《计算机应用》唯一官方网站 ›› 2024, Vol. 44 ›› Issue (10): 3200-3208.DOI: 10.11772/j.issn.1001-9081.2023101416

• 多媒体计算与计算机仿真 • 上一篇    下一篇

结合加权原型和自适应张量子空间的小样本宫颈细胞分类

谢莉1(), 舒卫平1, 耿俊杰1, 王琼2, 杨海麟3   

  1. 1.江南大学 物联网工程学院,江苏 无锡 214122
    2.南京医科大学附属无锡人民医院 医学检验科,江苏 无锡 214023
    3.江南大学 生物工程学院,江苏 无锡 214122
  • 收稿日期:2023-10-19 修回日期:2023-12-21 接受日期:2023-12-22 发布日期:2024-10-15 出版日期:2024-10-10
  • 通讯作者: 谢莉
  • 作者简介:谢莉(1985—),女,重庆人,副教授,博士,主要研究方向:图像处理、软测量建模、系统辨识 xieli@jiangnan.edu.cn
    舒卫平(1998—),男,安徽黄山人,硕士研究生,主要研究方向:深度学习、图像处理
    耿俊杰(2000—),男,江苏南通人,硕士研究生,主要研究方向:深度学习、图像处理
    王琼(1981—),女,湖北咸宁人,副主任医师,博士,主要研究方向:细胞图像识别、血液系统疾病发病机制与诊断标志物
    杨海麟(1971—),男,江苏无锡人,教授,博士,主要研究方向:生物细胞图像处理、发酵过程优化与控制。
  • 基金资助:
    国家重点研发计划项目(2022YFC3401302)

Few-shot cervical cell classification combining weighted prototype and adaptive tensor subspace

Li XIE1(), Weiping SHU1, Junjie GENG1, Qiong WANG2, Hailin YANG3   

  1. 1.School of Internet of Things Engineering,Jiangnan University,Wuxi Jiangsu 214122,China
    2.Department of Medical Laboratory,Wuxi People’s Hospital Affiliated to Nanjing Medical University,Wuxi Jiangsu 214023,China
    3.School of Biological Engineering,Jiangnan University,Wuxi Jiangsu 214122,China
  • Received:2023-10-19 Revised:2023-12-21 Accepted:2023-12-22 Online:2024-10-15 Published:2024-10-10
  • Contact: Li XIE
  • About author:SHU Weiping, born in 1998, M. S. candidate. His research interests include deep learning, image processing.
    GENG Junjie, born in 2000, M. S. candidate. His research interests include deep learning, image processing.
    WANG Qiong, born in 1981, Ph. D., associate chief physician. Her research interests include cell image recognition, pathogenesis and diagnostic markers of hematological diseases.
    YANG Hailin, born in 1971, Ph. D., professor. His research interests include biological image processing, optimization and control of fermentation process.
  • Supported by:
    National Key Research and Development Program of China(2022YFC3401302)

摘要:

基于深度学习的图像分类算法通常依赖大量训练数据,然而对于医学领域中的宫颈细胞分类任务,难以实现收集大量的图像数据。为了在少量图像样本的条件下正确分类宫颈细胞,提出一种结合加权原型和自适应张量子空间的小样本分类算法(CWP-ATS)。首先,结合预训练技术和元学习,保证特征提取网络从元训练集中学习更多的先验知识;其次,在原型计算过程中采用最大均值差异算法为每个支持集样本赋予合适的权重,并采用转导学习算法修正,以获得更准确的原型;最后,利用多线性主成分分析算法将每类样本投影至各自的低维张量子空间,从而在不破坏原始张量特征自然结构的前提下,在低维空间中学习高效的自适应子空间分类器。在小样本Herlev宫颈细胞图像的2-way 10-shot和3-way 10-shot分类任务中,与DeepBDC(Deep Brownian Distance Covariance)算法相比,CWP-ATS的分类准确度分别提高了2.43和3.23个百分点;当元测试集中30%的样本受噪声干扰时,与原型网络相比,CWP-ATS的分类准确度有超过20个百分点的提升。实验结果表明,CWP-ATS有效提高了对小样本宫颈细胞的分类准确度和鲁棒性。

关键词: 图像分类, 宫颈细胞, 小样本学习, 加权原型, 自适应张量子空间

Abstract:

Deep learning image classification algorithms rely on a large amount of training data typically. However, for cervical cell classification tasks in the medical field, collecting large amount of image data is difficult. To accurately classify cervical cells with a limited number of image samples, a few-shot classification algorithm Combining Weighted Prototype and Adaptive Tensor Subspace (CWP-ATS) was proposed. Firstly, the pre-training technique was combined with meta-learning to ensure that the feature extraction network learned more priori knowledge from the meta-training set. Subsequently, the maximum mean discrepancy algorithm was adopted in the prototype computation procedure to assign appropriate weight to each support set sample, and the transductive learning algorithm was further employed for making corrections and obtaining more accurate prototypes. Finally, the multilinear principal component analysis algorithm was utilized to project each class of samples into their respective low-dimensional tensor subspaces, enabling efficient adaptive subspace classifiers in the low-dimensional space to be learned without compromising the natural structures of the original tensor features. In the 2-way 10-shot and 3-way 10-shot classification tasks of few-shot Herlev cervical cell images, compared with the DeepBDC (Deep Brownian Distance Covariance) algorithm, CWP-ATS improved classification accuracy by 2.43 and 3.23 percentage points, respectively; when 30% samples of the meta-test set were interfered by noise, in comparison with the prototype network, the classification accuracy of CWP-ATS was improved by more than 20 percentage points. The experimental results demonstrate that the proposed algorithm can effectively improve the classification accuracy and robustness of few-shot cervical cells.

Key words: image classification, cervical cell, few-shot learning, weighted prototype, adaptive tensor subspace

中图分类号: