MSCT孤立肺小结节的神经网络模型的应用研究分析

The Study of Application Analysis of SVM in SPN During MSCT Scanning

目的研究神经网络模型(SVM)在MSCT孤立肺小结节(SPN)良恶性判断中的应用。方法收集2000例经病理证实的SPN临床表现及影像学特征,采用SPSS13.0统计学软件将各观察指标进行多因素回归分析,与SPN病理相关指标作为SVM分析参数,对1500例进行训练,另对500例进行预测,对照病理结果分析。结果与病理对照,500例SVM中良性正确预测238(98.5%)例,错误预测4(1.5%)例;恶性正确预测250(97%)例,错误预测8(3%)例。结论 SVM能有效地对SPN良恶性进行较精确的预测,对医生正确诊断有很好的帮助。

Objective To study the application of SVM in differential diagnosis of benign and malignant SPN.Method Clinical and imaging characteristics of 2000 SPN cases confirmed by pathology were collected.The SPSS 13.0 system was applied to analyze the data Multiple Regression Analysis on all observed indexes.We used pathological relative index of SPN as analysis parameters of SVM.1500 cases were trained and 500 were predicted.It was analyzed comparing with the pathological results.Results Result of 500 SVM cases were correlated with pathology,of which correct prediction were 238 cases（98.5%）and false prediction 4 cases（1.5%）in benign while correct prediction were 250 cases（97%）and false prediction 8（3%）in malignant respectively.Conclusions SVM can accurately forecast benign and malignant SPN and it is very helpful to doctor＇s accurate diagnosis.