Sparse self-attention aggregation networks for neural sequence slice interpolation

doi:10.1186/s13040-021-00236-z

CORC > 自动化研究所 > 中国科学院自动化研究所 > 类脑智能研究中心 > 微观重建与智能分析

	Sparse self-attention aggregation networks for neural sequence slice interpolation
	Wang,Zejin 1,4; Liu,Jing 1,4; Chen,Xi 4; Li,Guoqing 4; Han,Hua 2,3,4
刊名	BioData Mining
	2021-02-01
卷号	14 期号:1 页码:19
关键词	Slice interpolation Biological tissue recovery EM images Sparse self-attention network Adaptive style-balance loss
ISSN号	1756-0381
DOI	10.1186/s13040-021-00236-z
通讯作者	Li,Guoqing(guoqing.li@ia.ac.cn) ; Han,Hua(hua.han@ia.ac.cn)
英文摘要	AbstractBackgroundMicroscopic imaging is a crucial technology for visualizing neural and tissue structures. Large-area defects inevitably occur during the imaging process of electron microscope (EM) serial slices, which lead to reduced registration and semantic segmentation, and affect the accuracy of 3D reconstruction. The continuity of biological tissue among serial EM images makes it possible to recover missing tissues utilizing inter-slice interpolation. However, large deformation, noise, and blur among EM images remain the task challenging. Existing flow-based and kernel-based methods have to perform frame interpolation on images with little noise and low blur. They also cannot effectively deal with large deformations on EM images.ResultsIn this paper, we propose a sparse self-attention aggregation network to synthesize pixels following the continuity of biological tissue. First, we develop an attention-aware layer for consecutive EM images interpolation that implicitly adopts global perceptual deformation. Second, we present an adaptive style-balance loss taking the style differences of serial EM images such as blur and noise into consideration. Guided by the attention-aware module, adaptively synthesizing each pixel aggregated from the global domain further improves the performance of pixel synthesis. Quantitative and qualitative experiments show that the proposed method is superior to the state-of-the-art approaches.ConclusionsThe proposed method can be considered as an effective strategy to model the relationship between each pixel and other pixels from the global domain. This approach improves the algorithm’s robustness to noise and large deformation, and can accurately predict the effective information of the missing region, which will greatly promote the data analysis of neurobiological research.
资助项目	NSFC[61701497] ; Instrument function development innovation program of Chinese Academy of Sciences[E0S92308] ; Instrument function development innovation program of Chinese Academy of Sciences[282019000057] ; Strategic Priority Research Program of Chinese Academy of Science[XDB32030200]
WOS研究方向	Mathematical & Computational Biology
语种	英语
出版者	BioMed Central
WOS记录号	BMC:10.1186/S13040-021-00236-Z
资助机构	NSFC ; Instrument function development innovation program of Chinese Academy of Sciences ; Strategic Priority Research Program of Chinese Academy of Science
内容类型	期刊论文
源URL	[http://ir.ia.ac.cn/handle/173211/42242]
专题	类脑智能研究中心_微观重建与智能分析
通讯作者	Li,Guoqing; Han,Hua
作者单位	1.School of Artificial Intelligence, University of Chinese Academy of Sciences 2.School of Future Technology, University of Chinese Academy of Sciences 3.Center for Excellence in Brain Science and Intelligence Technology Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences 4.National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences
推荐引用方式 GB/T 7714	Wang,Zejin,Liu,Jing,Chen,Xi,et al. Sparse self-attention aggregation networks for neural sequence slice interpolation[J]. BioData Mining,2021,14(1):19.
APA	Wang,Zejin,Liu,Jing,Chen,Xi,Li,Guoqing,&Han,Hua.(2021).Sparse self-attention aggregation networks for neural sequence slice interpolation.BioData Mining,14(1),19.
MLA	Wang,Zejin,et al."Sparse self-attention aggregation networks for neural sequence slice interpolation".BioData Mining 14.1(2021):19.