毕业论文--基于FCM算法的PET医学图像分割.doc

湖南大学毕业设计(论文) 第 = 3 \* ROMAN \* MERGEFORMAT 页 HUNAN UNIVERSITY 毕业设计(论文) 设计论文题目： 基于FCM的PET医学图像分割 学生姓名： 学生学号： 专业班级： 自动化2班 学院名称： 电气与信息工程学院 指导老师： 学院院长：  2015 年 月 日 PAGE \* MERGEFORMAT I 湖南大学毕业设计(论文) PAGE \* MERGEFORMAT I 摘 要 正电子发射断层扫描（Positron Emission Computed Tomography）功能分子生物影像系统已广泛运用于恶性肿瘤放射治疗的临床研究。准确的将PET图像中的肿瘤区域分割出来具有重要的临床意义。 PET作为一种常用于头颈部癌靶区勾画的功能影像，其PET SUV是判断肿瘤良恶性最常用的半定量指标。临床实验中最简单的是手动勾画。但是由于PET图像自身的低分辨率等特点，模糊的边缘，以及不均匀性让此方法更加难以准确分割出肿瘤部分，所以分割结果也不是特别准确。 为了更准确的区分肿瘤组织和正常组织，本课题选择基于模糊集理论提出的模糊C均值(FCM)算法进行实验。模糊聚类因其具有较强的描述不良问题的能力，所以将模糊理论引入图像处理与分析领域，使基于模糊集理论的图像分割方法有更好的图像分割效果。 本文对近几年来出现的图像分割方法作了较为全面的综述，探讨了图像分割技术的发展方向，然后选择了模糊聚类中最为经典的方法—模糊C均值算法进行了研究，同时通过实验验证了方法的有效性。 关键词：PET，图像分割，FCM，模糊集 Abstract Positron emission tomography (Positron Emission Computed Tomography) imaging system has the function of molecular biology in clinical studies are widely used in cancer radiotherapy. PET images accurately dividing tumor region out of great clinical significance. PET functional imaging as a common target delineation in head and neck cancer, the PET SUV judging benign and malignant tumors most commonly used semi-quantitative indicators. With this method, the simplest is to manually outline. However, due to its low-resolution PET image characteristics such as blurred edges, as well as non-uniformity make this method more difficult to accurately segment the tumor section, so the result is not particularly accurate segmentation. In order to more accurately distinguish between tumor and normal tissue, the subject selected for the experiment Fuzzy C-Means (FCM) fuzzy set theory proposed algorithm more. Fuzzy clustering because of its strong ability to describe the problems of poor, so the fuzzy theory into the field of image processing and analysis, so that there is a better method of image segmentation image segmentation based on fuzzy set theory. In this paper, image segmentation emerged in r