Cellplex benchmarks predicting single cell proteins from he slides
CellPlex: Benchmarking Single-Cell Spatial Proteomics Prediction from H&E
Computer Vision and Pattern Recognition
Summary
Measuring many proteins in each cell of a tissue is expensive, but simple stained slides called H&E are common. The authors introduce CellPlex, a large and diverse dataset that compares methods for predicting protein levels in single cells from these common H&E images. They find that models adapted from general pathology foundation models work best across different tissues, though accuracy drops when tested on new patient groups. Which proteins can be predicted depends on the tissue type, not on the number of cells measured. CellPlex provides a shared standard for testing future methods that could bring detailed molecular info from routine slides.
What this means in practice
- •For pathology software developers: Develop and compare models that predict detailed cell protein levels from routine H&E slides using the CellPlex benchmark.
- •For clinical diagnostics teams: Use protein predictions from H&E images to supplement expensive spatial proteomics in patient tissue analysis.
Authors
Zijun Gao, Yijiang Chen, Xiangde Luo, Pheng-Ann Heng
Abstract
Spatial proteomics measures tens of proteins in every cell of a tissue section, but its cost confines it to small research cohorts, whereas hematoxylin and eosin (H\&E) slides exist for nearly every patient. Inferring single-cell spatial proteomics from H\&E would bring its analyses to routine pathology, and a growing number of virtual staining models pursue this goal. Each is evaluated on its own cohort, panel and pixel-level metric, and released models transfer poorly to new cohorts, leaving open which approach to build on and which proteins H\&E can reveal. We introduce CellPlex, a benchmark that poses the task at the level of the cell: seven cohorts with same-section H\&E and multiplexed immunofluorescence, from colorectal and breast cancer to lymph-node metastases, bone marrow and glioma, with 717 samples, 44.8 million cells and 149 proteins, processed by one pipeline into per-cell targets. Pathology foundation models (PFMs) are adapted with one cell-level protocol, specialist models are fine-tuned under a matched budget, and all are scored on the same cells. Adapted PFMs are the most accurate models on every cohort, ahead of the specialists even after fine-tuning, but keep less than half of their accuracy on cohorts they were not trained on. Which proteins can be predicted depends on the tissue, not on how many cells were measured for them. By fixing the cells, targets and splits on which predictions are judged, CellPlex sets a common standard for inferring costly molecular measurements from routine H\&E and shows how far H\&E can stand in for spatial proteomics. We will release the data, splits, evaluation code and baselines as a testbed for developing and comparing future pathology foundation models and virtual staining methods.