1.南京中医药大学附属南京市中医院, 江苏省南京市秦淮区大明路157号,210022
2.南京邮电大学物联网学院
fsyy00237@njucm.edu.cn
收稿:2025-10-30,
修回:2026-01-20,
纸质出版:2026-04-02
移动端阅览
张诗杰,张澳,王康等.痔病理图像自动化分割视觉大模型的构建及其在中医临床证型分析中的应用[J].中医杂志,2026,67(07):764-769.
ZHANG Shijie,ZHANG Ao,WANG Kang,et al.Construction of An Automated Segmentation Visual Foundation Model for Pathological Images of Hemorrhoids and Its Application in Traditional Chinese Medicine Clinical Syndrome Analysis[J].Journal of Traditional Chinese Medicine,2026,67(07):764-769.
张诗杰,张澳,王康等.痔病理图像自动化分割视觉大模型的构建及其在中医临床证型分析中的应用[J].中医杂志,2026,67(07):764-769. DOI: 10.13288/j.11-2166/r.2026.07.011.
ZHANG Shijie,ZHANG Ao,WANG Kang,et al.Construction of An Automated Segmentation Visual Foundation Model for Pathological Images of Hemorrhoids and Its Application in Traditional Chinese Medicine Clinical Syndrome Analysis[J].Journal of Traditional Chinese Medicine,2026,67(07):764-769. DOI: 10.13288/j.11-2166/r.2026.07.011.
提出一种融合视觉大模型与扩散模型的两阶段方法,采用分割一切模型(SAM)作为视觉大模型代表,结合SegRefiner扩散模型构建SAM-SegRefiner模型框架,用于痔核组织病理图像中水肿、炎症及血栓区域的自动化分割,为病理形态的客观量化及其中医证候学研究提供可重复的技术工具。通过多中心回顾性数据进行训练与验证,SAM-SegRefiner模型在独立测试集上平均像素准确率达95.32%,平均交并比为66.81%,显著优于U-Net、MixU-Net、SAM-Med2D等对比模型,并展现出良好的跨中心泛化能力。进一步将模型量化分割结果与患者中医证型进行关联分析,探索病理形态学特征与中医辨证分型之间的潜在联系,发现不同证型间的炎症浸润与血栓形成程度未呈现统计学显著差异,提示局部病理改变与整体证候表征之间关系复杂。
This paper proposes a two-stage method integrating visual foundation models (VFM) and diffusion models. The segment anything model (SAM) as VFM is combined with the SegRefiner diffusion model to construct the SAM-SegRefiner framework for automated segmentation of edema, inflammation, and thrombus regions in histopathological images of hemorrhoidal tissue, providing a reproducible technical tool for the objective quantification of pathological morphology and its application in traditional Chinese medicine (TCM) syndrome research. Trained and validated on multi-center retrospective data, the SAM-SegRefiner model achieved an average pixel accuracy of 95.32% and a mean intersection over union (mIoU) of 66.81% on an independent test set, significantly outperfor-ming comparative models such as U-Net, MixU-Net, and SAM-Med2D, and also demonstrating robust cross-center generalization capability. Furthermore, by correlating the quantitatively segmented results from the model with the patients' TCM syndrome types, the potential associations between pathomorphological features and TCM syndrome differentiation have been explored. The analysis revealed no statistically significant differences in the degree of inflammatory infiltration and thrombus formation among different syndrome types, suggesting a complex relationship between local pathological changes and systemic syndrome manifestations.
中国中西医结合学会大肠肛门病专业委员会 . 中国痔病诊疗指南(2020) [J]. 结直肠肛门外科 , 2020 , 26 ( 5 ): 519 - 533 .
MEN V , KAUR D , BAHL P , et al . Is routine histopathological analysis of hemorrhoidectomy specimens necessary? A systematic review and meta-analysis [J]. World J Surg , 2024 , 48 ( 8 ): 1981 - 1989 .
刘宁远 , 刘殿龙 , 郑丽华 . 痔的治疗现状与展望 [J]. 中国临床医生杂志 , 2024 , 52 ( 1 ): 1 - 3 .
卞修武 , 平轶芳 . 我国病理学科发展面临的挑战和机遇 [J]. 第三军医大学学报 , 2019 , 41 ( 19 ): 1815 - 1817 .
孙兴 , 蔡肖红 , 李明 , 等 . 视觉大模型sam在医学图像分割中的应用综述 [J]. 计算机工程与应用 , 2024 , 60 ( 17 ): 1 - 16 .
LI L , WANG C , GENG Y , et al . Segment anything model for gastric cancer [J]. Cancer Med , 2025 , 14 ( 18 ): e71246 .
KIRILLOV A , MINTUN E , RAVI N , et al . Segment anything [EB/OL]. ( 2023-04-05 ) [ 2025-12-16 ]. https://arxiv.org/abs/2304.02643 https://arxiv.org/abs/2304.02643 .
WANG M , DING H , LIEW JH , et al . SegRefiner: towards model-agnostic segmentation refinement with discrete diffusion process [EB/OL]. ( 2023-12-19 ) [ 2025-12-16 ]. https://arxiv.org/abs/2312.12425 https://arxiv.org/abs/2312.12425 .
TRIGKA M , DRITSAS E . A comprehensive survey of deep learning approaches in image processing [J]. Sensors , 2025 , 25 ( 2 ): 531 .
梁仲达 , 仇大伟 , 刘静 . 基于多尺度特征维度频域筛选的医学图像分割算法 [J/OL]. 激光与光电子学进展 .( 2026-01-22 )[ 2026-02-18 ]. https://link.cnki.net/urlid/31.1690.TN.20260121.1807.080 https://link.cnki.net/urlid/31.1690.TN.20260121.1807.080 .
卢力玮 , 汪洋 , 柳杨 , 等 . 基于多尺度通道融合注意力的皮肤癌U-Net分割模型 [J]. 信息与控制 , 2025 , 54 ( 5 ): 787 - 800 .
RONNEBERGER O , FISCHER P , BROX T . U-net: convolutional networks for biomedical image segmentation [EB/OL]. ( 2015-03-18 ) [ 2025-12-16 ]. https://arxiv.org/abs/1505.04597 https://arxiv.org/abs/1505.04597 .
XIE E , WANG W , YU Z , et al . SegFormer: simple and efficient design for semantic segmentation with transformers [EB/OL]. ( 2021-03-31 ) [ 2025-12-16 ]. https://arxiv.org/abs/2105.15203 https://arxiv.org/abs/2105.15203 .
丁维龙 , 宗泽永 , 朱红波 , 等 . 基于CNN和Transformer的组织病理图像分割方法 [J]. 浙江工业大学学报 , 2024 , 52 ( 6 ): 591 - 600 .
CHENG J , YE J , DENG Z , et al . SAM-Med 2 D[EB/OL]. ( 2023-08-30 ) [ 2025-12-16 ]. https://arxiv.org/abs/2308.16184 https://arxiv.org/abs/2308.16184 .
席作武 , 杨娅娟 . 内痔病理学本质与中医证型相关性研究 [J]. 国医论坛 , 2021 , 36 ( 5 ): 23 - 25 .
MA J , HE Y , LI F , et al . Segment anything in medical images [J]. Nat Commun , 2024 , 15 ( 1 ): 654 .
LIU H , YANG H , VAN DIEST PJ , et al . WSI-SAM: multi-resolution segment anything model (SAM) for histopathology whole-slide images [EB/OL]. ( 2023-04-18 ) [ 2025-12-16 ]. https://arxiv.org/abs/2304.09257 https://arxiv.org/abs/2304.09257 .
CHEN T , ZHU L , DING C , Et al . SAM fails to segment anything? SAM-adapter: adapting SAM in underperformed scenes: camouflage, shadow, medical image segmentation, and more[EB/OL]. ( 2023-04-18 ) [ 2025-12-16 ]. https://arxiv.org/abs/2304.09148 https://arxiv.org/abs/2304.09148 .
MAZUROWSKI MA , DONG H , GU H , et al . Segment anything model for medical image analysis: an experimental study [J]. Med Image Anal , 2023 , 89 : 102918 . doi: 10.1016/j.media.2023.102918 http://dx.doi.org/10.1016/j.media.2023.102918 .
CAZZANIGA G , DEL CARRO F , ECCHER A , et al . Improving the annotation process in computational patho-logy: a pilot study with manual and semi-automated approaches on consumer and medical grade devices [J]. J Imaging Inform Med , 2024 , 38 ( 2 ): 1112 - 1119 .
SHEN Z , SIMARD M , BRAND D , et al . A deep lear-ning framework deploying segment anything to detect pan-cancer mitotic figures from haematoxylin and eosin-stained slides [J]. Commun Biol , 2024 , 7 ( 1 ): 1674 .
WANG C , WAN Y , LI S , et al . SegAnyPath: a foundation model for multi-resolution stain-variant and multi-task pathology image segmentation [J]. IEEE Trans Med Imaging , 2025 , 44 ( 10 ): 3924 - 3937 .
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11010102007558号