肺癌CT图像特征的ISODATA模糊聚类分析

Fuzzy Cluster Analysis Based on CT Image Features of Lung Cancer

针对肺癌早期所表现的医学征象,根据其特征进行分类,使医生在病变初期就能够对可疑病变做出分类。使用由56名肺癌患者组成的样本集,在MATLAB编程环境下,采用ISODATA模糊聚类方法,对关键的医学征象进行分析,将患者分类,反复迭代得到理想的聚类结果。结果显示,ISODATA模糊聚类算法可将孤立性肺结节肺癌CT图像特征相似的归为一类;样本集中的全部患者可分为3类。并进一步研究了肺部CT图像特征与肺癌诊断之间的关系,辅助医生更深刻地理解孤立性肺结节肺癌CT图像的医学特征,对肺癌的早期诊断及临床治疗具有十分重要的意义。

According to the medical signs in the early stage of lung cancer,the classification is made according to its characteristics,so that the doctors can classify the suspicious lesions at the early stage.In the MATLAB programming environment,the method of ISODATA fuzzy clustering was adopted to analyze the key medical signs,56 lung cancer patients were composed into a sample set and classified,and the ideal clustering results were obtained through repeated iteration.The results show that ISODATA fuzzy clustering algorithm classifies the CT images of solitary pulmonary nodules lung cancer with similar features into one category,and the clustering results are 3 categories,which accord with the expected results.The relationship between CT features of lung and diagnosis of lung cancer was further studied,to assist doctors to better understand the medical characteristics of isolated lung nodules of lung cancer in the CT images,which is of great significance for the early diagnosis and clinical treatment of lung cancer.