Mapping Brain Morphology with Micro-CT and Topological Data Analysis: From Drosophila to General Applications

Few organisms offer a window as precise and tractable into the exploration of biological form as Drosophila melanogaster. Yet the extraordinary natural variation in its brain morphology has remained largely uncharted, restrained by limitations in imaging, throughput, and analytical frameworks. This thesis introduces a novel, high-resolution, high-throughput imaging and analysis pipeline that redefines how we measure, compare, and interpret brain structure in flies and beyond. The heart of this thesis is TopoTome, an unsupervised algorithm grounded in topological data analysis (TDA), which encodes complex 3D brain images into persistent substructures without requiring predefined landmarks or reference atlases. This natively three-dimensional and reference-free approach allows for the formal description of shape using topology rather than geometry. Indeed, TopoTome extracts local variation in brain morphology while leveraging the global structural coherence delivered by topology. This shift is essential to capture subtle and previously unresolvable variation in brain structure. We applied our method across hundreds of genetically diverse Drosophila lines, combining it with behavioural profiling to expose how structural brain variation aligns with phenotypic diversity. Our findings reveal that even subtle morphological differences carry meaningful behavioural signatures, often imperceptible through conventional geometric methods. Through multivariate and hierarchical statistical analyses, we further uncover how environmental parameters, such as developmental temperature, imprint themselves on brain architecture, cascading into divergent behavioural expression. Beyond the methodological innovations, this thesis makes a conceptual statement: the brain's structure is not a fixed template but a distributed landscape of possibilities, where variation is not noise but signal. By formalising this landscape through topology, we move closer to a biologically grounded view of the brainâ a view that respects both the individuality of organisms, the multiscale nature of the brain, and the mathematical soundness of topological data analysis. Finally, we demonstrate the pipeline's generalisability beyond Drosophila by applying it to systems ranging from vertebrate models to human biopsies. In doing so, my thesis builds a bridge between foundational neuroscience, topology, and translational biomedical research, offering a toolset as rigorous as it is versatile.