摘要
目的评价DR影像多频域后处理方式对肺内结节检出的影响。方法选取经胸部CT证实的50例有肺内结节和50例正常者的DR正位胸片,采用多频域后处理得到标准、高通过及低通过3组图像,由低、中、高年资医师各2名进行评价,评价结果用受试着工作特征(ROC)曲线进行统计分析。结果标准、高通过及低通过3组图像ROC曲线下面积分别为0.739±0.019、0.697±0.020、0.789±0.017(P<0.05),三组中,低通过组结节检出率最高,高通过组结节检出率最低、假阴性率最高,标准组假阳性率最高。结论DR影像多频域后处理对肺结节的检出有影响,低频增强处理可以提高肺结节的检出率,高频增强处理漏诊率增加。
Objective To evaluate the effect of multiscale processing in digital chest radiography on pulmonary nodules detection.Methods Fifty normal cases and fifty small nodule patient cases confirmed by chest CT were selected.Each case had been taken a chest radiograph obtained with a digital radiography system,and the acquired image was processed by three types of multiscale processing with different structure preference values.Three groups of different experienced radiologists independently reviewed and recorded the results of pulmonary nodules detection.All observations were evaluated by the receiver operating characteristic（ROC）analysis.Results The mean area under the ROC curve（Az）values of default images,high-pass images and low-pass images were 0.739±0.019,0.697±0.020,0.789±0.017（P〈0.05）,respectively.Among the three kinds of different processing images,the Az value of low-pass was highest.The high-pass images had the lowest Az value and highest false negative rate,and the false positive rate of default images was highest.Conclusion Multiscale processing in digital chest radiography affects the pulmonary nodules detection.The low-pass processing increased the relevance rate of the pulmonary nodules detection,while the high-pass processing increased the omission rate.
出处
《中国医学影像技术》
CSCD
北大核心
2007年第11期1727-1729,共3页
Chinese Journal of Medical Imaging Technology
关键词
数字化放射影像
图像后处理
多频域后处理
肺结节
ROC曲线
Digital radiography
Image processing
Multiscale processing
Pulmonary nodule
Receiver operating characteristic curve
