This work introduces a novel and specialized dataset of high-resolution fluorescence microscopic images focused on astrocytic gap junctions, aiming to get insights into intercellular communication in both healthy and pathological brain conditions. The dataset includes 20 z-stack image series -- 10 from human glioblastoma tissue and 10 from healthy rat brain tissue -- each containing between 24 and 126 optical slices. These images were acquired using standardized protocols following immunofluorescence labeling with antibodies against connexin 43 (Cx43), enabling visualization of gap junction localization at the subcellular level. The dataset is uniquely tailored to support detailed morphological and quantitative analyses of gap junction networks, featuring metadata including species, age, sex, imaging parameters, tissue characteristics, and localization of gap junctions in images provided by bounding boxes and image masks. Its structure facilitates comparative studies across physiological states and species, enhancing translational and evolutionary perspectives on astrocytic connectivity. Given the labor-intensive nature of manual gap junction quantification and 3D statistical analysis for Z-stacks, this dataset serves as a resource for the development of machine learning tools capable of automating the detection and analysis of Cx43-positive signals.