Disentangled representations of microscopy images

2025/06/25 by Jacopo Dapueto, Vito Paolo Pastore, Dapueto, Jacopo +5
Biochemistry, Genetics and Molecular Biology · #Artificial Intelligence (cs.AI) #Cell Image Analysis Techniques #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Machine Learning (cs.LG)

paper · pdf · doi:10.48550/arxiv.2506.20649

openalex publication_date 2025/06/25 · openalex created_date 2025/10/09 · openalex updated_date 2026/07/28

Abstract

Microscopy image analysis is fundamental for different applications, from diagnosis to synthetic engineering and environmental monitoring. Modern acquisition systems have granted the possibility to acquire an escalating amount of images, requiring a consequent development of a large collection of deep learning-based automatic image analysis methods. Although deep neural networks have demonstrated great performance in this field, interpretability, an essential requirement for microscopy image analysis, remains an open challenge. This work proposes a Disentangled Representation Learning (DRL) methodology to enhance model interpretability for microscopy image classification. Exploiting benchmark datasets from three different microscopic image domains (plankton, yeast vacuoles, and human cells), we show how a DRL framework, based on transferring a representation learnt from synthetic data, can provide a good trade-off between accuracy and interpretability in this domain.

Citations

• Transferring disentangled representations: bridging the gap between synthetic and real images
• Flow Factorized Representation Learning
• DisDiff: Unsupervised Disentanglement of Diffusion Probabilistic Models
• Disentangled Representation Learning
• An Empirical Study on Disentanglement of Negative-free Contrastive Learning
• A Contrastive Objective for Learning Disentangled Representations
• Emerging Properties in Self-Supervised Vision Transformers
• Where and What? Examining Interpretable Disentangled Representations
• Learning Disentangled Representation by Exploiting Pretrained Generative Models: A Contrastive Learning View
• On the Transfer of Disentangled Representations in Realistic Settings
• Theory and Evaluation Metrics for Learning Disentangled Representations
• Are Disentangled Representations Helpful for Abstract Visual Reasoning?
• Towards a Definition of Disentangled Representations
• Robustly Disentangled Causal Mechanisms: Validating Deep Representations\n for Interventional Robustness
• Learning Deep Disentangled Embeddings with the F-Statistic Loss
• A Survey of Inductive Biases for Factorial Representation-Learning
• InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets
• Ladder Variational Autoencoders
• Deep Convolutional Inverse Graphics Network
• Representation Learning: A Review and New Perspectives
• Representation Learning: A Review and New Perspectives
• Greedy function approximation: A gradient boosting machine.

Related

• Discovering interpretable models of scientific image data with deep learning
• Roadmap on Deep Learning for Microscopy
• Attention Pooling Enhances NCA-based Classification of Microscopy Images
• Multi-domain learning CNN model for microscopy image classification
• Masked Autoencoders are Scalable Learners of Cellular Morphology