Scholarlink

<p>Glaucoma is a common eye disease affecting several people worldwide. Blindness can be avoided with proper treatment and regular examination. Delayed diagnosis of eye disease causes serious damage to the optic nerve, resulting in loss of vision and blindness. As a result, early disease detection is crucial, and the current research has employed time-consuming machine-learning techniques. It is also difficult to detect eye diseases using computer-aided diagnostic systems because they rely on manually designed features to assess the disease. In order to detect glaucoma, the proposed work presented a novel hybrid deep learning-based GD Pooled Attention assisted Transformer model. The initial step in the proposed detection method is to remove noise and enhance the contrast of the fundus image from the databases using a cross-guided bilateral filter (Cr-GBF). The glaucoma-affected images are collected from ORIGA and RIMONE datasets. Glaucoma is primarily identified by structural deformation in the optic disc region. Thus, significant features&#xa0;are extracted from the images using improved channel spatial attention with AlterNet-K (Im-ChspAN), and disease detection is performed using the proposed Optimized hybrid GD pooled attention former (OpHGpoTr) model. The proposed model’s performance is evaluated using the following metrics: accuracy of 98.68%, precision of 97.06%, sensitivity of 97.07%, specificity of 98.57%, and f-score of 98.31%. The classifier showed better performance than the existing studies.</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

GDPooled transformer: glaucoma detection using pooled attention based transformer with attention mechanism

  • 期刊论文

  • 发表日期：

    2026年1月31日

查看全文
  • V. C. Bharathi,
  • Sharmila Shaik

摘要

Glaucoma is a common eye disease affecting several people worldwide. Blindness can be avoided with proper treatment and regular examination. Delayed diagnosis of eye disease causes serious damage to the optic nerve, resulting in loss of vision and blindness. As a result, early disease detection is crucial, and the current research has employed time-consuming machine-learning techniques. It is also difficult to detect eye diseases using computer-aided diagnostic systems because they rely on manually designed features to assess the disease. In order to detect glaucoma, the proposed work presented a novel hybrid deep learning-based GD Pooled Attention assisted Transformer model. The initial step in the proposed detection method is to remove noise and enhance the contrast of the fundus image from the databases using a cross-guided bilateral filter (Cr-GBF). The glaucoma-affected images are collected from ORIGA and RIMONE datasets. Glaucoma is primarily identified by structural deformation in the optic disc region. Thus, significant features are extracted from the images using improved channel spatial attention with AlterNet-K (Im-ChspAN), and disease detection is performed using the proposed Optimized hybrid GD pooled attention former (OpHGpoTr) model. The proposed model’s performance is evaluated using the following metrics: accuracy of 98.68%, precision of 97.06%, sensitivity of 97.07%, specificity of 98.57%, and f-score of 98.31%. The classifier showed better performance than the existing studies.