Saved in:
Bibliographic Details
Main Authors: Han, Jiashu, Liu, Kunzan, Kim, Yeojin, Sinha, Saurabh, You, Sixian
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.13401
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914392515280896
author Han, Jiashu
Liu, Kunzan
Kim, Yeojin
Sinha, Saurabh
You, Sixian
author_facet Han, Jiashu
Liu, Kunzan
Kim, Yeojin
Sinha, Saurabh
You, Sixian
contents Bridging microscopy and omics would allow us to read molecular states from images-at single-cell resolution and tissue scale-without the cost and throughput limits of omics technologies. Self-supervised pretraining offers a scalable approach with minimal labels, yet how to encode single-cell identity within tissue environments-and the extent of biological information such models can capture-remains an open question. Here, we introduce MAD (microenvironment-aware distillation), a pretraining strategy that learns cell-centric embeddings by jointly self-distilling the morphology view and the microenvironment view of the same indexed cell into a unified embedding space. Across diverse tissues and imaging modalities, MAD achieves state-of-the-art prediction performance on downstream tasks including cell subtyping, transcriptomic prediction, and bioinformatic inference. MAD even outperforms foundation models with a similar number of model parameters that have been trained on substantially larger datasets. These results demonstrate that MAD's dual-view joint self-distillation effectively captures the complexity and diversity of cells within tissues. Together, this establishes MAD as a general tool for representation learning in microscopy, enabling virtual spatial omics and biological insights from vast microscopy datasets.
format Preprint
id arxiv_https___arxiv_org_abs_2603_13401
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle MAD: Microenvironment-Aware Distillation -- A Pretraining Strategy for Virtual Spatial Omics from Microscopy
Han, Jiashu
Liu, Kunzan
Kim, Yeojin
Sinha, Saurabh
You, Sixian
Computer Vision and Pattern Recognition
Artificial Intelligence
Optics
Bridging microscopy and omics would allow us to read molecular states from images-at single-cell resolution and tissue scale-without the cost and throughput limits of omics technologies. Self-supervised pretraining offers a scalable approach with minimal labels, yet how to encode single-cell identity within tissue environments-and the extent of biological information such models can capture-remains an open question. Here, we introduce MAD (microenvironment-aware distillation), a pretraining strategy that learns cell-centric embeddings by jointly self-distilling the morphology view and the microenvironment view of the same indexed cell into a unified embedding space. Across diverse tissues and imaging modalities, MAD achieves state-of-the-art prediction performance on downstream tasks including cell subtyping, transcriptomic prediction, and bioinformatic inference. MAD even outperforms foundation models with a similar number of model parameters that have been trained on substantially larger datasets. These results demonstrate that MAD's dual-view joint self-distillation effectively captures the complexity and diversity of cells within tissues. Together, this establishes MAD as a general tool for representation learning in microscopy, enabling virtual spatial omics and biological insights from vast microscopy datasets.
title MAD: Microenvironment-Aware Distillation -- A Pretraining Strategy for Virtual Spatial Omics from Microscopy
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Optics
url https://arxiv.org/abs/2603.13401