摘要
目的研究两种ADC值测量方法的差异及测量个体间差异,评价两种方法对前列腺外周带癌的诊断效能差异。方法54例前列腺病例进行DWI扫描,非癌31例,外周带癌23例。采用两种方法分别测量外周带ADC值。第一种方法:大小为30mm2左右的圆形感兴趣区(ROI),放置于外周带ADC值最低处,每层图像右叶、左叶各放一个;第二种方法:沿外周带边缘手工绘制不规则形ROI,每层图像分别绘制右叶及左叶两个ROI。对ADC值诊断前列腺癌的效能进行ROC分析。在54例研究对象中随机抽取12例,由两位医师采用两种方法分别测量,以Bland-Altman分析检验不同测量者间的个体差异。结果54例患者两种方法各获得549个ROI。第二种方法所得ADC值明显大于第一种方法,两种ADC值的相关系数为0.958。两种方法的ROC曲线下面积分别为0.856和0.850,无统计学差异,但诊断临界点有较大差异,第二种方法大于第一种方法。12例患者两种方法各获得123个ROI,两位医师之间的重复性好,且第二种方法重复性稍好于第一种方法。结论两种不同的ADC值测量方法所得数值有一定差异,但对前列腺外周带癌的诊断效能影响不大,其中第一种方法操作简单而第二种方法重复性稍好。
Objective To evaluate the differences between the two methods of ADC values measurement and to investigate the diagnostic values in the prostate cancer（PCa）in peripheral zone.Methods DWI scans were performed to 54 patients among which 31 were non-cancerous and 23 had PCa.Two methods for measuring ADC values were employed.Method one：round region of interest（ROI）areas about 30 mm2 were placed at peripheral zone with lowest ADC value.Method two：the whole volumes of the unilateral peripheral zones were measured using a freehand method on the ADC map.The diagnostic efficacy of ADC value in PCa was evaluated by ROC analysis.Twelve patients were selected to test the interobserver variance by Bland-Altman analysis.Results All of 549 ROIs were obtained by two methods.The ADC values of method two were significantly higher than the ones of method one with a corelation coefficient of 0.958.The ROC areas were 0.856 and 0.850 for the two methods with no statistic difference.But the diagnostic cut point of method two is higher than method one.The interobserver reproducibility is good with method two better than method one.Conclusion The ADC values obtained by the two methods were different but had no significant impact on the diagnosis of PCa.Method one was easier while method two has better reproducibility.
出处
《中国医学影像技术》
CSCD
北大核心
2008年第8期1236-1239,共4页
Chinese Journal of Medical Imaging Technology
基金
2007年度高等学校博士学科点专项科研基金新教师项目(20070001748)
关键词
前列腺
磁共振成像
扩散加权成像
Prostate
Magnetic resonance imaging
Diffusion-weighted imaging
