Towards Label-Free Single-Cell Phenotyping Using Multi-Task Learning

2026-05-14

AI Summary

This study addresses the challenge that label-free brightfield images lack sufficient information to directly infer cellular molecular phenotypes, thereby limiting non-invasive single-cell analysis. To overcome this, the authors propose a unified multi-task deep learning framework that integrates CNN-derived texture features with global representations from a Vision Transformer, augmented by a learnable cross-branch gating mechanism. This architecture jointly performs white blood cell classification and continuous protein expression regression (e.g., CD16). Furthermore, a large language model is leveraged to generate biologically interpretable summaries of cellular states. Evaluated on the BSCCM dataset, the method achieves 91.3% accuracy in white blood cell classification and a Pearson correlation coefficient of 0.72 for CD16 expression regression, demonstrating the feasibility and potential of label-free single-cell phenotyping using differential phase contrast (DPC) imaging.

Abstract

Label-free single-cell imaging offers a scalable, non-invasive alternative to fluorescence-based cytometry, yet inferring molecular phenotypes directly from bright-field morphology remains challenging. We present a unified Deep Learning (DL) framework that jointly performs White Blood Cell (WBC) classification and continuous protein-expression regression from label-free Differential Phase Contrast (DPC) images. Our model employs a Hybrid architecture that fuses convolutional fine-grained texture features with transformer-based global representations through a learnable cross-branch gating module, enabling robust morpho-molecular inference from DPC images. To support downstream interpretability, we further incorporate a Large Language Model (LLM) that generates concise, biologically grounded summaries of the predicted cell states. Experiments on the Berkeley Single Cell Computational Microscopy (BSCCM) and Blood Cells Image benchmarks demonstrate strong performance, achieving a 91.3% WBC classification accuracy and a 0.72 Pearson correlation for CD16 expression regression on BSCCM. These results underscore the promise of label-free single-cell imaging for cost-effective hematological profiling, enabling simultaneous phenotype identification and quantitative biomarker estimation without fluorescent staining. The source code is available at https://github.com/saqibnaziir/Single-Cell-Phenotyping.

Problem

• label-free imaging
• single-cell phenotyping
• morpho-molecular inference
• protein expression regression
• white blood cell classification

Innovation

• label-free single-cell imaging
• multi-task learning
• hybrid CNN-Transformer architecture
• morpho-molecular inference
• large language model interpretability