Automated Detection and Segmentation of Astrocytes in GFAP-labeled Micrographs using YOLOv8

Astrocytes, the most abundant subtype of glial cells, play a critical role in the central nervous system (CNS), including neural defense and metabolism, and are implicated in the pathophysiology of various psychiatric and neurological disorders. Since astrocytes manifest their function via a range of morphological alterations, there is a particular interest in quantifying such alterations accurately and efficiently. To facilitate this task, we introduce an improved deep learning framework for automated detection and segmentation of GFAP-labeled images of astrocytes based on YOLOv8, a recent version of the popular YOLO object detection deep learning platform. We conducted extensive numerical experiments showing that our approach yields state-of-the-art detection and segmentation performance over a wide range of images of astrocytes. We also show that the proposed model can be easily adapted to deal with images of other glial cells, such as microglia.