A large-scale expert-annotated 3D image dataset of dermal tissue macrophages

Tissue macrophages are highly versatile immune cells whose morphologies shift dramatically depending on their activation state and tissue environment. This morphological plasticity reflects their diverse origins and functions but also poses significant challenges for both manual and automated image analysis. In the skin, macrophages display particularly intricate shapes, ranging from rounded to highly elongated, irregular, and non-convex forms. We developed a deep learning–based approach for the automated identification, 3D segmentation, and quantitative morphological profiling of dermal tissue macrophages, facilitating large-scale analysis of cellular volume and elongation. This allowed us to build a dataset of 3D-segmented dermal tissue macrophages, comprising 947 single cells. To ensure accuracy, automatically generated segmentations were refined using a custom-designed Napari plugin. To illustrate technical validity and potential reuse, we quantified macrophage volume and elongation in adult steady-state and five-day post-Staphylococcus aureus infection samples, showing that the dataset captures biologically interpretable morphological variation.

The released dataset comprises fluorescence microscopy z-stack images from adult steady-state and adult-infected mouse skin, accompanied by expert-reviewed 3D instance segmentation masks of individual macrophages.