多维密集注意网络用于色彩底部图像中光盘的像素分段
Sreema Ma1, Jayachandran A2, Sudarson Rama Perumal T3
1Department of Electronics and Communication Engineering, Arunachala College of Engineering for Women, Manavilai, India.
概括
这项研究介绍了MDDA-Net,这是一个深度学习模型,用于在视网膜图像中准确的光盘细分. MDDA-Net实现了高精度,超越了现有的早期疾病检测模型.
科学领域:
- 眼科医生 眼科 眼科
- 医疗成像医学成像
- 计算机视觉 计算机视觉
背景情况:
- 视网膜结构如光盘 (OD) 的准确细分对于诊断糖尿病视网膜病变和玻璃眼等疾病至关重要.
- 目前用于OD细分的深度学习模型面临的挑战包括模糊的界限,有限的上下文表示和不充分的特征处理,导致细节丢失.
研究的目的:
- 提出一种新的多维密集注意网络 (MDDA-Net),用于在视网膜图像中精确的像素对光盘的细分.
- 通过增强上下文表示,特征处理和多尺度信息融合来解决现有模型的局限性.
主要方法:
- 拟议的MDDA-Net包含了一个密集的关注区块,用于强大的上下文获取.
- 使用三重注意 (TA) 块来改善像素间关系提取和特征表示.
- 实现多尺度上下文融合 (MCF),以捕获全面的多尺度上下文信息.
主要成果:
- 在具有挑战性的数据集上,MDDA-Net实现了高细分精度,在MESSIDOR上达到99.28%,在ORIGA上达到98.95%.
- 实验结果表明MDDA-Net的性能优于最先进的深度学习模型.
结论:
- MDDA-Net有效地克服了现有的光盘细分方法的局限性.
- 拟议的模型为视网膜图像分析提供了更准确和更强大的解决方案,有助于早期发现疾病.
相关概念视频
01:20Anatomy of the Eyeball
6.8K
The eye is a spherical, hollow structure composed of three tissue layers. The outer layer — the fibrous tunic, comprises the sclera — a white structure — and the cornea, which is transparent. The sclera encompasses some of the ocular surface, most of which is not visible. However, the 'white of the eye' is distinctively visible in humans compared to other species. The cornea, a clear covering at the front of the eye, enables light penetration. The eye's middle...
6.8K
01:24Vision
53.1K
Vision is the result of light being detected and transduced into neural signals by the retina of the eye. This information is then further analyzed and interpreted by the brain. First, light enters the front of the eye and is focused by the cornea and lens onto the retina—a thin sheet of neural tissue lining the back of the eye. Because of refraction through the convex lens of the eye, images are projected onto the retina upside-down and reversed.
53.1K