A deep-learning framework reveals whole-body perturbations at cell level

2026-05-20 · Source: Machine learning : nature.com subject feeds · Field: Science & Research — Artificial Intelligence & Machine Learning, Life Sciences & Biology, Health & Medical Research · Depth: Expert, extended

Summary

MouseMapper is a deep-learning framework designed for comprehensive, high-resolution analysis of whole-body disease perturbations at the cellular level in mice. This suite of foundation-model-based algorithms enables multi-system analysis, automatically segmenting 31 organs and tissues while quantitatively analyzing nerves and immune cells, including fine axonal branches and immune-cell clusters. Applied to diet-induced obesity, MouseMapper revealed significant structural alterations in the infraorbital branch of the trigeminal ganglia, leading to functional sensory deficits in whisker sensing. It also identified conserved proteomic changes in the trigeminal ganglion of obese mice and humans, affecting axon remodeling and complement pathways. The framework further generated detailed three-dimensional inflammation maps, showing shifts to larger Cd68-eGFP+ immune cell clusters in various tissues of obese mice. MouseMapper exhibits robust generalizability across diverse imaging resolutions and datasets, offering a scalable method to link animal model insights to human conditions.

Key takeaway

For research scientists investigating systemic diseases, MouseMapper offers a powerful tool to overcome limitations in whole-body, cell-level analysis. You should consider integrating this AI-driven framework to precisely quantify nerve and immune cell perturbations across 31 organs and tissues, enabling the identification of novel disease mechanisms and therapeutic targets. This approach provides a comprehensive spatial and molecular context, crucial for translating animal model findings to human conditions.

Key insights

MouseMapper provides a scalable, AI-driven platform for whole-body, cell-level disease analysis, revealing systemic pathologies and their molecular underpinnings.

Principles

• Foundation models enhance segmentation robustness and generalizability across diverse datasets.
• Whole-body imaging reveals systemic disease effects beyond localized tissue changes.
• Multi-modal data integration bridges structural, functional, and molecular insights.

Method

MouseMapper uses three deep-learning modules (Nerve, Immune, Tissue) fine-tuned on VR-annotated 3D light-sheet microscopy data, leveraging a pre-trained foundation model (VesselFM) for robust whole-body segmentation and quantification.

In practice

• Utilize foundation models for robust 3D biomedical image segmentation tasks.
• Integrate whole-body imaging with spatial proteomics for systemic disease insights.
• Quantify nerve density and immune cell cluster shifts in animal disease models.

Topics

• Deep Learning
• Whole-Body Imaging
• Biomedical Image Segmentation
• Obesity Research
• Neuroinflammation
• Spatial Proteomics

Code references

• erturklab/mouseMapper

Best for: AI Scientist, Research Scientist, Computer Vision Engineer

Related on AIssential

• Toward Generalizable Whole Brain Representations with High-Resolution Light-Sheet Data — CANVAS provides a large-scale LSFM benchmark for whole mouse brain data with extensive cell annotations.
• Systematically decoding pathological morphologies and molecular profiles with unified multimodal embedding — Multi-Embed unifies morphological and molecular data for superior disease understanding and prediction.
• AI algorithm enables tracking of vital white matter pathways — An AI tool precisely segments brainstem white matter bundles in live MRI, revealing disease-specific structural changes and recovery.
• SP-Mind: An Autonomous Reasoning Agent for Spatial Proteomics Analysis — SP-Mind is an autonomous AI agent unifying spatial proteomics analysis from natural language queries to phenotype discovery, demonstrating superior performance.
• Deep learning-based in silico labeling for analyzing morphological features of MSCs to predict immunomodulatory capacity — A dual AI model non-invasively predicts MSC immunomodulatory capacity from morphology, enhancing cell therapy quality control.
• SMART: A Flexible, Interpretable, and Scalable Spatio-temporal Brain Atlas from High-Resolution Imaging Data — SMART creates flexible, interpretable brain atlases by decoupling global disease dynamics from individual anatomy using differential equations and Neural Cellular Automata.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine learning : nature.com subject feeds.