Histopathology is diagnosis based on visual examination of tissue sections under a microscope. With the growing number of digitally scanned tissue slide images, computer-based segmentation and classification of these images is a high-demand area of research. Convolutional neural networks (CNNs) constitute the most popular classification architecture for a variety of image classification problems. However, applying CNNs to histology slides is not a trivial task and has several challenges, ranging from variations in the colors of slides to excessive high resolution and lack of proper labeling. In this advanced review, we introduce the application of CNN-based architectures to digital histological image analysis, discuss some problems associated with such analysis, and look at possible solutions.

中文翻译：

---

组织病理学中的深度学习：综述

组织病理学是基于在显微镜下组织切片的视觉检查的诊断。随着数字扫描组织切片图像的数量不断增加，这些图像的基于计算机的分割和分类是一个高需求的研究领域。卷积神经网络 (CNN) 构成了用于各种图像分类问题的最流行的分类架构。然而，将 CNN 应用于组织学载玻片并不是一项微不足道的任务，并且面临着一些挑战，从载玻片颜色的变化到过高的分辨率和缺乏适当的标签。在这篇高级评论中，我们介绍了基于 CNN 的架构在数字组织学图像分析中的应用，讨论了与此类分析相关的一些问题，并研究了可能的解决方案。