西北大学 数学学院,陕西 西安 710127
赵婷婷,女,副教授,硕士生导师,从事生物数学研究,ztt0701@nwu.edu.cn。
收稿:2026-05-20,
修回:2026-06-05,
纸质出版:2026-10-25
移动端阅览
赵婷婷, 张元元. 基于双编码结构的病理性近视病灶分割[J]. 西北大学学报(自然科学版), 2026,56(5):1115-1124.
Zhao Tingting, Zhang Yuanyuan. Pathological myopia lesion segmentation using dual encoding structure[J]. Journal of Northwest University (Natural Science Edition), 2026, 56(5): 1115-1124.
赵婷婷, 张元元. 基于双编码结构的病理性近视病灶分割[J]. 西北大学学报(自然科学版), 2026,56(5):1115-1124. DOI: 10.16152/j.cnki.xdxbzr.2026-05-012.
Zhao Tingting, Zhang Yuanyuan. Pathological myopia lesion segmentation using dual encoding structure[J]. Journal of Northwest University (Natural Science Edition), 2026, 56(5): 1115-1124. DOI: 10.16152/j.cnki.xdxbzr.2026-05-012.
病理性近视病灶的精准自动化分割在眼科临床具有关键意义,可减轻眼科医生分析眼底图像负担、提速诊断并为治疗提供可靠依据。但病灶形状大小各异,部分边界模糊且与周围组织相似,这使确定其位置和大小任务复杂,限制了深度学习模型的分割性能。为此,该文提出一种基于双编码器的深度网络,用于处理病灶复杂形态特征。该网络使用ResNet-34编码器提取图像局部特征,同时,引入PVTv2编码器获取全局特征,全面把握病灶及周围组织信息。借助软注意力机制融合双编码器特征,并依据病灶特征自动分配权重,聚焦关键区域,提升不规则及边界模糊病灶的分割精度。该文设计了空间通道增强注意力模块,使模型敏锐捕捉病灶与周围组织的细微特征差异,精准确定病灶边界和范围。在PALM19数据集上的实验结果显示,所提方法的Dice系数、交并比IoU分别为84.63%、75.71%,为病理性近视病灶精准分割提供了有效解决方案。
In ophthalmic clinical practice
the precise automated segmentation of pathological myopia lesions holds great significance. It can relieve ophthalmologists of the burden of analyzing fundus images
accelerate diagnosis
and offer a reliable basis for treatment. However
lesions vary significantly in shape and size
with some boundaries being blurred and similar to the surrounding tissues. This makes determining their location and size a complex task and restricts the segmentation performance of deep learning models. Therefore
this article puts forward a deep network based on dual encoders for handling the complex morphological features of lesions. This network uses ResNet-34 encoder to extract local features of images. Meanwhile
it introduces a PVTv2 encoder to obtain global features
thus comprehensively grasping the information of lesions and surrounding tissues. By leveraging the soft attention mechanism to fuse the features of the dual encoders and automatically assign weights according to lesion characteristics
focusing on key regions
the segmentation accuracy of irregular and boundary-blurred lesions is enhanced. Simultaneously
this article designs a spatial-channel enhanced attention module
enabling the model to sensitively capture the subtle feature differences between lesions and surrounding tissues
and accurately determine the boundaries and ranges of lesions. Finally
the numerical experimental results on the PALM19 dataset demonstrated that the Dice coefficient and intersection of union (IOU) of the proposed method were 84.63% and 75.71%
providing an effective solution for the accurate segmentation of pathological myopia lesions.
王春燕 , 艾育德 . 病理性近视病因学研究的最新进展 [J ] . 内蒙古医学杂志 , 2004 , 36 ( 1 ): 41 - 44 .
Wang C Y , Ai Y D . Latestprogress of study on etiology of pathological myopia [J ] . Inner Mongolia Medical Journal , 2004 , 36 ( 1 ): 41 - 44 .
张望 , 刘康 , 高建华 , 等 . 国际近视研究院关于病理性近视的报告 [J ] . 中华实验眼科杂志 , 2023 , 41 ( 4 ): 366 - 387 .
Yorston D . Retinal diseases and VISION 2020 [J ] . Community Eye Health , 2003 , 16 ( 46 ): 19 - 20 .
李永安 . 基于深度学习的病理性近视识别及病灶分割技术研究与实现 [D ] . 桂林 : 桂林理工大学 , 2023 .
茹月 , 师燕芸 . 超广角眼底成像在病理性近视周边视网膜形态改变中的诊疗价值 [J ] . 中国药物与临床 , 2021 , 21 ( 19 ): 3277 - 3280 .
Ronneberger O , Fischer P , Brox T . U-Net: Convolutional networks for biomedical image segmentation [C ] // Medical Image Computing and Computer-Assisted Intervention-MICCAI 2015 . Cham : Springer , 2015 : 234 - 241 .
Oktay O , Schlemperer J , Folgoc L L , et al . Attention U-Net: Learning where to look for the pancreas [J ] . Medical Image Analysis , 2020 , 58 : 101563 .
Chen J N , Lu Y Y , Yu Q H , et al . TransUNet: Transformers make strong encoders for medical image segmentation [EB/OL ] . ( 2021-09-08 ) [ 2025-09-20 ] . https://arxiv.org/abs/2102.04306 https://arxiv.org/abs/2102.04306 .
Vaswani A , Shazeer N , Parmar N , et al . Attention is all you need [C ] // Proceedings of the 31st International Conference on Neural Information Processing Systems . Red Hook : Curran Associates Inc. , 2017 : 5998 - 6008 .
Zhang Y D , Liu H Y , Hu Q . TransFuse: Fusing transformers and CNNs for medical image segmentation [C ] // Medical Image Computing and Computer Assisted Intervention-MICCAI 2021 . Cham : Springer , 2021 : 14 - 24 .
He K M , Zhang X Y , Ren S Q , et al . Deep residual learningforimage recognition [C ] // 2016 IEEE Conference on Computer Vision and Pattern Recognition . Las Vegas : IEEE , 2016 : 770 - 778 .
Wu H S , Chen S H , Chen G L , et al . FAT-Net: Feature adaptive transformers for automated skin lesion segmentation [J ] . Medical Image Analysis , 2022 , 76 : 102327 .
Wang W H , Xie E Z , Li X , et al . PVT v2: Improved baselines with pyramid vision transformer [J ] . Computational Visual Media , 2022 , 8 ( 3 ): 415 - 424 .
Jha D , Tomar N K , Biswas K , et al . CT liver segmentation via PVT-based encoding and refined decoding [C ] // 2024 IEEE International Symposium on Biomedical Imaging (ISBI) . Athens : IEEE , 2024 : 1 - 5 .
Tang F H , Ding J R , Quan Q , et al . CMUNEXT: An efficient medical image segmentation network based on large kernel and skip fusion [C ] // 2024 IEEE International Symposium on Biomedical Imaging (ISBI) . Athens : IEEE , 2024 : 1 - 5 .
Fu H Z , Li F , Orlando J I , et al . PALM: Pathologic myopia challenge [DB/OL ] . IEEE Dataport, 2019 . [ 2025-09-20 ] . https://doi.org/10.21227/55pk-8z03 https://doi.org/10.21227/55pk-8z03 .
Rauf N , Gilani S O , Waris A . Automatic detection of pathologic myopia using machine learning [J ] . Scientific Reports , 2021 , 11 : 16570 .
符灵利 . 基于深度学习的眼底图像血管分割与病理性近视检测研究 [D ] . 武汉 : 华中师范大学 , 2023 .
Rahman M M , Munir M , Marculescu R . EMCAD: Efficient multi-scale convolutional attention decoding for medical image segmentation [C ] // 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) . Seattle : IEEE , 2024 : 11769 - 11779 .
Li H , Zhai D H , Xia Y Q . ERDUnet: An efficient residual double-coding Unet for medical image segmentation [J ] . IEEE Transactions on Circuits and Systems for Video Technology , 2024 , 34 ( 4 ): 2083 - 2096 .
0
浏览量
0
下载量
0
CSCD
关联资源
相关文章
相关作者
相关机构
京公网安备11010802024621