Bronson R T, Bishop Y, Hedley-Whyte E T
J Comp Neurol. 1978 Mar 1;178(1):177-86. doi: 10.1002/cne.901780110.
Several aspects of data collection and analyses of peripheral nerve experiments employing light and electron microscopic morphometric techniques have not been adequately discussed in the literature. From statistical tests performed on nerve data, it was found that light compared with electron microscopic morphometry underestimates the number of small fibers. An optimum sampling strategy must take into account a potential bias toward small fibers introduced by measuring fibers from electronmicrographs. It must also take into account a potential bias introduced by the non-random distribution of nerve fibers of different sizes in nerves. These biases are offset by sampling a large enough number of fibers from large enough area electron micrographs. A method is presented for analysing periopheral nerve data using the nested analysis of variance. This requires first dividing the usual bimodal nerve fiber distribution into component normally distributed parts. The number of fibers in the two portions of a bimodal distribution must be considered in data analysis. Knowledge of the variances of parameters to be studied in any particular nerve is necessary for optimum sampling strategies.
在利用光学和电子显微镜形态计量技术进行的周围神经实验中，数据收集和分析的几个方面在文献中尚未得到充分讨论。通过对神经数据进行统计测试发现，与电子显微镜形态计量相比，光学显微镜形态计量会低估小纤维的数量。最佳抽样策略必须考虑到从电子显微照片测量纤维时对小纤维引入的潜在偏差。它还必须考虑到不同大小的神经纤维在神经中分布的非随机性所引入的潜在偏差。通过从足够大区域的电子显微照片中抽取足够数量的纤维，可以抵消这些偏差。本文提出了一种使用方差嵌套分析来分析周围神经数据的方法。这首先需要将通常的双峰神经纤维分布划分为成分正态分布的部分。在数据分析中必须考虑双峰分布两部分中的纤维数量。对于最佳抽样策略而言，了解任何特定神经中待研究参数的方差是必要的。
