| Abstrakt: |
Researchers from the University of Hong Kong have developed a new technique called "Cyto-Morphology Adversarial Distillation" (CytoMAD) that addresses the challenges of image-based cytometry. CytoMAD is a self-supervised multi-task learning strategy that distills biologically relevant cellular morphological information from batch variations, allowing for integrated analysis across multiple data batches without complex data assumptions or extensive manual annotation. The technique has been shown to enhance diagnostic and screening workflows, including drug screening and tumor biopsy evaluation, by providing robust, label-free cellular insights across diverse datasets. This research was supported by the Hong Kong Special Administrative Region of China and the Hong Kong Research Grants Council. [Extracted from the article] |
