Rai G S, Haggith J W, Fenwick J D, James O
Br J Radiol. 1979 Feb;52(614):116-23. doi: 10.1259/0007-1285-52-614-116.
A prospective study to assess the clinical usefulness of computer processing of liver scans has been carried out on 203 patients. All patients have had six months follow up to confirm the diagnostic accuracy of the scan results. Four presentations have been studied using ROC analysis: (i) original gamma-camera pictures; (ii) processed images obtained using a non-stationary filter; (iii) images processed using nine-point smoothing; (iv) images obtained by linear interpolation at 25 isocount levels of display. It is concluded that pictures obtained by use of a non-stationary filter at 25 isocount levels are superior to a conventional gamma-camera pictures and displays obtained by linear interpolation only. All three are considerably superior to the images obtained by nine-point smoothing in detection of focal space-occupying lesions. Based on the results obtained, a strategy for the rational use of computer processing of gamma-camera liver scans is proposed.
一项评估肝脏扫描计算机处理临床实用性的前瞻性研究已在203例患者中开展。所有患者均已随访6个月以确认扫描结果的诊断准确性。使用受试者工作特征（ROC）分析研究了四种呈现方式：（i）原始γ相机图像；（ii）使用非平稳滤波器获得的处理后图像；（iii）使用九点平滑处理的图像；（iv）在25个等计数显示水平通过线性插值获得的图像。得出的结论是，在25个等计数水平使用非平稳滤波器获得的图像优于传统γ相机图像以及仅通过线性插值获得的显示。在检测局灶性占位性病变方面，这三种方式均明显优于通过九点平滑获得的图像。基于所获得的结果，提出了合理使用γ相机肝脏扫描计算机处理的策略。
