INTERNATIONAL JOURNAL OF SCIENTIFIC DEVELOPMENT AND RESEARCH International Peer Reviewed & Refereed Journals, Open Access Journal ISSN Approved Journal No: 2455-2631 | Impact factor: 8.15 | ESTD Year: 2016
open access , Peer-reviewed, and Refereed Journals, Impact factor 8.15
Microaneurysms Detection with Enhanced U-Net Using Fundus Images
Authors Name:
Srijan Chandrakar
, Siddhant Shekhar , Mrs. Nousheen Taj
Unique Id:
IJSDR2207098
Published In:
Volume 7 Issue 7, July-2022
Abstract:
A major factor in diabetes-related blindness is diabetic retinopathy. Microaneurysms (MA) are an early sign of diabetic retinopathy. Therefore, early detection and treatment can help you avoid eyesight loss. Although analysing fundus images with human vision can occasionally be difficult, MA in fundus images is detected by identifying tiny red dots. In this study, a model that can identify microaneurysms was built utilising a deep learning model with a U-Net architecture. Segmentation is a technique for identifying or locating a predetermined area (technically known as pixels) and giving it particular labels. To train a segmentation model, we provide training images and training masks that correspond to those images. Training masks are the equivalent of the images’ “ground truth,” or the labels that have previously been determined and located in the images. We used data from IDRiD and E- Ophtha to train our model. Deep learning has recently gained popularity as a method for enhancing performance in a number of industries, including the categorization and analysis of medical picture data.
Keywords:
Microaneurysms, Diabetic Retinopathy
Cite Article:
"Microaneurysms Detection with Enhanced U-Net Using Fundus Images", International Journal of Science & Engineering Development Research (www.ijsdr.org), ISSN:2455-2631, Vol.7, Issue 7, page no.645 - 651, July-2022, Available :http://www.ijsdr.org/papers/IJSDR2207098.pdf
Downloads:
000337078
Publication Details:
Published Paper ID: IJSDR2207098
Registration ID:201097
Published In: Volume 7 Issue 7, July-2022
DOI (Digital Object Identifier):
Page No: 645 - 651
Publisher: IJSDR | www.ijsdr.org
ISSN Number: 2455-2631
Facebook Twitter Instagram LinkedIn