Willemse F, Nap M, Henzen-Logmans S C, Eggink H F
Department of Pathology, University Hospital Groningen, The Netherlands.
Anal Quant Cytol Histol. 1994 Oct;16(5):357-64.
Most image analysis systems (IAS) use black-and-white cameras. However, true color IASs are considered to be useful for quantification of immunohistologically stained structures. Using a true color IAS, we evaluated two methods of segmentation for quantification of area percentage of staining: one using fixed, preset thresholds and one using thresholds interactively set per image. Furthermore, the effect of shading correction was evaluated, and measurements in both color and black-and-white mode were compared. The results of segmentation with fixed thresholds did not differ significantly from those of control percentages, established by interactive morphometry using a grid, which served as reference. Interactive segmentation was significantly different from the reference (t test, P = .0001). The effect of shading correction was negligible. Measurements with and without this procedure correlated highly (r = .99, P < .001). Comparison of the results obtained in color and black-and-white mode showed a significant difference in the latter from the reference (t test, P = .005). We conclude that it is possible to quantify, in a reliable way, area percentage of positive staining using a true color IAS with application of a segmentation method with fixed thresholds.
大多数图像分析系统(IAS)使用黑白相机。然而,真彩色IAS被认为对免疫组织化学染色结构的定量分析很有用。我们使用真彩色IAS评估了两种分割方法以定量染色面积百分比:一种使用固定的预设阈值,另一种使用针对每个图像交互式设置的阈值。此外,还评估了阴影校正的效果,并比较了彩色和黑白模式下的测量结果。使用固定阈值的分割结果与通过使用网格的交互式形态测量法确定的对照百分比没有显著差异,该网格用作参考。交互式分割与参考有显著差异(t检验,P = 0.0001)。阴影校正的效果可忽略不计。有无此程序的测量结果高度相关(r = 0.99,P < 0.001)。彩色和黑白模式下获得的结果比较表明,后者与参考有显著差异(t检验,P = 0.005)。我们得出结论,使用真彩色IAS并应用固定阈值的分割方法可以可靠地定量阳性染色的面积百分比。