【摘要】 在酿酒酵母(Saccharomyces cerevisiae)的细胞衰老研究中，微流控芯片可以实现单个酵母细胞的捕获和培养，同时用光学显微镜进行时序图像拍摄，最后分析图像得到细胞的复制寿命。针对基于“漏碗”式捕获结构的微流控芯片长时序酵母细胞实验图像，为提高处理效率，提出了一种基于卷积神经网络的高效算法，用于提取酵母细胞在衰老过程中的出芽特征。经验证，提出的算法模型准确率达到95%,F1 score指标达到平均90%以上，具有较高的可靠性。为深度学习方法在生物医学图像处理领域的应用提供了新的思路。更多还原

【Abstract】 In yeast aging studies with budding yeast(Saccharomyces cerevisiae),microfluidic chips are used for yeast cell immobilization and cell culture on single cell scale. Based on time-lapse imaging, replicative lifespan(RLS)can be determined through image analysis. However, manual image analysis is time-consuming and laborious. Here, an algorithm based on convolutional neural network is proposed to extract the budding events of single yeast cell, which is tailored for microscopic images of cells on microfluidic chips with “leaky bowl”traps. The effectiveness of the algorithm is proved by high accuracy of 95% and F1 score of 90% on average. Therefore, the proposed algorithm provides a novel method for single-cell analysis, and it could be used for similar work in the field of biomedicine image analysis.更多还原