In January I submitted the my master thesis in Informatics about „Tracing and Analysis of Microglia“. You can download the final version here.
Analyzing microglial morphology can reveal information about mechanisms in the brain e.g. during injury, neurodegeneration or aging. However, this is challenging as manually tracing microglia from scans is error-prone and tedious. In this thesis we introduce a new approach to automatically extract microglia from volumetric confocal microscopy scans. First we first segment somas, then we extract processes and finally we combine the obtained structures complete cells. By visual evaluation our approach provides a new level of accuracy for extracted microglia. Applying our tracing framework to scans of mouse brain tissue, we extract nearly 3000 microglia cells and analyze their morphology in classification, clustering and regression experiments. In various ways we quantify the morphology of microglia in different activation states and show a continuous morphological transition between the microglia cells close to an injury site and those further away. In this way we demonstrate that microglial morphology can be used to predict injury severity.