On sampling bones for microcracks

doi:10.1016/j.bone.2006.11.014

2007-04-01

PRE

R. Bruce Martin *

Oscar C. Yeh

David P. Fyhrie

DOI：10.1016/j.bone.2006.11.014

原文链接

原文求助

摘要

En 中文

This paper addresses the problem of designing experiments to measure microcrack density in cortical bone. Microcracks are relatively scarce in bone cross-sections, and their size requires microscope settings having small fields of view. Thus, substantial time is required to count cracks in each cross-section. Consequently, most studies evaluate a relatively small cross-sectional area from each specimen, the chance of finding a crack in any given field is small, and there is a significant chance of not finding even one crack in the specimens representing a particular subject. Therefore, a statistical model for microcrack counting was created to develop guidelines for sampling bones for microcracks. Three questions were addressed. 1) What are the relationships of sample size to variability in microcrack density results and the probability of crackless specimens? 2) How can sample size be chosen a priori so as to reduce the probability of crackless specimens and the associated variability in the data to an acceptable level? 3) What are the confidence intervals for the mean density of microcracks measured using microscopic counting? Using a Poisson model for the distribution of microcracks within microscope fields the total area (mm(2)) that should be examined for each specimen is given by A(s) = -In(F)/Cr.Dn, where Cr.Dn is the expected microcrack density for an individual sample and F is the desired probability (expressed as a fraction) that the individual sample will contain no microcracks. This equation is validated against 8 results from three different experiments. (c) 2006 Elsevier Inc. All rights reserved.

Keyword:

microcrack

microdamage

experimental design

Poisson distribution