Fig. 1: Framework of the PhenoProfiler for morphology representations.
From: PhenoProfiler: advancing phenotypic learning for image-based drug discovery
a Flowchart comparison of end-to-end PhenoProfiler with existing non-end-to-end methods. b PhenoProfiler includes a gradient encoder to enhance edge gradients, improving clarity and contrast in cell morphology. A transformer encoder then captures high-dimensional dependencies and intricate relationships, enriching image representations. A designed multi-objective learning module is utilized for accurate morphological representation learning. c For model inference, PhenoProfiler uses phenotype correction strategy (PCs) with hyperparameter α to identify morphological changes between treated and control conditions. Created in BioRender. Song, Q. (2025) https://BioRender.com/v0nqs11.
