Automatic best reference slice selection for smooth volume reconstruction of a mouse brain from histological images.

Abstract

In this paper, we present a novel and effective method for registering histological slices of a mouse brain to reconstruct a 3-D volume. First, intensity variations in images are corrected through an intensity standardization process so that intensity values remain constant across slices. Second, the image space is transformed to a feature space where continuous variables are taken as high fidelity image features for accurate registration. Third, in order to improve the quality of the reconstructed volume, an automatic best reference slice selection algorithm is developed based on iterative assessment of image entropy and mean square error of the registration process. Fourth, a novel metric for evaluating the quality of the reconstructed volume is developed. Finally, the effect of optimal reference slice selection on the quality of registration and subsequent reconstruction is demonstrated.