3D Quantification and Description of the Developing Zebrafish Cranial Vasculature

Background: Zebrafish are an excellent model to study cardiovascular development and disease. Transgenic reporter lines and state-of-the-art microscopy allow 3D visualization of the vasculature in vivo. Previous studies relied on subjective visual interpretation of vascular topology without objective quantification. Thus, there is the need to develop analysis approaches that model and quantify the zebrafish vasculature to understand the effect of development, genetic manipulation or drug treatment. Aim: To establish an image analysis pipeline to extract quantitative 3D parameters describing the shape and topology of the zebrafish vasculature, and examine how these are impacted during development, disease, and by chemicals. Methods: Experiments were performed in zebrafish embryos, conforming with UK Home Office regulations. Image acquisition of transgenic zebrafish was performed using a Z.1 Zeiss light-sheet fluorescence microscope. Pre-processing, enhancement, registration, segmentation, and quantification methods were developed and optimised using open-source software, Fiji (Fiji 1.51p; National Institutes of Health, Bethesda, USA). Results: Motion correction was successfully applied using Scale Invariant Feature Transform (SIFT), and vascular enhancement based on vessel tubularity (Sato filter) exceeded general filter outcomes. Following evaluation and optimisation of a variety of segmentation methods, intensity-based segmentation (Otsu thresholding) was found to deliver the most reliable segmentation, allowing 3D vascular volume measurement. Following successful segmentation of the cerebral vasculature, a workflow to quantify left-right intra-sample symmetry was developed, finding no difference from 2-to-5dpf. Next, the first vascular inter-sample registration using a manual landmark-based approach was developed and it was found that conjugate direction search allowed automatic inter-sample registration. This enabled extraction of age-specific regions of similarity and variability between different individual embryos from 2-to-5dpf. A workflow was developed to quantify vascular network length, branching points, diameter, and complexity, showing reductions in zebrafish without blood flow. Also, I discovered and characterised a previously undescribed endothelial cell membrane behaviour termed kugeln. Conclusion: A workflow that successfully extracts the zebrafish vasculature and enables detailed quantification of a wide variety of vascular parameters was developed.