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
Diabetic retinopathy (DR) remains a significant cause of vision loss globally, necessitating accurate diagnosis and timely intervention. This research presents a comprehensive methodology leveraging the U-Net architecture for retinal lesion segmentation, disease staging, and treatment recommendation in diabetic retinopathy. The U- Net model is trained on annotated fundus images from the IDRID dataset, enabling precise segmentation of retinal lesions, particularly hard exudates. Disease staging is performed based on the quantified area of retinal lesions, classified into three stages: Non-proliferative DR, diabetic macular edema, and proliferative DR. Treatment recommendations, including medications, surgeries, and laser treatments, are tailored to each disease stage. Evaluation of the methodology encompasses segmentation accuracy, disease staging performance, and treatment recommendation validity. Key metrics such as Intersection over Union, Dice coefficient, and classification metrics are employed to assess model performance.
Keywords:
Diabetic Retinopathy, UNet Architecture, Disease Staging, Deep Learning, Medical Imaging, Hard Exudates.
Cite Article:
"Segmentation of Hard Exudates and Disease staging of Diabetic Retinopathy using UNET architecture", International Journal of Science & Engineering Development Research (www.ijsdr.org), ISSN:2455-2631, Vol.9, Issue 4, page no.1374 - 1385, April-2024, Available :http://www.ijsdr.org/papers/IJSDR2404201.pdf
Downloads:
000338172
Publication Details:
Published Paper ID: IJSDR2404201
Registration ID:211041
Published In: Volume 9 Issue 4, April-2024
DOI (Digital Object Identifier):
Page No: 1374 - 1385
Publisher: IJSDR | www.ijsdr.org
ISSN Number: 2455-2631
Facebook Twitter Instagram LinkedIn