深度学习技术在生物显微图像中的应用

Application of deep learning technique in biological microscopic images

近年来深度学习技术已经逐步渗入到各行各业之中,尤其是计算机视觉领域,深度学习凭借其强大的图像高维信息的处理能力,将人类所不能理解的信息进行提取及自动化分析。随着生物医学技术的不断发展,研究者们不断深入了解生物的本质,生物显微技术在其中就起到了十分重要的作用。而随之产生的大量生物显微图像,如数字病理图、荧光细胞图像等,正好与数据驱动的深度学习技术相契合,使生物显微图像的分析得到快速的发展。本文先简要对深度学习技术进行了介绍,然后根据图像分类、目标检测、分割、超分辨率等在生物显微图像上的子任务对深度学习的应用及创新进行详细论述,最后展望了深度学习在生物显微图像上的发展趋势及所面临的挑战。

In recent years,deep learning technology has gradually penetrated into all aspects of life,especially in the field of computer vision.With its powerful processing ability of high-dimensional image information,deep learning can extract and automate the analysis of information that cannot be understood by humans.Thanks to the continuous development of biomedical technology,researchers have studied biology at a micro level,in which biological microscopy plays critical role.Subsequently,a large number of biological microscopic images,such as digital pathological images and fluorescent cell images,coincided with the data-driven deep learning technology,have brought a rapid progress in biological microscopic image analysis.In this paper,the deep learning technology is briefly introduced,and the application and innovation of deep learning are discussed in detail according to the sub-tasks of image classification,target detection,segmentation and super resolution,etc.Finally,the development trends and challenges of deep learning in biological microscopic images are prospected.