External marker-based automatic congruencing: A new method of 3D reconstruction from serial sections

Abstract

Background Computer‐based three‐dimensional (3D) visualizations reconstructed from sectional images represent a valuable tool in biomedical research and medical diagnosis. Particularly with those imaging techniques that provide virtual sections, such as CT, MRI, and CLSM, 3D reconstructions have become routine. Reconstructions from physical sections, such as those used in histological preparations, have not experienced an equivalent breakthrough, due to inherent shortcomings in sectional preparation that impede automated image‐processing and reconstruction. The increased use of molecular techniques in morphological research, however, generates an overwhelming amount of 3D molecular information, stored within series of physical sections. This valuable information can be fully appreciated and interpreted only through an adequate method of 3D visualization. Methods and results:In this paper we present a new method for a reliable and largely automated 3D reconstruction from physically sectioned material. The ‘EMAC‘ concept (External Marker‐based Automatic Congruencing) successfully approaches the three major obstacles to automated 3D reconstruction from serial physical sections: misalignment, distortion, and staining variation. It utilizes the objectivity of external markers for realignment of the sectional images and for geometric correction of distortion. A self‐adapting dynamic thresholding technique compensates for artifactual staining variation and automatically selects the desired object contours. Conclusions Implemented on a low‐cost hardware platform, EMAC provides a fast and efficient tool that largely facilitates the use of computer‐based 3D visualization for the analysis of complex structural, molecular, and genetic information in morphological research. Due to its conceptual versatility, EMAC can be easily adapted for a broad range of tasks, including all modern molecular‐staining techniques, such as immunohistochemistry and in situ hybridization. Anat. Rec. 248:583‐602, 1997. © 1997 Wiley‐Liss, Inc.

A three-dimensional, histological and deformable atlas of the human basal ganglia. I. Atlas construction based on immunohistochemical and MRI data

“…Second, this 3D block needed to be registered with the corresponding MRI data. As the data acquisition protocol included photographs taken during frozen brain cutting with fiducial markers (the rivets) included in the field of view, marker-based alignment of these photographs produced intermediate 3D blocks that possess the 3D shape properties of the specimen (Malandain et al, 2004;Mega et al, 1997;Schormann and Zilles, 1998;Streicher et al, 1997). Alignment was performed using rigid transforms in order to correct for small variations of the cryomicrotome stage position during photography.…”

A low temperature embedding and section registration strategy for 3D image reconstruction of the rat brain from autoradiographic sections

“…Since very little operator input is needed, the method is objective and the required processing time for the operator is only a couple of minutes. Our approach dedicated to frozen tissue samples differs from the approach proposed by Streicher et al (1997) that requires room temperature manipulations of the sample and additional chemical treatments (like dehydration). As in our case, the positions of the markers are defined by the metallic mold, they can be thinner and better aligned (parallel trajectories) than holes drilled independently in the embedding material.…”

“…However, internal markers are difficult to place inside small frozen specimens, they damage the tissue, and they are sometimes difficult to identify. We therefore developed an approach using external markers (Streicher et al, 1997). After some practice, it becomes very easy to setup the markers, and in general, they stand sectioning very well.…”

“…A basic problem in image registration is that consecutive slices are naturally dissimilar, even in the absence of data corruption. Finally, the non-contiguous sampling of images induces global deformations due to the shape of the object (socalled "banana problem") (Malandain et al, 2004;Streicher et al, 1997). These registration difficulties can be avoided by using extrinsic markers, either inside or outside the tissue sample.…”

“…Classical embedding methods, whether applied with or without external markers, are performed at room, or even at higher temperatures (Bjarkam et al, 2001;Sorensen et al, 2000;Streicher et al, 1997;Williams and Doyle, 1996). They cannot be applied to frozen tissues, in particular to autoradiographic sections that may not defrost during processing.…”