Scholarlink

<p>Age-related Macular Degeneration (AMD) is a leading cause of vision loss globally; necessitating early detection and precise diagnosis for effective intervention. This study presents a computer-aided diagnosis (CAD) system for AMD detection and diagnosis over two main stages; that are SegNet-MobileNet for segmentation and feature extraction for classification. The SegNet-MobileNet architecture merges the SegNet’s precise segmentation with the MobileNet’s efficacy; achieving high accuracy in delineating AMD lesions from retinal images. Also, the features extracted from regions of interest (ROIs) capture diverse aspects of retinal structure and texture mandatory for the AMD diagnosis. A dataset of 864 retinal images (accompanied by detailed demographic and clinical information) was collected and analyzed. Through the proper evaluation of ML algorithms, we reported that ensemble methods (such as CatBoost, Extra Trees, and XGBoost) demonstrate superior performance in distinguishing AMD cases, with accuracies exceeding 97%. Model explainability is utilized through the SHapley Additive exPlanations (SHAP) technique; providing insights into feature importance and model decision-making. This study demonstrates the potential of advanced methodologies in contributing to AMD diagnosis and laying the groundwork for future research directions in the ophthalmic imaging and diagnostic field. While promising, these results indicate the need for larger, multi-center validation to ensure generalizability.</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

Advanced computer-aided diagnosis for age-related macular degeneration: integrating segmentation and feature extraction for precise diagnosis

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月9日

查看全文
  • Hossam Magdy Balaha,
  • Mohamed T. Azam,
  • Ali Mahmoud,
  • Mohammed Ghazal,
  • Ayman El-Baz

摘要

Age-related Macular Degeneration (AMD) is a leading cause of vision loss globally; necessitating early detection and precise diagnosis for effective intervention. This study presents a computer-aided diagnosis (CAD) system for AMD detection and diagnosis over two main stages; that are SegNet-MobileNet for segmentation and feature extraction for classification. The SegNet-MobileNet architecture merges the SegNet’s precise segmentation with the MobileNet’s efficacy; achieving high accuracy in delineating AMD lesions from retinal images. Also, the features extracted from regions of interest (ROIs) capture diverse aspects of retinal structure and texture mandatory for the AMD diagnosis. A dataset of 864 retinal images (accompanied by detailed demographic and clinical information) was collected and analyzed. Through the proper evaluation of ML algorithms, we reported that ensemble methods (such as CatBoost, Extra Trees, and XGBoost) demonstrate superior performance in distinguishing AMD cases, with accuracies exceeding 97%. Model explainability is utilized through the SHapley Additive exPlanations (SHAP) technique; providing insights into feature importance and model decision-making. This study demonstrates the potential of advanced methodologies in contributing to AMD diagnosis and laying the groundwork for future research directions in the ophthalmic imaging and diagnostic field. While promising, these results indicate the need for larger, multi-center validation to ensure generalizability.