摘要
针对甲状旁腺超声图像灰度分布不均匀、甲状旁腺病灶多样化的特点,利用图像全局和局部信息,采用基于图像局部熵的混合水平集模型进行甲状旁腺超声图像分割。针对不同超声图像灰度分布差异大的难题,利用图像局部熵确定全局项权重,提高模型的自适应能力。为避免局部项区域尺度设定大易出现过分割,区域尺度设定小计算效率低的问题,利用两尺度进行曲线演化。实验结果表明,本文提出的混合水平集模型对差异性大的甲状旁腺超声图像具有较强的自适应能力,能使演化曲线自动收敛于目标轮廓,具有更高的分割准确率和计算效率。
Aiming at the characteristics of the intensity inhomogeneous and diversiform parathyroid lesions in the ultrasound images of the parathyroid gland, we propose a hybrid level set model for parathyroid gland segmentation based on local entropy of images. The proposed model uses both global and local image information. To address the problem of the inhomogeneous intensity distribution in ultrasound images,local entropy of images is used to determine the weight of the global term to improve the model’s adaptivity. In addition, two scales are adopted to prevent over-segmentation and calculation inefficiency on the large and small scales, respectively. Experimental results show that the proposed model can adapt to different ultrasound images of parathyroid gland, which makes the evolution curve converge to the target contour automatically. In addition, this model has high segmentation accuracy and computational efficiency.
作者
毛林
赵利强
于明安
魏莹
王颖
Mao Lin;Zhao Liqiang;Yu Ming’an;Wei Ying;Wang Ying(College of Information Science and Technology,Beijing University of Chemical Technology,Beijing 100029,China;Interventional Ultrasourid Medicine,China-Japan Friendship Hospital,Beijing 100029,China;Institute of Microelectronics,Chinese Academy of Sciences,Beijing 100029,China;University of Chinese Academy of Sciences,Beijing100049,China)
出处
《光学学报》
EI
CAS
CSCD
北大核心
2019年第12期248-256,共9页
Acta Optica Sinica
基金
国家自然科学基金(61340056)
北京化工大学-中日友好医院生物医学转化工程联合基金(PYBZ1804)
关键词
医用光学
甲状旁腺超声图像
水平集
图像局部熵
自适应
两尺度
medical optics
ultrasound images of parathyroid gland
level set
image local entropy
adaptation
two scales
