeneity of the dataset, the performance of the proposed model is satisfactory. The accuracy of the model can be increased by using other complex denoising techniques.

SYSTEM REQUIREMENTS
SOFTWARE REQUIREMENTS:
• Programming Language : Python
• Font End Technologies : TKInter/Web(HTML,CSS,JS)
• IDE : Jupyter/Spyder/VS Code
• Operating System : Windows 08/10

HARDWARE REQUIREMENTS:

 Processor : Core I3
 RAM Capacity : 2 GB
 Hard Disk : 250 GB
 Monitor : 15″ Color
 Mouse : 2 or 3 Button Mouse
 Key Board : Windows 08/10

For More Details of Project Document, PPT, Screenshots and Full Code
Call/WhatsApp – 9966645624
Email – info@srithub.com

Facebook
Twitter
WhatsApp
LinkedIn

Enquire Now

Leave your details here for more details.