【摘要】 为缓解空间转录组学数据因高离散性、稀疏性和分辨率受限导致空间域识别精度低的问题，基于多尺度邻域特征融合策略，本研究提出了一种空间域识别方法stMSAA。本研究在人类背外侧前额叶皮层数据集和人类乳腺癌数据集上进行实验，调整兰德系数、归一化互信息和聚类评价指数分别达到0.452、0.603和0.560。结果表明，stMSAA可有效整合多尺度的空间邻域信息，更准确地识别层次结构和边缘信息，提高空间域识别的精度和鲁棒性。更多还原

【Abstract】 To alleviate the problem of low spatial domain recognition accuracy due to high discreteness, sparsity and limited resolution of spatial transcriptomic data, based on multi-scale neighborhood feature fusion strategy, we proposed a spatial domain recognition method stMSAA. The experiment was conducted on the human dorsolateral prefrontal cortex dataset and the human breast cancer dataset. The adjusted rand index, normalized mutual information, and fowlkes-mallows index reached 0.452, 0.603 and 0.560, respectively. The results show that stMSAA can effectively integrate different scale spatial neighborhood information, accurately identify hierarchical structures, edge information and improve the accuracy and robustness of spatial domain identification.更多还原