计算机应用 ›› 2020, Vol. 40 ›› Issue (11): 3332-3339.DOI: 10.11772/j.issn.1001-9081.2020030355

• 虚拟现实与多媒体计算 • 上一篇    下一篇

基于改进的Mask R-CNN的染色体图像分割框架

冯涛1,2,3, 陈斌1,2, 张跃飞1,2,3   

  1. 1. 中国科学院 成都计算机应用研究所, 成都 610041;
    2. 中国科学院 广州电子技术研究所, 广州 510075;
    3. 中国科学院大学 计算机科学与技术学院, 北京 101408
  • 收稿日期:2020-03-26 修回日期:2020-05-29 出版日期:2020-11-10 发布日期:2020-06-10
  • 通讯作者: 陈斌(1970-),男,四川广汉人,研究员,博士,主要研究方向:机器视觉、深度学习;chenbin306@sohu.com
  • 作者简介:冯涛(1994-),男,广西钦州人,硕士研究生,主要研究方向:计算机视觉、深度学习;张跃飞(1990-),男,山西忻州人,硕士研究生,主要研究方向:目标检测、实例分割

Chromosome image segmentation framework based on improved Mask R-CNN

FENG Tao1,2,3, CHEN Bin1,2, ZHANG Yuefei1,2,3   

  1. 1. Chengdu Institute of Computer Applications, Chinese Academy of Sciences, Chengdu Sichuan 610041, China;
    2. Guangzhou Institute of Electronic Technology, Chinese Academy of Sciences, Guangzhou Guangdong 510075, China;
    3. School of Computer Science and Technology, University of Chinese Academy of Sciences, Beijing 101408, China
  • Received:2020-03-26 Revised:2020-05-29 Online:2020-11-10 Published:2020-06-10

摘要: 针对染色体图像的人工分割耗时费力且当前自动分割方法精度不佳的问题,基于改进的Mask R-CNN提出了一种染色体图像分割框架——Mask Oriented R-CNN,引入方向信息对染色体图像进行实例分割。首先,新增有向包围框回归分支,以预测紧实包围框并获取方向信息;然后,提出新的交并比(IoU)度量——角度加权交并比(AwIoU),从而结合方向信息与边的关系以改进冗余包围框的判据;最后,实现有向卷积通路结构,通过拷贝掩模分支通路并依据实例的方向信息选择训练路径来减少掩模预测中的干扰。实验结果表明,相较于基准模型Mask R-CNN,Mask Oriented R-CNN在IoU阈值为0.5时的平均精度均值指标提升了10.22个百分点,IoU阈值为0.5~0.95时的平均指标提升了4.91个百分点。研究结果显示,Mask Oriented R-CNN框架相较于基准模型取得了更好的染色体图像分割结果,有助于实现染色体图像自动分割。

关键词: 卷积神经网络, 实例分割, Mask R-CNN, 染色体图像分割, 图像分割, 非极大值抑制, 交并比

Abstract: The manual segmentation of chromosome images is time-consuming and laborious, and the accuracy of current automatic segmentation methods is not poor. Therefore, based on improved Mask R-CNN (Mask Region-based Convolutional Neural Network), a chromosome image segmentation framework named Mask Oriented R-CNN (Mask Oriented Region-based Convolutional Neural Network) was proposed, which introduced orientation information to perform instance segmentation of chromosome images. Firstly, the regression branch of oriented bounding boxes was added to predict the compact bounding boxes and obtain orientation information. Secondly, a novel Intersection-over-Union (IoU) metric called AwIoU (Angle-weighted Intersection-over-Union) was proposed to improve the criterion of redundant bounding boxes by combining the relationship between the orientation information and edges. Finally, the oriented convolutional path structure was realized to reduce the interference in mask prediction by copying the path of mask branch and selecting the training path according to the orientation information of the instances. Experimental results show that compared with the baseline model Mask R-CNN, Mask Oriented R-CNN has the mean average precision increased by 10.22 percentage points when the IoU threshold is 0.5, and the mean metric increased by 4.91 percentage points when the IoU threshold is from 0.5 to 0.95. Experimental results show that the Mask Oriented R-CNN framework achieves better segmentation results than the baseline model in chromosome image segmentation, which is helpful to achieve automatic segmentation of chromosome images.

Key words: Convolutional Neural Network (CNN), instance segmentation, Mask R-CNN (Mask Region-based Convolutional Neural Network), chromosome image segmentation, image segmentation, Non-Maximum Suppression (NMS), Intersection-over-Union (IoU)

中图分类号: