冠状动脉CT血管造影图像标注方法的研究  

Research on images annotation of coronary CT angiography

作　　者：尹燕伟 朱敏 董祥梅 胡嘉航 李永生[3] Yanwei Yin;Min Zhu;Xiangmei Dong;Jiahang Hu;Yongsheng Li(Department of Imaging,Hongqi Hospital Affiliated to Mudanjiang Medical University,Mudanjiang 157011,China;Department of Pathology,Hongqi Hospital Affiliated to Mudanjiang Medical University,Mudanjiang 157011,China;School of Medical Imaging,Mudanjiang Medical University,Mudanjiang 157011,China)

机构地区：[1]牡丹江医学院附属红旗医院影像科,黑龙江省157011 [2]牡丹江医学院附属红旗医院院病理科,黑龙江省157011 [3]牡丹江医学院医学影像学院,黑龙江省157011

出　　处：《中华消化病与影像杂志（电子版）》2022年第5期281-285,共5页Chinese Journal of Digestion and Medical Imageology(Electronic Edition)

基　　金：牡丹江市应用技术研究与开发计划项目(HT2020JG054);黑龙江省教育厅基本科研业务费项目(2021kyywf0508)

摘　　要：医学图像的分割和标注问题是目前医学图像处理方向研究的难点和热点。目前对于冠状动脉CT医学图像人工标注的标准尚未有统一的定义和一致的标注方法。为此根据医学影像中的器官位置探讨冠状动脉CT血管造影的图像标注方法,本文讨论了冠状动脉CT血管造影图像标注格式、原则和策略。结合分段标注原则在标注时将整个冠状动脉CT血管造影图像细化为十二段。结合具体实例阐述了右冠状动脉支分段和右后降支分段图像的具体标注方法,即选择标注、如何标注、标注形式等。这种针对医学影像器官位置的CT血管造影分段标注方法可以为医学图像的自动标注构建高效统一的训练数据集,提高以监督学习或者半监督学习方式为基础的自动图像标注的效率。The problem of segmentation and annotation of medical images is the difficulty and hotspot of current medical image processing research.At present,there is no unified definition and consistent annotation method for manual annotation of coronary CT medical images.Therefore,the image annotation method of coronary CT angiography is discussed according to the position of organs in medical images.This paper discusses the annotation format,principles and strategies of coronary CT angiography.According to the segmented image annotation principle,the whole coronary CT angiography image is subdivided into twelve segments.Combined with specific examples,the specific annotation methods of segmented images of right coronary artery branch and right posterior descending branch are expounded,that is,selective annotation,how to annotate and annotation form,etc.This segmented annotation method of CT angiography for organ positions in medical images can build an efficient and unified training data set for automatic annotation of medical images,and improve the efficiency of automatic image annotation based on supervised learning or semi-supervised learning.

关 键 词：冠状动脉 CT血管造影 CT图像标注 医学图像处理 

分 类 号：R816.2[医药卫生—放射医学]