ME-VAE: Multi-Encoder Variational AutoEncoder for Controlling Multiple Transformational Features in Single Cell Image Analysis

Image-based cell phenotyping relies on quantitative measurements as encoded representations of cells; however, defining suitable representations that capture complex imaging features is challenging since there are many obstacles, including segmentation and identifying subcellular compartments for feature extraction. Variational autoencoder (VAE) approaches produce encouraging results by mapping from an image to a representative descriptor, and outperform classical hand-crafted features for morphology, intensity, and texture at differentiating data. Although VAEs show promising results for capturing morphological and organizational features in tissue, single cell image analyses based on VAEs often fail to identify biologically informative features due to the intrinsic amount of uninformative variability. Herein, we propose a multi-encoder VAE (ME-VAE) in single cell image analysis using transformed images as a self-supervised signal to extract transform-invariant biologically meaningful features. We show that the proposed architecture improves analysis by making distinct populations more separable compared to traditional VAEs and intensity measurements by enhancing phenotypic differences between cells and by improving correlations to other modalities.