In Other News… Digital Pathology

A sMASHing discovery
 

Metabolic dysfunction-associated steatohepatitis (MASH) therapies have previously been hindered by a lack of human translational models and limitations in fibrosis analysis techniques – until now (PMID: 38467661). A team of researchers across Switzerland and the USA combined digital pathology with artificial intelligence analysis to create the FibroNest image analysis platform. Applying the FibroNest to their MASH three-dimensional (3D) InSight human liver microtissue (hLiMT) model, researchers demonstrated that using measurements of fibrosis, alongside analyzing the secretion of fibrotic biomarkers and studying gene expression, opens avenues for fibrosis drug discovery.

Neglected no longer
 

Despite being a vital part of the immune system, there is no standardized method for assessing the structure and function of the thymus. In response, researchers created an integrated and orthogonal digital pathology approach that allows for morphometric analysis of the thymic epithelial cell (TEC) network. This pipeline is versatile and applicable to different conditions affecting the thymus – from acute involution to autoimmune diseases like myasthenia gravis – and opens avenues for investigating thymic function and disease in basic and translational immunology labs. 

Illuminating multidimensional pathology
 

Pathologists are starting to explore the possibilities of 3D pathology for analyzing complex 3D tissue structures and imaging thick tissue that isn’t possible with slide-based methods. Researchers at Optica Biophotonics have progressed this ideal by incorporating a swept illumination source into an open-top light-sheet microscope. Compared with previous microscope designs, this new system quadruples the field of view and doubles optical sectioning without compromising volumetric imaging speed. 

UNIfied digital pathology
 

With hopes of improving evaluation and annotation of images in computational pathology, researchers introduce UNI – a self-supervised model trained on a dataset of over 100,000 diagnostic images across 20 major tissue types. UNI introduces new capabilities, such as resolution-agnostic tissue classification and disease subtyping. The technology should allow for more efficient and accurate AI models for diverse and diagnostically challenging tasks and clinical workflows (PMID: 38504018).