| 梁礼明,王成斌,康婷,钟奕.多层级语义融合与特征耦合Transformer的视网膜血管分割[J].电子测量与仪器学报,2025,39(12):155-166 |
| 多层级语义融合与特征耦合Transformer的视网膜血管分割 |
| Multi-level semantic fusion and feature-coupled Transformer forretinal vessel segmentation |
| |
| DOI： |
| 中文关键词: 视网膜血管分割 列非均匀校正模块 特征耦合Transformer 多层级语义融合 |
| 英文关键词:retinal vessel segmentation column non-uniformity correction module feature-coupled Transformer multi-level semantic fusion |
| 基金项目:国家自然科学基金资助项目(51365017,61463018)、江西省自然科学基金资助项目(20192BAB205084)、江西省教育厅科学技术研究青年项目（GJJ2200848）资助 |
|
|
| Author | Institution |
| Liang Liming | School of Electrical Engineering and Automation, Jiangxi University of Science and Technology, Ganzhou
341000, China |
| Wang Chengbin | School of Electrical Engineering and Automation, Jiangxi University of Science and Technology, Ganzhou
341000, China |
| Kang Ting | School of Electrical Engineering and Automation, Jiangxi University of Science and Technology, Ganzhou
341000, China |
| Zhong Yi | School of Electrical Engineering and Automation, Jiangxi University of Science and Technology, Ganzhou
341000, China |
|
| 摘要点击次数: 550 |
| 全文下载次数: 453 |
| 中文摘要: |
| 针对视网膜血管分割中存在微细血管提取困难、成像对比度低及病灶信息干扰等问题,提出一种多层级语义融合与特征耦合Transformer的视网膜血管分割算法。首先,采用列非均匀校正模块构建双联合特征提取端,有效保留血管纹理信息,增强模型提取微细血管的能力；然后,在编解码连接处引入特征耦合Transformer模块,增强血管特征的表达能力,使算法能更准确辨别血管语义特征；最后,在编码端加入多层级语义融合模块,有效抑制背景噪声干扰,着重关注血管区域特征。在公共数据集DRIVE、STARE和CHASE_DB1上进行实验,其灵敏度分别为80.30%、80.84%和82.43%,准确率分别为97.11%、97.61%和97.63%,F1值分别为82.96%、83.76%和81.48%。实验结果表明,该方法在血管分割精度、微细血管结构完整性保持以及复杂病灶区域处理方面均表现优异,整体性能优于现有大部分先进算法,并在泛化能力和鲁棒性方面展现出良好潜力,为视网膜血管疾病的智能辅助诊断提供更为可靠的技术支持。 |
| 英文摘要: |
| A retinal vessel segmentation algorithm based on multi-level semantic fusion and feature-coupled Transformer was proposed to address challenges such as the difficulty in extracting fine vessels, low imaging contrast, and interference from lesion information in retinal images. First, a column non uniformity correction module was used to construct a dual-branch joint feature extraction module, which effectively preserved vessel texture information and enhanced the model’s ability to extract fine vessels. Then, a feature-coupled Transformer module was introduced at the encoder-decoder connection to enhance the representation of vessel features, enabling more accurate recognition of vascular semantic information. Finally, a multi-level semantic fusion module was added to the encoder to suppress background noise interference and focus on vessel-related features. Experiments were conducted on public datasets DRIVE, STARE, and CHASE_DB1. The sensitivity achieved was 80.30%, 80.84% and 82.43%, respectively, the accuracy reached 97.11%, 97.61% and 97.63%, respectively, and the F1-scores were 82.96%, 83.76% and 81.48%, respectively. The experimental results indicate that the proposed method achieves superior segmentation accuracy, preserves the integrity of fine vascular structures, and effectively handles complex lesion regions; overall performance surpasses that of most existing advanced methods and exhibits strong generalization and robustness, thereby providing more reliable technical support for intelligent auxiliary diagnosis of retinal vascular diseases. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
