When Pathology Goes 3D
Stratifying cancers with non-destructive 3D pathology
This image shows a 3D pathology dataset of a prostate biopsy stained with a fluorescent analogue of H&E (left). The researchers perform deep learning-based image translation to convert the H&E dataset into a synthetic dataset that looks like it has been immunolabeled to highlight a cytokeratin biomarker (brown) that is expressed by the epithelial cells in all prostate glands. In turn, this synthetically immunolabeled dataset allows for relatively straightforward and accurate 3D segmentation of the prostate gland epithelium (yellow) and lumen spaces (red). This also allows the researchers to extract the “skeleton” of the branching-tree gland network (magenta). Quantitative features derived from these segmented 3D structures are used to train a machine classifier to stratify between aggressive (recurrent) versus indolent (non-recurrent) cancer.
Jonathan Liu, a professor at the University of Washington and senior investigator on this project (1), remarks that “this is the first of hopefully many studies to come that will demonstrate the value of 3D pathology versus conventional 2D pathology for guiding critical treatment decisions for cancer patients. We are combining the best of modern optical technologies, tissue-clearing methods, and big data machine learning tools to transform the field of pathology that is so foundational to diagnostic medicine.”
Click here to see a video summary of their work.
- W Xie et al., Cancer Res, 82, 334 (2022). PMID: 34853071.