A contactless automated dynamic monitoring method for organoid morphology on the time axis

Xun Deng,Lun Hu, Zekun Jiang, Lu Liu,Peng-Wei Hu

2023 23RD IEEE INTERNATIONAL CONFERENCE ON DATA MINING WORKSHOPS, ICDMW 2023(2023)

引用 0|浏览0
暂无评分
摘要
Organoids, as in vitro 3D cultures with structures and functional characteristics similar to corresponding tissues, offer crucial biotechnological support for physiology, pathology, and drug testing. In the field of organoid imaging, the primary approach involves observing the dynamic changes in organoid morphology and movement during cultivation. Currently, there are limited and somewhat restrictive methods for organoid image analysis. Common techniques, such as those based on fluorescence staining, can impact organoid viability. Methods relying on bounding-box detection and tracking fail to capture organoid morphological characteristics, while threshold-based or deep neural network methods often suffer from segmentation inaccuracies. this paper presents the MacrOrga model, which is based on a combination of a multi-axis attention mechanism and convolutional residual networks. MacrOrga represents an innovative approach to organoid segmentation. MacrOrga effectively captures long-distance pixel relationships within organoid images, facilitating precise organoid segmentation. It demonstrates robust generalization capabilities across various datasets. Based on the results of the exact segmentation, the tracking of individual organoids is achieved using the KuhnMunkres algorithm to minimize the cost matrix of organoid assignment. The combination of precise segmentation and tracking functions offers a contactless, automated monitoring method for organoids in drug experiments, physiological assessments, and other research applications.
更多
查看译文
关键词
Organoids,segmentation,tracking,automated,contactless
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要