文献类型：期刊文献

英文题名：Artificial intelligence-driven multimodal image analyzer for genetically encoded fluorescence sensors on cell division metabolism dynamics

作者：Xu, Hang[1];Lu, Shijie[2];Song, Yike[2];Zhou, Jiale[1];Shen, Bin[1];Zou, Yejun[3];Zhang, Zhuo[2];Zhao, Yuzheng[2];Wang, Huifeng[1,2]

机构：[1]East China Univ Sci & Technol, Key Lab Smart Mfg Energy Chem Proc, Minist Educ, Shanghai 200237, Peoples R China;[2]East China Univ Sci & Technol, Optogenet & Synthet Biol Interdisciplinary Res Ctr, Shanghai Frontiers Sci Ctr Optogenet Tech Cell, State Key Lab Bioreactor Engn, Shanghai 200237, Peoples R China;[3]Tianfu Jincheng Lab, Chengdu 610212, Peoples R China

年份：2026

卷号：37

期号：8

外文期刊名：CHINESE CHEMICAL LETTERS

收录：;WOS:【SCI-EXPANDED(收录号:WOS:001764819400001)】；

基金：This work was supported in part by the National Key R&D Pro-gram of China (No. 2023YFA1802002) and the National Natural Sci-ence Foundation of China (No. 62103148) .

语种：英文

外文关键词：Genetically encoded fluorescence sensor; Multimodal images; Artificial intelligence; Single-cell metabolism dynamics; Cell division metabolism; Cell tracking

摘要：Genetically encoded fluorescence (FL) sensors play vital roles in monitoring cell metabolism dynamics, gene transcription, DNA repair, apoptosis, and nutrient sensing. In genetically encoded FL sensors study, microscope acquires multimodal images that contain bright-field (BF) and FL images. BF and FL images can provide different information on cell phenotypes and gray level. Here, we developed an artificial intelligence (AI)-driven system "AI-cell metabolism dynamic analyzer (AI-CMDA)" for high-throughput automatically cell division metabolism dynamics processing by fusing multimodal information. The system consists of a deep learning model based on multimodal images dedicated to division nodes extraction and an adaptive correlation filter tracker for associating cell sequences. The division extraction method enables fast filtering of dividing cells without manual selection, and the adaptive correlation filter can achieve robust, accurate tracking during cell transforms with time. We apply the system with 3 genetically encoded FL sensors for nicotinamide adenine dinucleotides (NAD+ /NADH) ratio, reduced nicotinamide adenine dinucleotide phosphate (NADPH) and H2O2, respectively, to monitor the redox metabolism within cell division sequences. The results show that this system can reduce the processing time to seconds compared with several hours' manual labeling and can achieve accuracy and fastness. (c) 2026 Published by Elsevier B.V. on behalf of Chinese Chemical Society and Institute of Materia Medica, Chinese Academy of Medical Sciences.