Heckermann D, Fink U, Schätzl M, Fink B, Kenn W, Miller P, Pistitsch C, Herrmann K, Reiser M
Institut für Radiologische Diagnostik, Klinikum Grosshadern, Ludwig-Maximilians-Universität München.
Rofo. 1998 Jul;169(1):38-44. doi: 10.1055/s-2007-1015047.
To compare softcopy and hardcopy reading of chest radiographs a software tool was designed for creating lesions with clearly defined size, location and contrast.
An ROC study was performed using a set of chest radiographs with 300 simulated small nodules and linear details displayed on film and on a 1 K monitor. Six observers participated in the study. ROC curves were generated on the basis of 7200 decisions.
There was no significant difference in observer performance overall (AUC = Area under curve). Monitor: 0.84 +/- 0.08 vs. AUC, Film: 0.84 +/- 0.08). Slight, but not statistical significant differences were found in the detection of low-contrast lung nodules and of linear lung details.
Soft-copy reading using a high quality 1 K CRT display provides the same accuracy for the detection of subtle computer-simulated lesions as hardcopy reading. A software tool could be used for assigning optimal characteristics of CRT- and hardcopy displays using clearly defined lesions as a gold standard.
为比较胸部X光片的软拷贝阅读和硬拷贝阅读，设计了一种软件工具来创建具有明确大小、位置和对比度的病变。
使用一组胸部X光片进行ROC研究，这些片子上有300个模拟的小结节和线性细节，分别显示在胶片和1K显示器上。六名观察者参与了该研究。基于7200个判断生成ROC曲线。
观察者的总体表现没有显著差异（AUC = 曲线下面积）。显示器：0.84±0.08 vs. AUC，胶片：0.84±0.08）。在检测低对比度肺结节和线性肺细节方面发现了轻微但无统计学意义的差异。
使用高质量1K CRT显示器进行软拷贝阅读在检测细微的计算机模拟病变方面与硬拷贝阅读具有相同的准确性。可以使用一种软件工具，以明确界定的病变作为金标准来确定CRT显示器和硬拷贝显示器的最佳特性。
Zotero 插件