当前位置: X-MOL 学术Displays › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Early detection of diabetic retinopathy from big data in hadoop framework
Displays ( IF 3.7 ) Pub Date : 2021-08-27 , DOI: 10.1016/j.displa.2021.102061
Amartya Hatua 1 , Badri Narayan Subudhi 2 , Veerakumar T. 3 , Ashish Ghosh 4
Affiliation  

In this article, we have designed a fast and reliable Diabetic Retinopathy (DR) detection technique in Hadoop framework, which can identify the early signs of diabetes from eye retinal images. In the proposed scheme the retinal images are classified into five categories: No Diabetic Retinopathy (DR), Mild DR, Moderate DR, Severe DR and Proliferative DR. The proposed scheme follows three distinct steps for classification of the diabetic retinopathy images: feature extraction, feature reduction and image classification. In the initial stage of the algorithm, the Histogram of Oriented Gradients (HOG) is used as a feature descriptor to represent each of the Diabetic Retinopathy images. Principal Component Analysis (PCA) is used for dimensional reduction of HOG features. In the final stage of the algorithm, K-Nearest Neighbors (KNN) classifier is used, in a distributed environment, to classify the retinal images to different classes. Experiments have been carried out on a substantial number of high-resolution retinal images taken under an assortment of imaging conditions. Both left and right eye images are provided for every subject. To handle such large datasets, Hadoop platform is used with MapReduce and Mahout framework for programming. The results obtained by the proposed scheme are compared with some of the close competitive state-of-the-art techniques. The proposed technique is found to provide better results than the existing approaches in terms of some standard performance evaluation measures.



中文翻译:

hadoop框架下大数据早期检测糖尿病视网膜病变

在本文中,我们在 Hadoop 框架中设计了一种快速可靠的糖尿病视网膜病变 (DR) 检测技术,可以从眼睛视网膜图像中识别糖尿病的早期迹象。在提议的方案中,视网膜图像分为五类:无糖尿病视网膜病变 (DR)、轻度 DR、中度 DR、重度 DR 和增殖性 DR。所提出的方案遵循三个不同的糖尿病视网膜病变图像分类步骤:特征提取、特征减少和图像分类。在算法的初始阶段,使用定向梯度直方图 (HOG) 作为特征描述符来表示每个糖尿病视网膜病变图像。主成分分析 (PCA) 用于对 HOG 特征进行降维。在算法的最后阶段,使用 K-Nearest Neighbors (KNN) 分类器,在分布式环境中,将视网膜图像分类为不同的类。已经对在各种成像条件下拍摄的大量高分辨率视网膜图像进行了实验。为每个对象提供左眼和右眼图像。为了处理如此大的数据集,Hadoop 平台与 MapReduce 和 Mahout 框架一起用于编程。通过所提出的方案获得的结果与一些竞争激烈的最先进技术进行了比较。发现所提出的技术在某些标准性能评估措施方面比现有方法提供了更好的结果。已经对在各种成像条件下拍摄的大量高分辨率视网膜图像进行了实验。为每个对象提供左眼和右眼图像。为了处理如此大的数据集,Hadoop 平台与 MapReduce 和 Mahout 框架一起用于编程。通过所提出的方案获得的结果与一些竞争激烈的最先进技术进行了比较。发现所提出的技术在某些标准性能评估措施方面比现有方法提供了更好的结果。已经对在各种成像条件下拍摄的大量高分辨率视网膜图像进行了实验。为每个对象提供左眼和右眼图像。为了处理如此大的数据集,Hadoop 平台与 MapReduce 和 Mahout 框架一起用于编程。通过所提出的方案获得的结果与一些竞争激烈的最先进技术进行了比较。发现所提出的技术在某些标准性能评估措施方面比现有方法提供了更好的结果。通过所提出的方案获得的结果与一些竞争激烈的最先进技术进行了比较。发现所提出的技术在某些标准性能评估措施方面比现有方法提供了更好的结果。通过所提出的方案获得的结果与一些竞争激烈的最先进技术进行了比较。发现所提出的技术在某些标准性能评估措施方面比现有方法提供了更好的结果。

更新日期:2021-09-04
down
wechat
bug