Automatically detecting antibody labelled cells within three-dimensional datasets such as whole-brain image stacks generated with light-sheet fluorescence microscopy is a complex task. In this study, we introduce DELiVR, a virtual reality (VR) trained deep learning pipeline for identifying c-Fos+ cells, which serve as markers of neuronal activity, in cleared mouse brains. By employing VR annotation, we significantly accelerated training data generation, resulting in DELiVR surpassing current state-of-the-art methods for cell segmentation. Our pipeline, which encompasses cell detection, brain atlas registration and visualization, is conveniently packaged in a single Docker container. It operates seamlessly via the user-friendly interface of the open-source software Fiji, making it accessible to users without coding expertise. Additionally, we designed a re-training option that allows researchers to train the deep learning model on custom data sets. We highlight this feature by re-training DELiVR to detect microglia somata in the brain. Applying DELiVR to examine cancer-related brain activity, we discovered a distinct activation pattern that distinguishes weight-stable cancer from cancers associated with weight loss. Overall, DELiVR represents a robust deep learning solution for analyzing whole-brain imaging data in health and disease, eliminating the need for coding skills.