Researchers at the Paul Scherrer Institute (PSI) in Switzerland have successfully developed VISTACT, an artificial intelligence workflow that transforms high-resolution micro-computed tomography data into three-dimensional virtual histological stains. Led by physicist Goran Lovric and co-authored by Cristina Almagro-Pérez, the project, published in the Journal of the Royal Society Interface in 2026, overcomes the traditional two-dimensional constraints of disease diagnosis. Conventional histology requires manually slicing tissue into thin sections, applying chemical stains, and examining them under microscopes. This process is time-consuming, destructive, and confined to a single plane. While phase-contrast micro-CT captures detailed soft-tissue structures in three dimensions, it produces only grayscale images lacking the familiar color markers of chemical stains. VISTACT bridges this gap using a conditional generative adversarial network trained on paired datasets of physical histology slides and corresponding micro-CT scans. The model learns to map microscopic grayscale patterns to standard histological colors, translating volumetric data into digitally stained specimens. A key innovation is a novel multistage registration process that precisely aligns physical tissue slices with their exact coordinates within the CT volume. Standard mapping methods often fail due to tissue distortion during cutting, but VISTACT’s automated spatial correlation ensures high-fidelity image translation. During validation on lung tissue from pulmonary hypertension patients, the system accurately rendered vascular remodeling and tissue architecture in three dimensions, generating distinct color differentiation for blood vessels, collagen, and pulmonary surfaces. This breakthrough establishes a functional proof of concept for non-destructive, automated three-dimensional pathology. By eliminating physical sectioning and chemical processing, the workflow could substantially accelerate biomarker discovery and disease modeling, particularly for complex conditions like oncology or vascular pathology. The AI-driven approach also promises faster throughput and reduced sample waste. The technology remains experimental and is not yet approved for clinical diagnostics. High-resolution phase-contrast imaging currently requires large-scale facilities like PSI’s Swiss Light Source, and the resulting datasets are computationally intensive. Current resolutions also struggle to reliably resolve individual cell nuclei. Additionally, because the output represents statistical reconstruction rather than physical chemistry, clinical validation protocols are still required. Nevertheless, by merging computational imaging with classical pathology, VISTACT signals a major shift in tissue analysis, potentially redefining diagnostic research more than a century after Virchow’s foundational work.