Sundfors C, Collan Y
Department of Pathology, University of Turku, Finland.
Cell Mol Biol (Noisy-le-grand). 1995 Jul;41(5):671-81.
The reproducibility of quantitative image analytic assessment of agarose gel separated PCR products was tested in a computerised core facility image analysis unit serving several research groups. Routine type use of the instrumentation showed considerable variation (measurements of individuals bands: CV about 20%; ratio measurements: CV about 6.5%). Careful concentration on the methodology improved the result (CV about 9% and 4.5%, respectively). Focusing into the methodology clearly showed that the baseline correction was necessary. The magnification of the image and the width of the measurement window were important variation sources. If these were kept constant, the reproducibility improved remarkably (CVs of 1.69% and 2.90%, respectively). Further analysis showed that the results did not change during the warm up period of the instrumentation. Measurements based on integrated optical density or related variables were well correlated, but did not show linear relation with peak height estimates. The most reproducible results were achieved when the thresholds were placed at the point of 45 degrees inclination of the image density curve of the band, or at the half height points of the curve.
在为多个研究小组服务的计算机化核心设施图像分析单元中，对琼脂糖凝胶分离的PCR产物进行定量图像分析评估的可重复性进行了测试。仪器的常规使用显示出相当大的差异（单个条带的测量：CV约为20%；比率测量：CV约为6.5%）。仔细关注方法学改进了结果（分别为CV约9%和4.5%）。专注于方法学清楚地表明基线校正很有必要。图像的放大倍数和测量窗口的宽度是重要的变异来源。如果将这些保持恒定，可重复性会显著提高（分别为CV 1.69%和2.90%）。进一步分析表明，在仪器的预热期间结果没有变化。基于积分光密度或相关变量的测量具有良好的相关性，但与峰高估计值没有线性关系。当阈值设置在条带图像密度曲线45度倾斜点或曲线半高处时，可获得最可重复的结果。
