Abstract
Purpose :
Neovascular (NV) tufts are hallmarks of retinopathy of prematurity or diabetic retinopathy and can lead to impaired vision or blindness. A recent study showed in an oxygen induced retinopathy (OIR) mouse model that they form complex, entangled structures. Assessing their 3D morphology in a quantitative, reproducible way remains challenging. We set out to address this by (1) utilising advanced volumetric light and electron microscopy and (2) developing an automated image analysis workflow to quantify tuft morphology at the vascular and cellular scale.
Methods :
The OIR mouse model was used throughout this study. For LSFM retinas were fixed at five timepoints after hypoxia (D2-D6), labelled, and cleared. High resolution spinning disk confocal microscopy was performed with an optimised sample mounting to preserve the curvature of the tissue. Serial block-face scanning electron microscopy (SBF-SEM) was used to investigate the structure of tufts independent of specific labels. Image analysis was performed in Python.
Results :
Our multi-scale imaging approach provided a visualisation of NV tufts from the tissue to subcellular scale. In LSFM smaller tufts were observed at all timepoints, whereas later stage tufts had a larger lateral extent. To image sub-cellular details, we mounted retinal tissue free-floating and used a high NA (>1), long working distance (0.8 mm) objective. This approach preserved the curvature of the retina, which is distorted in the often performed flat-mounting. Individual filopodia formed by endothelial cells were resolved within the tissue context. SBF-SEM identified larger tufts contained a cell-free region that was topologically connected to the vitreous. Fibrillar components in the vitreous radially arranged around the tuft.
To automatically detect NV tufts our image analysis workflow is based on the classification of a multi-scale segmentation of the vascular network. We calculate quantitative features describing the tuft morphology, connectivity, and its composing cells.
Conclusions :
We present a novel image analysis platform for light microscopy data of NV tufts. Together with our volumetric imaging approaches we can both image the structure of tufts in detail and make automated, quantitative measurements. Future work will apply our analysis to other tuft-associated cell types and genetically altered OIR retinas, to address hypotheses of tuft formation and advance our understand of their formation.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.