Some pulmonary carcinomas display a solid pattern, and immunohistochemistry is commonly used for tumor differentiation. Micro–computed tomography (micro-CT), with its ability to produce detailed three-dimensional images using small voxel sizes, may offer additional insights. This study investigates whether three solid tumor types, solid adenocarcinoma (sAC), non-keratinizing squamous cell carcinoma, and carcinoid tumor (CaT), can be differentiated using micro-CT. Methods: Fifteen paraffin blocks, five for each type, were scanned with micro-CT (Skyscan 1275, Bruker). These images were compared to whole slide images (WSIs) of the same tumors. Consequently, tumoral (n = 74) and non-tumoral (n = 49) regions of interest (tumor ROIs [tROIs] and non-tumor ROIs [ntROIs]) were selected on the micro-CT images and evaluated in terms of certain structural variables (percent object volume, structure model index, structure thickness, structure linear density, connectivity, connectivity density, open porosity, closed porosity) to investigate whether tumors can be differentiated from normal parenchyma and from each other. Results: Although detailed images comparable to WSIs could not be obtained, it was considered an important advantage to be able to examine the entire depth of the paraffin blocks. tROIs and ntROIs could be distinguished based on all variables (p < .001). Additionally, sAC showed a notable difference from CaT in “percent object volume” (p = .011). Conclusions: With ongoing technological advancements, improving image quality without compromising tissue integrity will likely accelerate the adoption of micro-CT in pathology labs. Moreover, structural variables derived from micro-CT images may support differentiation among tumor types.

Digital and computational pathology are expanding rapidly worldwide, driven by advances in whole-slide imaging, AI algorithms, multimodal data integration, and improved digital infrastructure. Adoption continues to accelerate in the United States and internationally, supported by professional guidelines, emerging reimbursement pathways, and the growing need for remote workflows and collaborative diagnostics. Progress in interoperability standards, regulatory frameworks, and FDA approvals has strengthened the foundation for clinical deployment, while large-scale data repositories and federated learning approaches enable more robust and privacy-preserving model development. Foundation models, multimodal AI systems, and LLM-based copilots are reshaping diagnostic support, prognostication, workflow efficiency, clinical trials and drug discovery.