notesum.ai

Published at December 4

ASIGN: An Anatomy-aware Spatial Imputation Graphic Network for 3D Spatial Transcriptomics

cs.CV

Released Date: December 4, 2024

Authors: Junchao Zhu¹, Ruining Deng¹, Tianyuan Yao¹, Juming Xiong², Chongyu Qu, Junlin Guo², Siqi Lu, Mengmeng Yin³, Yu Wang, Shilin Zhao⁴, Haichun Yang, Yuankai Huo¹

Aff.: ¹Department of Computer Science Vanderbilt University; ²Department of Electrical and Computer Engineering Vanderbilt University; ³Department of Pathology, Microbiology and Immunology Vanderbilt University Medical Center; ⁴Department of Biostatistics Vanderbilt University Medical Center

Arxiv: http://arxiv.org/pdf/2412.03026v1

Dimension	Model	HER2[23]			DLPFC[25]			ST-Data[3]
Dimension	Model	MSE	MAE	PCC	MSE	MAE	PCC	MSE	MAE	PCC
2D WSI-wise	ST-Net[3]	0.663	0.633	0.430	0.496	0.552	0.738	0.804	0.705	0.517
	EGN [10]	0.580	0.585	0.442	0.401	0.490	0.757	0.674	0.644	0.554
	HisToGene[13]	0.669	0.644	0.432	0.348	0.472	0.753	0.575	0.598	0.552
	BLEEP [11]	0.743	0.665	0.340	0.304	0.429	0.737	0.581	0.590	0.496
	His2ST [15]	0.591	0.591	0.441	0.409	0.494	0.764	0.534	0.568	0.539
	ASIGN 2D (Ours)	0.579	0.590	0.476	0.275	0.411	0.766	0.528	0.566	0.552
3D Sample-wise	Overlap	0.461	0.532	0.625	0.292	0.422	0.749	0.447	0.525	0.663
	Similarity	0.449	0.517	0.649	0.278	0.412	0.762	0.579	0.548	0.696
	ASIGN 3D (Ours)	0.363	0.473	0.693	0.353	0.458	0.773	0.332	0.459	0.739