npj Imaging最新文献

Reliable biomedical imaging demands rigorous quality control, yet high-throughput microscopy remains prone to diverse artifacts. We present AutoQC-Bench, a software based on a reconstruction-driven diffusion model flagging abnormal images without prior knowledge, and along with a benchmark of 8000 images capturing common quality issues. The software outperforms existing methods, generalizes across modalities, and supports large-scale bioimaging studies. The software and benchmark are openly shared to advance robust microscopy quality control.

Cancer metastasis involves a complex cascade of events, where cancer cells migrate from their site of origin to secondary sites via the lymphatic and circulatory system. During this process, some cancer subclones will successfully 'seed' at distant organs to generate lethal metastases. Here, we optimised a method for tracking cancer cells in metastatic breast cancer tumours and investigated their complex interplay with the lung vasculature using lentiviral-based optical barcoding (LeGO). Given the regional heterogeneity in lung tissue microenvironments as well as lobar asymmetry, we used light sheet microscopy to perform three-dimensional (3D) imaging of wholemount lung lobes. The results revealed that polychromatic metastases occurred less frequently than monochromatic metastases and were more likely to be located nearer to blood vessels in both spontaneous (i.e. mammary fat pad injections) and experimental (i.e. tail vein injections) mouse assays of metastasis. This 3D imaging and analytic pipeline can provide unique insights about metastatic heterogeneity and dynamics, and represents a new avenue for studying therapeutic response across large volumes of lung tissue.

Chemical modification of monoclonal antibodies (mAbs) and their fragments gives rise to imaging probes and targeted therapies. Depending on the isotope used, radiolabeled mAbs enable positron emission tomography (PET) and single photon emission computed tomography (SPECT) imaging and can also be applied as cytotoxic therapies. Fluorescent mAb conjugates are used for a range of preclinical applications with clinical utility for intraoperative visualization of tumors. Antibody-drug conjugates (ADCs) enhance the therapeutic efficacy of mAbs and are the topic of extensive clinical development. In all these cases, chemical modifications can significantly affect mAb tumor targeting and clearance. Whole-body imaging techniques provide crucial insights into the in vivo consequences of these changes by directly tracking antibody conjugate distribution and clearance. This review examines in vivo imaging studies that compare "parental" and "modified" mAbs imaged under identical conditions to assess the effects of the cargo itself (e.g. fluorophore, chelator, drug), as well as the chemical conjugation methods. Additionally, we also describe studies that evaluate alternative strategies, including pretargeting, Fc modifications and pre- or co-dosing strategies that seek to tune the biodistribution of a given conjugate. Overall, we highlight the critical role of imaging in characterizing the in vivo performance of mAb conjugates, underscoring how these insights can inform both therapeutic efficacy and toxicity, and enable clinical translation.

Deuterium metabolic imaging (DMI) is a new imaging approach that provides unique, complementary information to anatomical MRI of brain tumors. Preclinical DMI studies have demonstrated excellent image contrast following intravenous infusion of deuterated choline (²H₉-Cho) at a severalfold higher dose than recommended for humans. We investigated DMI performance in rat glioblastoma models after oral administration of a ²H₉-Cho dose recommended for humans. DMI, following the three daily oral low doses, resulted in ²H₉-Cho concentrations in the tumor and tumor-to-normal-brain image contrast comparable to a single, high intravenous dose. Further, ²H and 2D ¹H-¹⁴N HSQC NMR on excised tumor tissue revealed that oral administration led to increased contributions from Cho-derived molecules that were products of tumor metabolism compared to intravenous infusion of ²H₉-Cho. These results can advance clinical translation of Cho-DMI as a noninvasive imaging tool for brain tumor characterization by demonstrating the feasibility of an oral intake approach using a clinical dose.

Cell analysis technologies play a critical role in biomedical research, enabling precise evaluation of essential parameters such as cell viability, density, and confluency. In this article, we introduce Quantella, a smartphone-based platform designed to perform comprehensive cell analysis encompassing these key metrics. Addressing limitations of conventional systems, such as high cost, hardware complexity, and limited adaptability, Quantella integrates low-cost optics, a rinsable flow cell, bluetooth-enabled hardware control, and a cloud-connected mobile application. Its adaptive image-processing pipeline employs multi-exposure fusion, thresholding, and morphological filtering for accurate, morphology-independent segmentation without requiring deep learning or user-defined parameters. System validation studies across diverse cell types showed deviations under 5% from flow cytometry. With the capacity to analyze over 10,000 cells per test, Quantella delivers high-throughput, reproducible results. Its accessible, scalable design makes it a promising tool for biomedical research, diagnostics, and education, particularly in resource-limited settings.

Image-based profiling is rapidly transforming drug discovery, offering unprecedented insights into cellular responses. However, experimental variability hinders accurate identification of mechanisms of action (MoA) and compound targets. Existing methods commonly fail to generalize to novel compounds, limiting their utility in exploring uncharted chemical space. To address this, we present a confounder-aware foundation model integrating a causal mechanism within a latent diffusion model, enabling the generation of balanced synthetic datasets for robust biological effect estimation. Trained on over 13 million Cell Painting images and 107 thousand compounds, our model learns robust cellular phenotype representations, mitigating confounder impact. We achieve state-of-the-art MoA and target prediction for both seen (0.66 and 0.65 ROC-AUC) and unseen compounds (0.65 and 0.73 ROC-AUC), significantly surpassing real and batch-corrected data. This innovative framework advances drug discovery by delivering robust biological effect estimations for novel compounds, potentially accelerating hit expansion. Our model establishes a scalable and adaptable foundation for cell imaging, holding the potential to become a cornerstone in data-driven drug discovery.

Ferroptosis emerged as a cell death modality against cancer, but there are currently no available biomarkers for imaging ferroptosis-based therapies. To address this, we evaluated phosphatidylserine exposure and perforation of lipid membranes during ferroptosis to explore potential targeting opportunities. We demonstrated that nano-sized gaps at late stage ferroptosis can serve as entry points for dyes that can bind to intracellular structures. These changes were accompanied with cellular signaling components similar to platelet activation, with phosphatidylserine exposure on the cell surface as a potential target for imaging programed cell death, including ferroptosis. We employed a novel tumor-seeking dye CJ215 that can also label apoptotic cells and showed that CJ215 accumulates in ferroptotic cells both in vitro and in vivo by binding to phosphatidylserine, which can be prevented with ferroptosis inhibition. Since phosphatidylserine exposure also occurs during apoptosis, CJ215 can monitor both apoptosis and ferroptosis-based therapies.