Semi-automated analysis of NMDA-mediated toxicity in digitised colour images from rat hippocampus

Abstract

The evaluation of neuronal cell survival after, for example, mechanical, hypoxic or drug-mediated injury requires the analysis of a high number of histological specimens. Since this is a time-consuming occupation, we have developed a semi-automated analysis routine for the determination of the distribution of live and dead cells. After digitalization of the histological preparations, 8-bit colour bitmaps were assessed using a compiled image-analysis programme of the software package Khoros. In the current study a detailed example of the application of this image-processing approach is described for the investigation of the cell survival after intraventricular application of N-methyl-d-aspartate (NMDA). The samples were prepared as fuchsin acid/toluidine blue stained hippocampal thin slices. The calculated areas of the live and dead cells were highly correlated with manual counts of live and dead cells in the 100 samples examined in this study. Twenty-four hours following NMDA-treatment animals (n=5) were found to have significantly fewer live and more dead hippocampal cells than the saline-treated animals (n=5), using either automated or manual examination techniques. The automated technique also revealed that NMDA treatment resulted in a reduction in the density of live cell distribution.

Introduction

It is of common interest to investigate cell survival in the brain after injuries such as local or global ischemic insults (Hara et al., 1993, Rami and Krieglstein, 1993), drug neurotoxicity (Lees and Leong, 1993), or after juvenile brain tissue transplantations (Shetty and Turner, 1995). It is believed that besides other factors, a variety of neurotransmitters are released during injury which are under cellular control, and which lead to the activation of several signal cascades responsible for the induction of necrotic or apoptopic cell death (Walker et al., 1988, Nowak and Jacewicz, 1994, Sims, 1995). To investigate the dynamics of cell survival or cell death, electrophysiological techniques in vivo or in vitro are generally used (Ashton et al., 1989, Stanton and Moskal, 1991, Opitz et al., 1995). The recording of neuronal field potentials is useful for the functional characterization of neuronal circuits, but this is not necessarily related to changes in cell survival. Therefore, histological characterization of cell survival using specific live/dead cell staining is necessary for interpretation of data obtained from electrophysiological measurements. The quantitative and qualitative analysis of pathological changes in brain tissue is a time-consuming occupation, complicated by subjective interpretation of dead or live cells in histological specimens. An automated analysis would be helpful in diminishing this subjectivity in histopathological specimen evaluation. However, the practical use of an automated image-analysis procedure is still restricted to objects which are clearly distinguishable from the background by their intensities, geometry or colours. Under such conditions cell characterization programs based on parameters such as shape, size, density of staining have been described (Eccles and Klevecz, 1986, Diller and Aggarwal, 1987, Clemens and Buzy, 1991).

The specific conditions encountered in hippocampal thin slices after staining with fuchsin acid/toluidine blue pose several problems which render existing image analysis protocols inadequate for the estimation of cell survival. The use of an automated image-analysis procedure is complicated however, by features of live cells which are present in the `cell free' region. Additional complications arise from the variation of colour, and overlap between the specimens. Thus one is forced to return to manual counting of the distinct stained cells. As an alternative to such manual counting, we have developed a semi-automated analysis routine for the determination of the area of live or dead cells stained with fuchsin acid/toluidine blue. In the present study, we present the application of this novel method (which uses digitalization of histological brain specimens and digital image-processing) for determination of the area of live/dead cells in histological preparations taken from rats which were treated in vivo with of N-methyl-d-aspartate (NMDA). This novel system is precise and efficient, considerably reduces the time required for histological evaluation of cell viability and cell death, and offers information about both the number of, the area occupied by and the population density of neuronal cells.