欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
为辅助医生诊断孤立性肺结节的良恶性，提出一种针对肺结节的PET/CT影像特征的PSO-SVM分类方法。该方法利用粒子群算法（PSO）对支持向量机（SVM）进行参数搜索，进而选择最合适的参数，得到合适的SVM分类模型。实验表明，利用粒子群优化算法对支持向量机模型中的参数进行优化，可以避免人为选择的随机性，在解决分类问题中有良好的表现。使用此方法得到的分类模型对良恶性肺结节进行分类，平均正确率可达到90％以上，且为医生诊断肺结节时选取的主要特征提供了理论依据。
英文摘要：
For assisting the doctors to diagnose the benign and malignant solitary pulmonary nodules, we propose a PSO-SVM classification method aiming at pulmonary nodule PET/CT imaging features. The method searches the parameters of support vector machine （SVM） by using particle swarm optimisation （PSO） and thus chooses the most appropriate parameters and gets the proper SVM classification model. Ex- periments show that to optimise the parameters in SVM with PSO, it is able to avoid the randomicity of artificial selection, and has good per- formance in solving the classification problems. Using the classification model derived by this method to classify the benign and malignant pul- monary nodules, the average accuracy rate can reach 90% and higher, and it provides a theoretical basis for the main features selected by the doctors in diagnosis of pulmonary nodules.
同期刊论文项目
同项目期刊论文
A cloud scheduling strategy based on the combination of Fuzzy Quotient Space Cluster and an Improved
Lung CT image segmentation based on combined multi-scales watershed method and region growing method
期刊信息
位置：