Analyzing high-resolution images of astrocytes is important in understanding diseases, such as glaucoma and retinal detachment, to which astrocytes are known to become reactive. Yet, little is known about the statistics of astrocyte patterns over the area of an entire retina.

We developed an interactive visualization system supporting the visual analysis of microscopy mosaics of entire mouse retinas, and present preliminary findings from the study of 8 full retina mosaics that were imaged in UCSB's Retinal Cell Biology Lab. We present the design and implementation of this visualization system, which required several steps of data preparation and normalization. We will briefly discuss our choices and validation approaches for parameter selection in the image normalization, segmentation, and visualization steps of our pipeline. We created tools to support the generation and visual comparison of possible network structures interconnecting the astrocyte cells. We will report our initial insights from using these tools on the 8 retinas.

Together with our biology collaborators, we identified 9 properties of astrocyte cells that held most promise to shed light on underlying principles and statistics in astrocyte distributions, and we will present the most interesting patterns that emerged from our visual analysis.