[1]石志良,范伟楠,甘梓博,等.骨边界增强滤波的图割算法[J].南京师大学报(自然科学版),2023,46(04):91-102.[doi:10.3969/j.issn.1001-4616.2023.04.013]
　Shi Zhiliang,Fan Weinan,Gan Zibo,et al.Graph-cut Algorithm for Bone Boundary Enhancement Filtering[J].Journal of Nanjing Normal University(Natural Science Edition),2023,46(04):91-102.[doi:10.3969/j.issn.1001-4616.2023.04.013]
点击复制

骨边界增强滤波的图割算法()
分享到：

《南京师大学报（自然科学版）》[ISSN:1001-4616/CN:32-1239/N]

卷:
第46卷
期数:
2023年04期
页码:
91-102
栏目:
计算机科学与技术
出版日期:
2023-12-15

文章信息/Info

Title:
Graph-cut Algorithm for Bone Boundary Enhancement Filtering
文章编号:
1001-4616(2023)04-0091-12
作者:
石志良1; 范伟楠1; 甘梓博1; 袁 琼2
(1.武汉理工大学机电工程学院,湖北 武汉 430070)
(2.湖北第二师范学院计算机学院,湖北 武汉 430205)
Author(s):
Shi Zhiliang1; Fan Weinan1; Gan Zibo1; Yuan Qiong2
(1.School of Mechanical and Electrical Engineering,Wuhan University of Technology,Wuhan 430070,China)
(2.College of Computer,Hubei University of Education,Wuhan 430205,China)
关键词:
骨组织; 自动分割; 图割; 骨边界增强滤波; Hessian矩阵
Keywords:
bone tissue;  automatic segmentation;  graph-cuts;  bone boundary enhancement filtering;  Hessian matrix
分类号:
TP391.41
DOI:
10.3969/j.issn.1001-4616.2023.04.013
文献标志码:
A
摘要:
精准的骨组织分割是辅助骨科疾病诊断和手术规划的必要步骤. 由于骨边界模糊、对比度低、骨关节处间隔狭窄,骨组织自动分割一直是医学图像分割中的难题. 传统的图割算法仅基于图像强度特征,在分割骨组织时不但需要人工设置种子点,且存在关节处骨边界分割不清晰等问题. 为此,结合骨边界增强滤波、形态学理论和图割算法,提出一种新的图像分割算法. 以图割算法为框架,基于Hessian矩阵设计增强骨边界的滤波器,将滤波结果转换为约束项加入到图割的能量函数中,求解图割模型获取初步分割结果; 基于形态学理论提出对初步结果图进行先腐蚀后简易图割的后处理方法,利用腐蚀图替代人工输入初始化图割模型,实现相邻骨组织的自动分离. 使用内部股骨数据集、癌症影像档案馆(TCIA)公开数据集进行算法验证. 实验结果表明,相比阈值分割、区域生长、传统图割,该算法在Dice、精确度、F1分数等指标上均取得了更好的结果,可作为临床诊断的可靠依据.
Abstract:
Accurate bone tissue segmentation is a necessary step to assist orthopedic disease diagnosis and surgical planning. Automatic segmentation of bone tissue has always been a difficult problem in medical image segmentation due to blurred bone boundaries,low contrast,and narrow bone joint interval. The traditional graph-cut algorithm is only based on intensity feature of the image,which not only requires manual setting seed points,but also has some problems such as unclear bone boundary at the joint. Therefore,a new image segmentation algorithm is proposed by combining bone boundary enhancement filtering,morphological theory and graph cut algorithm. Taking the graph-cut algorithm as the framework,bone boundary enhancement filtering is designed based on the Hessian matrix,and the filtering result is converted into a constraint term and added to the energy function of the graph-cut,and the graph-cut model is solved to obtain the preliminary segmentation results. The preliminary result is subjected to a post-processing method of first eroding and then simple graph-cut. The corrosion image is used to replace the manual input to initialize the graph-cut model to realize the automatic separation of adjacent bone tissue. Algorithm validation was performed using an internal femur dataset,the Cancer Image Archive(TCIA)public dataset. The experimental results show that compared with threshold segmentation,region growth,and traditional graph-cut,the algorithm achieves better results in Dice,accuracy,F1 score and other indicators,and can be used as a reliable basis for clinical diagnosis.

参考文献/References:

[1]ZHAO S W,WANG P J,HEIDARI A A,et al. Multilevel threshold image segmentation with diffusion association slime mould algorithm and Renyi's entropy for chronic obstructive pulmonary disease[J]. Computers in biology and medicine,2021,134:104427.
[2]ZHENG Z A B,ZHANG X,CHANG B,et al. Semi-automatic liver segmentation in CT images through intensity separation and region growing[J]. Procedia computer science,2018,131:220-225.
[3]CHENG Z,WANG J. Improved region growing method for image segmentation of three-phase materials[J]. Powder technology,2020,368:80-89.
[4]张桂梅,周飞飞,储珺. 一种改进的变分水平集的图像分割算法[J]. 图学学报,2015,36(5):740-746.
[5]LE Z,LTSA B,TW D,et al. A survey on regional level set image segmentation models based on the energy functional similarity measure[J]. Neurocomputing,2021,452:606-622.
[6]WU C,KANG Z. Robust entropy-based symmetric regularized picture fuzzy clustering for image segmentation[J]. Digital signal processing,2021,110:102905.
[7]康家银,张文娟. 用于图像分割的非局部空间约束的核FCM算法[J]. 南京师大学报(自然科学版),2019,42(3):122-128.
[8]朱红,何瀚志,方谦昊,等. 面向医学图像分割的蚁群密度峰值聚类[J]. 南京师大学报(自然科学版),2019,42(2):1-8.
[9]CHEN Y K,WU X M,CAI K,et al. CT Image Segmentation based on Clustering and Graph-Cuts[J]. Procedia engineering,2011,15:5179-5184.
[10]BOYKOV Y Y. Interactive graph cuts for optimal boundary & region segmentation of objects in n-d images[C]//Proc Eighth IEEE International Conference on Comput Vis. USA:IEEE Computer Society,2001,7:105-112.
[11]BOYKOV Y,FUNKA-LEA G. Graph cuts and efficient N-D image segmentation[J]. International journal of computer vision,2006,70(2):109-131.
[12]ABABNEH S Y,PRESCOTT J W,GURCAN M N. Automatic graph-cut based segmentation of bones from knee magnetic resonance images for osteoarthritis research[J]. Medical image analysis,2011,15(4):438-448.
[13]NAKAGOMI K,SHIMIZU A,KOBATAKE H,et al. Multi-shape graph cuts with neighbor prior constraints and its application to lung segmentation from a chest CT volume[J]. Medical image analysis,2013,17:62-77.
[14]胡局新,丁宾,邵晓根. 基于图切的交互式图像分割方法[J]. 南京师范大学学报(工程技术版),2014,14(4):62-65,70.
[15]王万耀,段先华,徐丹,等. 基于显著性的Grabcut图像分割方法[J]. 计算机工程,2018,44(7):230-243.
[16]HUANG Q,DING H,WANG X,et al. Fully automatic liver segmentation in CT images using modified graph cuts and feature detection[J]. Computers in biology & medicine,2018,95:198-208.
[17]JEBA J A,DEVI S N. Efficient graph cut optimization using hybrid kernel functions for segmentation of FDG uptakes in fused PET/CT images[J]. Applied soft computing,2019,85:105815.
[18]LIU Z,SONG Y Q,SHENG V S,et al. Liver CT sequence segmentation based with improved U-Net and graph cut[J]. Expert systems with application,2019,126:54-63.
[19]SIRIAPISITH T,KUSAKUNNIRAN W,HADDAWY P. Pyramid graph cut:Integrating intensity and gradient information for grayscale medical image segmentation[J]. Computers in biology and medicine,2020,126:103997.
[20]ZHEN Y,YQZA C,MIAO L,et al. Semi-automatic liver tumor segmentation with adaptive region growing and graph cuts[J]. Biomedical signal processing and control,2021,68:102670.
[21]JCGS A,MMAB C,MSA B,et al. Segmentation of bones in medical dual-energy computed tomography volumes using the 3D U-Net[J]. Physica medica,2020,69:241-247.

备注/Memo

备注/Memo:
收稿日期:2022-06-25.
基金项目:湖北省重点研发项目(2021BCA106)、国家重点研发计划项目(2018YFB1105503).
通讯作者:袁琼,博士,副教授,研究方向:图形图像处理、计算机视觉等. E-mail:77166821@qq.com
更新日期/Last Update: 2023-12-15