MorphGen: Controllable and Morphologically Plausible Generative Cell-Imaging

2025-10-01

AI Summary

This study addresses the high acquisition cost and labor-intensive annotation of real fluorescence microscopy images in high-content cellular imaging. We propose an end-to-end multi-channel fluorescence image generation method built upon a diffusion model architecture, capable of jointly synthesizing morphologically faithful, multi-organelle-resolved images under diverse perturbation conditions. Crucially, we introduce a representation alignment loss that enforces consistency between the latent-space representations of generated images and those of the biological foundation model OpenPhenom, thereby significantly improving morphological plausibility and biological fidelity. Quantitative evaluation shows our method reduces the Fréchet Inception Distance (FID) by over 35% compared to the state-of-the-art MorphoDiff. CellProfiler-based morphometric analysis further confirms strong agreement between synthetic and real images in cellular morphology features. The generated images are thus suitable for fine-grained drug screening and phenotypic analysis of genetic perturbations.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Generation

Application Category

• Web Mining and Content Analysis: Web data generation and simulation
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Simulating in silico cellular responses to interventions is a promising direction to accelerate high-content image-based assays, critical for advancing drug discovery and gene editing. To support this, we introduce MorphGen, a state-of-the-art diffusion-based generative model for fluorescent microscopy that enables controllable generation across multiple cell types and perturbations. To capture biologically meaningful patterns consistent with known cellular morphologies, MorphGen is trained with an alignment loss to match its representations to the phenotypic embeddings of OpenPhenom, a state-of-the-art biological foundation model. Unlike prior approaches that compress multichannel stains into RGB images -- thus sacrificing organelle-specific detail -- MorphGen generates the complete set of fluorescent channels jointly, preserving per-organelle structures and enabling a fine-grained morphological analysis that is essential for biological interpretation. We demonstrate biological consistency with real images via CellProfiler features, and MorphGen attains an FID score over $35%$ lower than the prior state-of-the-art MorphoDiff, which only generates RGB images for a single cell type. Code is available at https://github.com/czi-ai/MorphGen.

Problem

• Simulating cellular responses to interventions for drug discovery
• Generating fluorescent microscopy images with biological consistency
• Preserving organelle-specific details across multiple cell types

Innovation

• Diffusion-based generative model for microscopy imaging
• Alignment loss training with biological foundation model
• Joint generation of complete fluorescent channel set