Gil J, Marchevsky A M, Silage D A
Lab Invest. 1986 Feb;54(2):222-7.
We have studied the use in anatomic pathology of a computerized interactive morphometry system and conditions for reliability and proper interpretation of data obtained by tracings. Because of the ease of operation and direct visualization of the tracings, we prefer touch-sensitive screens to digitizing tablets as optimal interactive peripherals. Because touch screens have fewer sampling points (256 X 256) than tracing tablets, we studied the relationship between point density and accuracy of length and area measurements by tracing models of known dimensions on the touch screen by experimenting with different step sizes ranging from every second to every sixth valid sampling point; we obtained excellent results with step size 6 followed by two-fold smoothing by two-dimensional filtering. We conclude that because of staircase formation, a high point density may be more of a liability than an asset and that the touch screen is satisfactory. By tracing closed figures of variable shapes but known areas we observed that area measurements may be unreliable with certain irregular shapes and we discuss how perimeter to area ratios can be used to compute surface to volume ratios. Finally, the generation of simple calibrated graphic standards over routine pathologic slides makes the system useful for fast diagnostic applications.
我们研究了计算机交互式形态测量系统在解剖病理学中的应用,以及通过描记获得的数据的可靠性和正确解释的条件。由于操作简便且能直接查看描记结果,我们更倾向于将触摸屏作为最佳的交互式外围设备,而非数字化平板电脑。由于触摸屏的采样点(256×256)比描记板少,我们通过在触摸屏上对已知尺寸的模型进行描记,以不同步长(从每隔一个有效采样点到每隔五个有效采样点)进行实验,研究了点密度与长度和面积测量准确性之间的关系;我们发现步长为6并通过二维滤波进行两倍平滑处理时能获得极佳结果。我们得出结论,由于阶梯状形成,高的点密度可能更多是一种负担而非优势,且触摸屏是令人满意的。通过对形状可变但面积已知的封闭图形进行描记,我们观察到对于某些不规则形状,面积测量可能不可靠,并讨论了如何利用周长与面积之比来计算表面积与体积之比。最后,在常规病理切片上生成简单的校准图形标准使得该系统可用于快速诊断应用。