Computer-assisted image-averaging strategies for the topographic analysis of in situ hybridization autoradiographs

Abstract

We report the application of a computer-based image-averaging strategy to the quantitative topographic analysis of in situ hybridization autoradiographs, based upon a disparity-analysis algorithm. We illustrate this approach for a representative antisense riboprobe — the astrocytic marker, glial fibrillary acid protein (GFAP) — in the setting of fluid-percussion brain injury in rats. Sequential coronal autoradiographs in individual animals are first digitized and aligned by disparity analysis. Next, coronal sections of all brains of a given experimental group are placed in register with one another, using a common anatomic reference level. One brain of the series serves as a template, and corresponding sections of other brains are mapped into its contour at each level. In this manner, average and standard deviation image data sets may be generated. With thresholding techniques, individual data sets can be dichotomized with respect to a chosen threshold, and frequency maps can be generated at each coronal level, displaying numbers of brains showing supra-threshold levels of mRNA at each pixel location. Pixel-by-pixel statistical comparison of data sets obtained under two different conditions (e.g., 30 min vs. 24 h following brain trauma) is then feasible. A digitized functional-anatomic brain atlas may be fitted to the images to assist analysis. Computer-based image analysis of in situ hybridization autoradiographs greatly extends the utility and applicability of this technique.