Bootstrap methods for statistical inference from stereological estimates of volume fraction

Block bootstrap methods for the estimation of the intensity of a spatial point process with confidence bounds

This paper deals with the estimation of the intensity of a planar point process on the basis of a single point pattern, observed in a rectangular window. If the model assumptions of stationarity and isotropy hold, the method of block bootstrapping can be used to estimate the intensity of the process with confidence bounds. The results of two variants of block bootstrapping are compared with a parametric approximation based on the assumption of a Gaussian distribution of the numbers of points in deterministic subwindows of the original pattern. The studies were performed on patterns obtained by simulation of well-known point process models (Poisson process, two Matérn cluster processes, Matérn hardcore process, Strauss hardcore process). They were also performed on real histopathological data (point patterns of capillary profiles of 12 cases of prostatic cancer). The methods are presented as worked examples on two cases, where we illustrate their use as a check on stationarity (homogeneity) of a point process with respect to different fields of vision. The paper concludes with various methodological discussions and suggests possible extensions of the block bootstrap approach to other fields of spatial statistics.

A brief introduction to computer-intensive methods, with a view towards applications in spatial statistics and stereology

Computer-intensive methods may be defined as data analytical procedures involving a huge number of highly repetitive computations. We mention resampling methods with replacement (bootstrap methods), resampling methods without replacement (randomization tests) and simulation methods. The resampling methods are based on simple and robust principles and are largely free from distributional assumptions. Bootstrap methods may be used to compute confidence intervals for a scalar model parameter and for summary statistics from replicated planar point patterns, and for significance tests. For some simple models of planar point processes, point patterns can be simulated by elementary Monte Carlo methods. The simulation of models with more complex interaction properties usually requires more advanced computing methods. In this context, we mention simulation of Gibbs processes with Markov chain Monte Carlo methods using the Metropolis-Hastings algorithm. An alternative to simulations on the basis of a parametric model consists of stochastic reconstruction methods. The basic ideas behind the methods are briefly reviewed and illustrated by simple worked examples in order to encourage novices in the field to use computer-intensive methods.

Spatial arrangement of microglia in the mouse hippocampus: a stereological study in comparison with astrocytes

Microglia are classically considered to be immune cells in the brain, but have now been proven to be involved in neuronal activity as well. Here we stereologically analyzed the spatial arrangement of microglia in the mouse hippocampus. First, we estimated the numerical densities (NDs) of microglia identified by ionized calcium-binding adaptor molecule 1 (Iba1). Despite that microglia appeared to be evenly distributed throughout the hippocampal area, the NDs demonstrated significant dorsoventral, interregional, and interlaminar differences. Briefly, the NDs in the ventral hippocampus were significantly lower in the CA3 region than in the CA1 region and dentate gyrus, although no interregional differences were detectable in the dorsal hippocampus. Both in the CA1 and CA3 regions, the NDs were significantly higher in the stratum lacunosum-moleculare than in the remaining layers. Next, we investigated the spatial patterns of distribution of Iba1-labeled microglia and S100beta-labeled astrocytes. So far as we examined, the somato-somatic contacts were not seen among microglia or among astrocytes, whereas the close apposition between microglia and astrocytes were occasionally detected. The 3D point process analysis showed that the spatial distribution of microglia was significantly repulsive. Because the statistical territory of single microglia was larger than that estimated from process tracing, they are not likely to touch each other with their processes. Astrocytes were distributed slightly repulsively with overlapping areas. The 3D point process analysis also revealed a significant spatial attraction between microglia and astrocytes. The present findings provide a novel anatomical basis for glial research.

Application of the parametric bootstrap method to determine statistical errors in quantitative X-ray microanalysis of thin films

We applied the parametric bootstrap to the X-ray microanalysis of Si-Ge binary alloys, in order to assess the dependence of the Ge concentrations and the local film thickness, obtained by using previously described Monte Carlo methods, on the precision of the measured intensities. We show how it is possible by this method to determine the statistical errors associated with the quantitative analysis performed in sample regions of different composition and thickness, but by conducting only one measurement. We recommend the use of the bootstrap for a broad range of applications for quantitative microanalysis to estimate the precision of the final results and to compare the performances of different methods to each other. Finally, we exploited a test based on bootstrap confidence intervals to ascertain if, for given X-ray intensities, different values of the estimated composition in two points of the sample are indicative of an actual lack of homogeneity.

Statistical analysis of reduced pair correlation functions of capillaries in the prostate gland

Blood capillaries are thread-like structures that may be considered as an example of a spatial fibre process in three dimensions. At light microscopy, the capillary profiles appear as a planar point process on sections. It has recently been shown that the observed pair correlation function g(r) of the centres of the fibre profiles on two-dimensional sections may be used to estimate the reduced pair correlation function of stationary and isotropic fibre processes in three dimensions. In the present study, we explored how this approach may be extended to statistical analysis of reduced g-functions of capillaries from multiple specimens of different groups and with replicated observations. The methods were applied to normal prostatic tissue compared with prostate cancer. Confidence intervals for the mean reduced g-functions of groups were estimated for fixed r-values parametrically using the t-distribution, and by bootstrap methods. Each estimated reduced g-function was furthermore characterized in terms of its first maximum and minimum. The mean length of capillaries per unit tissue volume was significantly higher in prostate cancer tissue than in normal prostate tissue. Significant differences between the mean reduced g-functions of malignant and benign lesions could be demonstrated for two domains of r-values. In general, bootstrap-based confidence intervals were slightly wider than parametrically estimated confidence intervals. Falsely negative lower bounds of the intervals, which sometimes arose using the parametric approach, could be avoided by the bootstrap method. Testing of group mean values for significant differences by the bootstrap method yielded more conservative results than multiple t-tests. The functional value of the first maximum of the reduced g-function and a global statistical parameter of short-range ordering was significantly reduced in the carcinoma group. Prostate cancer tissue is more densely supplied with capillaries than normal prostate tissue and the three-dimensional arrangement of the vessels differs with respect to interaction at various distance ranges. In the local approach used here, bootstrap methods can be used as a robust statistical tool for the computation of confidence intervals and group comparisons of mean reduced g-functions at specific ranges of interaction.

Prediction of the variance of stereological volume estimates from systematic sections using computer-intensive methods

The Cavalieri method is an unbiased estimator of the total volume of a body from its transectional areas on systematic sections. The coefficient of error (CE) of the Cavalieri estimator was predicted by a computer-intensive method. The method is based on polynomial regression of area values on section number and simulation of systematic sectioning. The measurement function is modelled as a quadratic polynomial, with an error term superimposed. The relative influence of the trend and the error component is estimated by techniques of analysis of variance. This predictor was compared with two established short-cut estimators of the CE based on transitive theory. First, all predictors were applied to data sets from six deterministic models with analytically known CE. For these models, the CE was best predicted by the older short-cut estimator and by the computer-intensive approach, if the measurement function had finite jumps. The best prediction was provided by the newer short-cut estimator when the measurement function was continuous. The predictors were also applied to published empirical datasets. The first data set consisted of 10 series of areas of systematically sectioned rat hearts with 10-13 items, the second data set consisted of 13 series of systematically sampled transectional areas of various biological structures with 38-90 items. On the whole, similar mean values for the predicted CE were obtained with the older short-cut estimator and the computer-intensive method. These ranged in the same order of magnitude as resampling estimates of the CE from the empirical data sets, which were used as a cross-check. The mean values according to the newer short-cut CE estimator ranged distinctly lower than the resampling estimates. However, for individual data sets, it happened that the closest prediction as compared to the cross-check value could be provided by any of the three methods. This finding is discussed in terms of the statistical variability of the resampling estimate itself.