Journal of Computer Applications ›› 2024, Vol. 44 ›› Issue (3): 945-952.DOI: 10.11772/j.issn.1001-9081.2023040424

• Multimedia computing and computer simulation • Previous Articles     Next Articles

Real-time pulmonary nodule detection algorithm combining attention and multipath fusion

Kui ZHAO1,2, Huiqi QIU1,2(), Xu LI3, Zhifei XU4   

  1. 1.Shenyang Institute of Computing Technology,Chinese Academy of Sciences,Shenyang Liaoning 110168,China
    2.University of Chinese Academy of Sciences,Beijing 100049,China
    3.Institute of Future Artificial Intelligence Technology and Innovative Applications (Beijing),Beijing 100025,China
    4.School of Science and Engineering,The Chinese University of Hongkong,Shenzhen,Shenzhen Guangdong 518172,China
  • Received:2023-04-17 Revised:2023-07-14 Accepted:2023-07-14 Online:2023-12-04 Published:2024-03-10
  • Contact: Huiqi QIU
  • About author:ZHAO Kui, born in 1974, M. S., research fellow. His research interests include artificial intelligence, big data analytics, knowledge graph.
    LI Xu, born in 1996, engineer. His research interests include computer vision, medical artificial intelligence.
    XU Zhifei, born in 2000, M. S. candidate. His research interests include pattern recognition.


赵奎1,2, 仇慧琪1,2(), 李旭3, 徐知非4   

  1. 1.中国科学院 沈阳计算技术研究所, 沈阳 110168
    2.中国科学院大学, 北京 100049
    3.未来人工智能科创发展研究院(北京), 北京 100025
    4.香港中文大学(深圳) 理工学院, 广东 深圳 518172
  • 通讯作者: 仇慧琪
  • 作者简介:赵奎(1974—),男,辽宁沈阳人,研究员,硕士,主要研究方向:人工智能、大数据分析、知识图谱


Existing single-stage target detection algorithms are insensitive to nodule detection in lung nodule detection, multiple up-samplings during feature extraction by Convolutional Neural Network (CNN) has difficult feature extraction and poor detection effect, and the existing pulmonary nodule detection algorithm models are complex and not conductive to practical application employment and implementation. To address the above problems, a real-time pulmonary nodule detection algorithm combining attention mechanism and multipath fusion was proposed, based on which the up-sampling algorithm was improved to effectively increase the detection accuracy of lung nodules and speed of model inference, the model size was small and easy to deploy. Firstly, the hybrid attention mechanism of channel and space was fused in the backbone network part of feature extraction. Secondly, the sampling algorithm was improved to enhance the quality of generated feature maps. Finally, the channels were established between different paths in the enhanced feature extraction network part to achieve the fusion of deep and shallow features, so the semantic and location information at different scales was fused. Experimental results on LUNA16 dataset show that, compared to the original YOLOv5s algorithm, the proposed algorithm achieves an improvement of 9.5, 6.9, and 8.7 percentage points in precision, recall, and average precision, respectively, with a frame rate of 131.6 frames/s, and a model weight file of only 14.2 MB, demonstrating that the proposed algorithm can detect lung nodules in real time with much higher accuracy than existing single-stage detection algorithms such as YOLOv3 and YOLOv8.

Key words: deep learning, pulmonary nodule detection, attention mechanism, up-sampling algorithm, bi-directional feature pyramid


现有单阶段目标检测算法在肺结节检测中结节检出不敏感,卷积神经网络(CNN)在特征提取时多次上采样导致微小结节特征提取困难、检测效果差,并且现存肺结节检测算法模型复杂,不利于实际应用部署落地。针对上述问题,提出一种结合注意力机制和多路径融合的实时肺结节检测算法,并在此基础上改进上采样算法,提升肺部结节的检测精度和模型推理速度,且模型的权重小容易部署。首先,在特征提取的主干网络部分融合通道和空间的混合注意力机制;其次,改进采样算法,提高生成特征图的质量;最后在加强特征提取网络部分,在不同路径之间建立通道,实现深层和浅层特征的融合,将不同尺度的语义和位置信息融合。在LUNA16数据集的实验结果表明,相较于原始YOLOv5s算法,所提算法的精确率、敏感度和平均精度分别提升9.5、6.9和8.7个百分点,帧率达到131.6 frame/s,模型权重文件仅有14.2 MB,表明了所提算法可以实时检测肺结节,并且精度远高于YOLOv3和YOLOv8等现有单阶段检测算法。

关键词: 深度学习, 肺结节检测, 注意力机制, 上采样算法, 双向特征金字塔

CLC Number: