计算机X线摄影的观察者操作特性曲线测试探讨

Receiver operating characteristic curve evaluation on computed radiography: an experimental study

目的 通过选择计算机X线摄影 (CR)后处理参数 ,探测从输出影像上获得大信息量。方法 将直径 2mm的聚丙烯实球 10 0个放在成像板 (IP)上用X线曝光 ,再选择不同的旋转量 (GA)、灰阶 (层次 )曲线 (GT)、旋转中心 (GC)、密度变换 (GS)、频率等级 (RN)、频率种类 (RT)和频率增强 (RE)值组成 6组参数 ,在 1次X线曝光的条件下 ,获得 6张实验照片 ,由 4位观察者在照度为 6 0 0 0lx观片灯上 ,用 5值判断法取得数据 ,绘成ROC曲线。结果 采用后处理参数GA =1 0、GT =A、GC =1 6、GS=0 3、RN =4 0、RT =R、RE =3 0时 ,在低对比实验照片上获得ROC曲线 ,曲线面积平均值 (Az ) =0 96 ,在 6组后处理参数中最大 ,即获得最大信息量 ;其他参数不变 ,仅使有关参数变化为 :GA =0 8、GS =- 0 2、RE =0 5 ,在低对比实验照片上获得ROC曲线面积平均值Az =0 78,在 6组后处理参数中最小 ,即获得信息量最小。结论 只有恰当选择CR后处理参数 ,才能从输出影像上获得最大的信息量。

Objective To get the maximum information from computed radiography (CR) output images by changing post-processing parameters. Methods Six experimental photos of polypropylene balls of 2.0 mm in diameter were taken by means of one time X-ray exposing on an imaging plate (IP) with different post-processing parameters including rotation amount (GA), gray gradation type (GT), rotation center (GC), shifting amount (GS), frequency rank (RN), frequency type (RT), and degree of enhancement (RE). 6 photos were viewed by three students and one radiologist on a 6 000 lx illuminance viewbox. Receiver operating characteristic (ROC) curves were made by means of 5-value-differentiation method. Results The largest mean area value (Az) below ROC curves of a low contrast experimental photo with post-processing parameters GA=1.0,GT=A, GC=1.6, GS=0.3, RN=4.0, RT=R and RE=3.0 was 0.96, and the maximum information was obtained. The smallest mean area value (Az) was 0.78 with changed post-processing parameters GA=0.8, GS=-0.2 and RE=0.5 while other parameters were not changed. The minimum information was obtained from this photo. Conclusion In order to get the maximum information from a CR output image, the post-processing parameters should be suitably selected.