Microscopy generation improves with biologically meaningful cell controls
Rethinking Microscopy Generation: Co-Designed Diffusion for Biologically Interpretable Single-Cell Synthesis
Computer Vision and Pattern Recognition
Summary
Microscopy images usually show many cells, but controlling how each cell looks is hard. The authors designed a new way to create images by focusing on single cells and linking the input controls directly to meaningful biological traits like size and brightness. They built a special model called BioVAE that learns to organize cell features in a way that matches these traits. This joint design makes generated cells look more realistic and lets users change specific traits one at a time. Experiments with real data and biologists showed that this model produces believable and interpretable cell images.
What this means in practice
- •For computational biologists: Create synthetic single-cell microscopy images with precise control over cell features for better biological data augmentation and analysis.
- •For pharmaceutical image analysts: Generate realistic variations of cellular phenotypes to test and train automated drug screening algorithms more effectively.
Authors
Shumpei Takezaki, Duway Nicolas Lesmes-Leon, Ryoma Bise, Gillian Lovell, Bianca Migliori, Andreas Dengel, Sheraz Ahmed
Abstract
Generative models for microscopy can already produce realistic whole-image fields of view, but offer no principled way to control how each individual cell looks. We address this gap by recasting microscopy generation at the single-cell level and designing the input controls and the model's internal representation jointly, so that both are organized along the same biologically meaningful directions. On the input side, we aggregate a long list of standard cell measurements into a small set of semantic axes (size, elongation, brightness, texture, etc.) that are each named, measurable, and traversable. On the model side, we introduce BioVAE, an encoder-decoder whose latent space is explicitly aligned with the same set of measurements. A key finding is that the benefit of biologically structuring the latent emerges only when conditioning is also moved to the semantic axes, showing that the two ends must be designed jointly. Experiments on LIVECell show that semantic-axis conditioning substantially improves generative fidelity over high-dimensional conditioning, a blind study with biologists confirms the realism of generated cells, and per-axis traversals produce stable and interpretable phenotypic changes.