Scholarlink

<p>Major treats to visual health includes diabetic macular edema (DME), age-related macular degeneration (AMD) and retinal vein occlusion (RVO), which require prompt and correct interpretation for effective treatment. Optical coherence tomography (OCT) is an imaging modality, providing intense cross-sectional views of the retina to aid in diagnosis. Diagnosis and localization of retinal diseases were complicated by the structure of retinal fluids. In order to cope with these challenges, a deep learning architecture, the Adaptive Multi-Domain Fusion Network (AMDF-Net), is initiated to improve the detection of retinal diseases. AMDF-Net assimilates state of the art modules like Hybrid Spectral-Spatial Transformer (HSST) to gain insight about global and local features effectively. Moreover, the Dynamic Attention Fusion (DAF) module enhances the work of the network by specifying the features unique to retinal fluids, and Disease-Inclusive Segmentation (DIS) module makes it easier to accurately diagnose primary fluids. Extensive analyses of publicly available and real-time data reveal that AMDF-Net shows notable results with Dice coefficient of 98. 87% and classification accuracy of 98. 12%. These remarks highlight the potential of AMDF-Net to elevate automated retinal disease analysis and provide valuable assistance in the development of decisions focused on treatment.</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

Multi-scale adaptive fusion network for retinal layer and fluid segmentation in optical coherence tomography B-scans

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年3月30日

查看全文
  • Pavithra Mani,
  • Neelaveni Ramachandran,
  • V. Sowmya,
  • Vinayakumar Ravi,
  • Prasanna Venkatesh Ramesh,
  • Tahani Jaser Alahmadi

摘要

Major treats to visual health includes diabetic macular edema (DME), age-related macular degeneration (AMD) and retinal vein occlusion (RVO), which require prompt and correct interpretation for effective treatment. Optical coherence tomography (OCT) is an imaging modality, providing intense cross-sectional views of the retina to aid in diagnosis. Diagnosis and localization of retinal diseases were complicated by the structure of retinal fluids. In order to cope with these challenges, a deep learning architecture, the Adaptive Multi-Domain Fusion Network (AMDF-Net), is initiated to improve the detection of retinal diseases. AMDF-Net assimilates state of the art modules like Hybrid Spectral-Spatial Transformer (HSST) to gain insight about global and local features effectively. Moreover, the Dynamic Attention Fusion (DAF) module enhances the work of the network by specifying the features unique to retinal fluids, and Disease-Inclusive Segmentation (DIS) module makes it easier to accurately diagnose primary fluids. Extensive analyses of publicly available and real-time data reveal that AMDF-Net shows notable results with Dice coefficient of 98. 87% and classification accuracy of 98. 12%. These remarks highlight the potential of AMDF-Net to elevate automated retinal disease analysis and provide valuable assistance in the development of decisions focused on treatment.