Original title: EfNST: A composite scaling network of EfficientNet for improving spatial domain identification performance
Authors: Yanan Zhao,Chunshen Long,Na Yin,Zhihao Si,Wenjing Shang,Zhenxing Feng,Yongchun Zuo
This article discusses the use of Spatial Transcriptomics (ST) in studying tissue structure, tumor microenvironment, and biological development. ST combines gene expression profiling with spatial location and histological images, providing valuable insights into these areas of research. However, accurately identifying spatial domains using computational methods has been challenging due to the poor performance of existing algorithms. The authors propose a new deep learning algorithm called EfNST, specifically designed for the analysis of 10X Visium spatial transcriptomics data. They applied EfNST to three different datasets and compared its performance to five other advanced algorithms. EfNST outperformed the competing algorithms, achieving the best scores in terms of identifying spatial domains and discovering marker genes. It also demonstrated a high level of accuracy and improved running speed. In conclusion, EfNST offers a novel approach to analyzing spatial organization of cells, allowing for new insights in this field. The authors declare no competing interests.
Original article: https://www.biorxiv.org/content/10.1101/2023.12.03.569798v1