Scholarlink

<p>Retinopathy of Prematurity (ROP) is the major cause of blindness in preterm infants, making early detection and precise classification essential for effective intervention. This research analyzes the effectiveness of four transfer learning models (InceptionV3, DenseNet201, MobileNet, and Xception), along with several voting-based ensemble approaches, such as Soft Voting, Hard Voting, Weighted Average Voting, and Mix Voting for identifying ROP in binary and multi-class conditions. A novel dataset obtained from Aravind Eye Hospital, Chennai, was utilized for this classification task. Among the individual models, Xception performed better, with F1 score of 96.95% and an accuracy of 97.28%. InceptionV3 achieved a recall of 98.76%, while DenseNet201 excelled in specificity (96.59%). To improve prediction performance, base models were combined using ensemble techniques. Soft Voting aggregates predicted probabilities, Weighted Average Voting considers model-wise performance contributions, and Mix Voting resolves tie cases scenarios in Hard Voting. These ensemble approaches significantly improved performance in both binary and multi-class classification. Specifically, Soft Voting and Mix Voting achieved an accuracy of 98.37%, an F1 score of 98.14%, and a specificity of 99.02% in binary classification. Soft Voting and Mix Voting also performed well in multi-class classification. In general, the findings suggest that ensemble techniques that include voting procedures can greatly improve the accuracy and reliability of ROP classification, thereby facilitating improved clinical decision-making.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Early Detection of Retinopathy of Prematurity Using Voting Classifier-Based Ensemble Deep Learning Models

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年10月1日

查看全文
  • S. Suresh Krishna,
  • S. Sofana Reka

摘要

Retinopathy of Prematurity (ROP) is the major cause of blindness in preterm infants, making early detection and precise classification essential for effective intervention. This research analyzes the effectiveness of four transfer learning models (InceptionV3, DenseNet201, MobileNet, and Xception), along with several voting-based ensemble approaches, such as Soft Voting, Hard Voting, Weighted Average Voting, and Mix Voting for identifying ROP in binary and multi-class conditions. A novel dataset obtained from Aravind Eye Hospital, Chennai, was utilized for this classification task. Among the individual models, Xception performed better, with F1 score of 96.95% and an accuracy of 97.28%. InceptionV3 achieved a recall of 98.76%, while DenseNet201 excelled in specificity (96.59%). To improve prediction performance, base models were combined using ensemble techniques. Soft Voting aggregates predicted probabilities, Weighted Average Voting considers model-wise performance contributions, and Mix Voting resolves tie cases scenarios in Hard Voting. These ensemble approaches significantly improved performance in both binary and multi-class classification. Specifically, Soft Voting and Mix Voting achieved an accuracy of 98.37%, an F1 score of 98.14%, and a specificity of 99.02% in binary classification. Soft Voting and Mix Voting also performed well in multi-class classification. In general, the findings suggest that ensemble techniques that include voting procedures can greatly improve the accuracy and reliability of ROP classification, thereby facilitating improved clinical decision-making.