Cellular & Molecular Imaging

Purpose: Construct and analyse digital X-ray images in immobilised mice using an X-ray scanner or digital 3D reconstructions of immobilised mice using computed X-ray microtomography (μCT). Experimental design: 4M + 4F minimum number of animals; 14 weeks of age at test.

Multiscale 3D image datasets of mammalian placental tissue

Sequencing the human genome gave new insights into human biology and disease. However, the ultimate goal is to understand the dynamic expression of each of the approximately 20,000 protein-coding genes and the function of each protein. Uhlen et al. now present a map of protein expression across 32 human tissues. They not only measured expression at an RNA level, but also used antibody profiling to precisely localize the corresponding proteins.

The dataset is composed of 98 whole slide images (WSIs) of renal cell carcinoma, a type of kidney cancer. These WSIs are multiplex immunofluorescent (mIF) images. They have been labeled using tow distinct panels of protein markers to allow simultaneous visualisation of multiple targets withing the same tissue sample. Panel 1 includes markers TIM-3, PD1, PD-L1, and CD8 with a DNA counterstain (Hoechst). These markers are often associated with immune response and are crucial in understanding how the immune system interacts with cancer cells. Panel 2 includes OCT4a, ZEB1, Snail, and CD44, also combined with Hoechst for DNA visualisation. These markers are particularly important because they are linked to cancer stem cells (CSC) and the epithelial-to-mesenchymal transition (EMT), processes that play significant roles in cancer progression and metastasis. This dataset will be expected to be highly valuable in the development of tools and methods aimed at characterising the differences between primary renal cancer and its metastatic forms. By analyzing these images, researchers can gain deeper insights into how renal cancer evolves and spreads, potentially leading to improved prognostic tools and personalised treatment strategies for patients.

Imaging dataset for PMID: 39314434

Photon Absorption Remote Sensing (PARS) enables label-free imaging of subcellular morphology by observing biomolecule specific absorption interactions. Coupled with deep-learning, PARS produces label-free virtual Hematoxylin and Eosin (H&E) stained images in unprocessed tissues. This set of example of PARS virtual H&E images is from a clinical study available at https://doi.org/10.48550/arXiv.2504.18737. The purpose of this study is to evaluate the diagnostic performance of these PARS-derived virtual H&E images in benign and malignant excisional skin biopsies, including Squamous (SCC), Basal (BCC) Cell Carcinoma, and normal skin. Sixteen unstained formalin-fixed paraffin-embedded skin excisions were PARS imaged, virtually H&E stained, then chemically stained and imaged at 40x. Seven fellowship trained dermatopathologists assessed all 32 images in a masked randomized fashion. A subset of 3 PARS and 3 chemical H&E (BCC, SCC, and benign) are presented here.

All images used to generate the figures of this manuscript

Spatial omics is transforming our ability to interrogate local tissue microenvironments by enabling spatially resolved measurement of biomolecules such as transcripts, proteins, and metabolites. However, capturing the full biological complexity of tissues requires combining multiple modalities, which introduces both experimental as well as computational challenges. To address computational difficulties due to differences in resolution, noise levels, and available channels, we present MIMIC—a reproducible, semi-automated workflow that integrates Mass Spectrometry Imaging (MSI) and Imaging Mass Cytometry (IMC) for joint downstream analysis. MIMIC incorporates rigorous quality control, including registration error assessment, and supports pixel-level modeling to delineate analyte–cell type associations. We demonstrate the power of our approach with a proof-of-concept study on artificial tissue and apply it to human liver tissue affected by Metabolic dysfunction-associated steatotic liver disease (MASLD). Despite integration challenges, MIMIC provides a robust framework that successfully recovers known molecular associations and reveals novel spatial relationships across modalities.

Live undifferentiated KOLF2.1J human iPSCs were labelled with 7 organelle markers targeted to the nucleus, endoplasmic reticulum (ER), Golgi, mitochondria, peroxisomes, lysosomes, and lipid droplets. Complete z-stack and timelapse images were acquired in lambda mode (410-695 nm, 8.9 nm resolution) and subsequently linear unmixed into 8 different channels (channel 1 to 7 correspond to organelles, while channel 8 is the residual of the unmixing).

Live undifferentiated KOLF2.1J human iPSCs and iNeurons, were labelled with 8 organelle markers targeted to the nucleus*, endoplasmic reticulum (ER), Golgi, mitochondria, peroxisomes, lysosomes, lipid droplets and plasma membrane. Complete z-stack and timelapse images were acquired in lambda mode (410-695 nm, 8.9 nm resolution) and subsequently linear unmixed into 9 different channels (channel 1 to 8 correspond to organelles, while channel 9 is the residual of the unmixing). Images were next deconvolved using Huygens Essential software version 23.04 (Scientific Volume Imaging, The Netherlands, http://svi.nl). Next, using Infer-subc imaging pipeline (https://github.com/SCohenLab/infer-subc/blob/v2.0.0b1), images were segmented to obtain organelle segmentations, and cell region for iPSCs, or soma/neurites regions for iNeurons. These data were used in the following preprint (https://doi.org/10.64898/2026.02.10.704675)