Cataract Detection System Using Deep Convolutional Neural Network -Deep Learning Project

6,000.00

Cataract Detection System Using Deep Convolutional Neural Network -Deep Learning Project

100 in stock

SKU: Cataract Detection System Using Deep CNN Category:

Description

Cataract Detection System Using Deep Convolutional Neural Network -Deep Learning Project

The alarming cases of cataract within the last decade and the projection of cataract cases within the next few decades call for urgent intervention by early diagnosis. Formal ways of detecting cataract such as physical examination, tests and diagnosis are clinic and professional bound. Hence the need for automation process. Some works have been done on Computer Aided Diagnosis (CAD) of cataract with tools such as Expert systems, which are limited to their knowledgebase thus inaccurate. Early diagnosis of cataract enables quick intervention and treatment. This paper presents a web-based Computer Aided Diagnostic for cataract detection system using Convolutional Neural Network that can be used by any nonprofessional outside the clinic environment. The systems model trained on a data set of 100 eye images using transfer learning which were retrieved from google image search results of “normal human eyes” and “human eye cataract”. It utilized ImageNet model developed in ILSVRC2012 using the Convolutional Neural Network classifier and transferred its knowledge using Transfer learning to train a new model. The new model gained the ability to classify eye images into “Normal” and “Cataractious”. The system was designed to take images as inputs and achieved a Sensitivity of 69%, a Specificity of 86%, Precision of 86%, F-Score of 56% and AUC of 84.56%. Its accuracy score was 78% which was influenced using the model trained during the ImageNet image classification using deep convolutional neural network.

Reviews

There are no reviews yet.

Be the first to review “Cataract Detection System Using Deep Convolutional Neural Network -Deep Learning Project”

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.