Tools for shared annotation of correlative 2D light microscopy and 3D X-ray histology

Abstract

Histology is the microscopic study of biological tissues, typically performed by optical microscopy of thin sections. This informs biomedical research and clinical decision-making despite its limitation to two dimensions (2D). X-ray micro-computed tomography (CT) can non-destructively image tissues in 3D but lacks the biological specificity of classical histology. Combining optical microscopy with CT for correlative imaging of a single specimen joins the specificity of 2D histology with access to 3D microstructural information offered by CT, allowing exploration of tissue anatomy at a traditionally inaccessible level.
However, integrating these different, but complementary image techniques is challenging due to differences in spatial resolution, orientation, and image contrast mechanisms of these images. 2D histology images have much higher spatial resolution than the CT image slices. Non-linear deformations are also introduced between CT and histology due to mechanical sectioning. Image registration can correct these deformations, but automated methods based on the pixel intensities is challenging due to the dissimilarities in image contrast mechanisms from the different imaging techniques.
Conventional X-ray absorption-based CT has only recently been optimised for biological specimens1, so there is no established precedent on how biological structures may appear in CT compared to standard histology. Therefore, researchers often annotate histology images of matching regions of the specimen to identify biological structures of interest in CT datasets.
Here, we present a workflow for transferring annotations made on 2D histology images to the corresponding 3D CT volume, and how these correlative images together provide a comprehensive view into tissue microstructure. This workflow covers semi-automated registration to match orientations of the images, transferring the histology annotations to the registered images, and generating visualisations for intuitive interpretation of the results (Figure 1). An example application of the workflow to identify growth patterns of lung adenocarcinoma tumours will be shown.

More information

Identifiers

ORCID for Matthew Lawson: orcid.org/0000-0003-0115-1698

ORCID for Elena Konstantinopoulou: orcid.org/0000-0003-4077-9648

ORCID for Orestis L. Katsamenis: orcid.org/0000-0003-4367-4147

ORCID for Peter Lackie: orcid.org/0000-0001-7138-3764

ORCID for Philipp Schneider: orcid.org/0000-0001-7499-3576

Catalogue record

Export record