Improved YOLOv3 Network Combined with LSTM Model and Attention Module for Cerebral Aneurysm Detection
DOI:
https://doi.org/10.12974/2313-1047.2025.12.01Keywords:
YOLOv3, LSTM, Cerebral aneurysm, Attention mechanismsAbstract
Cerebral aneurysm is a kind of cerebrovascular disease, which is mainly diagnosed by reading the MRA slice data to diagnose whether it is suffering from cerebral aneurysm or not, and the medical image detection method based on deep learning can help doctors to improve the detection accuracy and efficiency. Small target detection and the interference of vascular region are the difficulties in cerebral aneurysm detection, which is prone to misdetection or missed detection. Aiming at these problems, we propose an improved method for cerebral aneurysm detection by introducing the LSTM model and the attention module on the basis of the YOLOv3 network, optimizing it in terms of feature processing, time series information construction, weight allocation, etc., using the structure of the YOLOv3 network to achieve effective feature extraction, the regression ability of the LSTM model to construct the time series information among the sliced sequences, and the attention module to assign weights to improve the detection ability of small targets and prevent the interference of blood vessels on the detected targets to improve the detection performance of the network. The experimental results prove the effectiveness of the above improved method which shows significant improvement in the accuracy and anti-interference detection. There is a significant improvement in the detection of cerebral aneurysms, with the precision index reaching 70.8%, an increase of 8.7%, the recall index reaching 76.2%, an increase of 5.0%, and the mAP index reaching 69.9%, an increase of 4.7%, which improves the ability to detect small targets and reduces the interference of blood vessels with the target of detection.
References
CHEN Meng, GENG Chen, LI Yu-xin, et al. Automatic Detection for Cerebral Aneurysms in TOF-MRA Images Based on Fuzzy Label and Deep Learning [J]. Chinese Journal of Magnetic Resonance. 2022; 39(3): 267-277.
Girshick, Ross, Donahue, Jeff, Darrell, Trevor, et al. Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation[C]//2014 IEEE Conference on Computer Vision and Pattern Recognition: 2014 27th IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2014), 23-28 June 2014, Columbus, Ohio.: Institute of Electrical and Electronics Engineers, 2014: 580-587. https://doi.org/10.1109/CVPR.2014.81
Mingjian Zhu. Dynamic Feature Pyramid Networks for Object Detection[C]//Fifteenth International Conference on Signal Processing Systems (ICSPS 2023): 17-19 November 2023. Xi an, China. 2024: 130911N.1-130911N.9.
Joseph Redmon, Santosh Divvala, Ross Girshick, et al. You Only Look Once: Unified, Real-Time Object Detection[C]//29th IEEE Conference on Computer Vision and Pattern Recognition: 29th IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 26 June – 1 July 2016, Las Vegas, Nevada.: Institute of Electrical and Electronics Engineers, 2016: 779-788. https://doi.org/10.1109/CVPR.2016.91
Wei Liu, Dragomir Anguelov, Dumitru Erhan, et al. SSD: Single Shot Multi-Box Detector[C]//Computer vision - ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, p.I.: Springer, 2016: 21-37. https://doi.org/10.1007/978-3-319-46448-0_2
Dan Yang, Lichun Yang, Dabiao Zhou. Stripe removal method for remote sensing images based on multi-scale variation model[C]//2019 IEEE International Conference on Signal Processing, Communications and Computing: ICSPCC 2019, Dalian, China, 20-22 September 2019.: Institute of Electrical and Electronics Engineers, 2019: 61-65. https://doi.org/10.1109/ICSPCC46631.2019.8960737
Mingxue Bi, Bingjie Hu, Handong Yu, et al. Long and short-term memory neural network multicomponent gas quantification correction algorithm based on sparrow search algorithm[C]//International Conference on Mechatronic Engineering and Artificial Intelligence (MEAI 2023), Part Two of Two Parts: 15-17 December 2023. Shenyang, China. 2024: 130712W.1-130712W.7.
Sepp Hochreiter, Yoshua Bengio, Paolo Frasconi, and Jürgen Schmidhuber. Gradient flow in recurrent nets: the difficulty of learning long-term dependencies, 2001.
Bo Jin. Detection of Cerebral Aneurysms Based on Deep Learning. Huazhong University of Science and Technology, 2020:19.
Hugo Larochelle, Geoffrey Hinton. Learning to combine foveal glimpses with a third-order Boltzmann machine[C]//Advances in Neural Information Processing Systems 23. vol. 2.: Neural Information Processing Systems, 2010: 1243-1251.
Artsiom Ablavatski, Shijian Lu, Jianfei Cai. Enriched Deep Recurrent Visual Attention Model for Multiple Object Recognition[C]//2017 IEEE Winter Conference on Applications of Computer Vision: [Volume 2 of 2] Pages 660-1314: IEEE Computer Society, 2017: 971-978. https://doi.org/10.1109/WACV.2017.113
Chunxi Wang, Maoshen Jia, Meiran Li, et al. Attention is All You Need for Blind Room Volume Estimation[C]//ICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2024), Vol.3: Seoul, South Korea.14-19 April 2024. 2024: 1341-1345. https://doi.org/10.1109/ICASSP48485.2024.10447723
Sheng-Hua Zhong, Yan Liu, Feifei Ren, et al. Video Saliency Detection via Dynamic Consistent Spatio-Temporal Attention Modelling[C]//Proceedings of the twenty-seventh AAAI conference on artificial intelligence and the twenty-fifth innovative applications of artificial intelligence conference: 14-18 July 2013, Bellevue, Washington, USA, v.2.: AAAI Press, 2013: 1063-1069. https://doi.org/10.1609/aaai.v27i1.8642
Tao Shen, Jing Jiang, Tianyi Zhou, et al. DiSAN: Directional Self-Attention Network for RNN/CNN-Free Language Understanding[C]//Proceedings of the Thirty-Second AAAI Conference on Artificial Intelligence: Thirtieth Innovative Applications of Artificial Intelligence Conference and Eighth Symposium on Educational Advances in Artificial Intelligence: New Orleans, Louisiana, USA, 2-7 February 2018, Volume Seven.: AAAI Press, 2018: 5446-5455.
Jingkuan Song, Lianli Gao, Zhao Guo, et al. Hierarchical LSTM with Adjusted Temporal Attention for Video Captioning[C]//International Joint Conferences on Artificial Intelligence: IJCAI 2017, Melbourne, Australia, 19-25 August 2017, Volume 3, Part A.: Curran Associates, Inc., 2019: 2737-2743. https://doi.org/10.24963/ijcai.2017/381
Christoph Feichtenhofer, Axel Pinz, Andrew Zisserman. Convolutional Two-Stream Network Fusion for Video Action Recognition[C]//29th IEEE Conference on Computer Vision and Pattern Recognition: 29th IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 26 June – 1 July 2016, Las Vegas, Nevada.: Institute of Electrical and Electronics Engineers, 2016: 1933-1941. https://doi.org/10.1109/CVPR.2016.213
Buracas GT, Boynton GM. The effect of spatial attention on contrast response functions in human visual cortex [J]. The Journal of Neuroscience: The Official Journal of the Society for Neuroscience, 2007; 27(1): 93-97. https://doi.org/10.1523/JNEUROSCI.3162-06.2007
Yao L, Torabi A, Cho K, et al. Video Description Generation Incorporating Spatio-Temporal Features and a Soft-Attention Mechanism [J]. Eprint Arxiv, 2015; 53: 199-211.
Tao Shen, Tianyi Zhou, Guodong Long, et al. Reinforced Self-Attention Network: A Hybrid of Hard and Soft Attention for Sequence Modeling[C]//27th International Joint Conference on Artificial Intelligence and 23rd European Conference on Artificial Intelligence: IJCAI-ECAI 2018, Stockholm, Sweden, 13-19 July 2018, Volume 6 of 8: Curran Associates, Inc., 2018: 4345-4352. https://doi.org/10.24963/ijcai.2018/604
David Dembinsky, Fatemeh Azimi, Federico Raue, et al. Sequential Spatial Transformer Networks for Salient Object Classification[C]//12th International Conference on Pattern Recognition Applications and Methods: ICPRAM 2023, Lisbon, Portugal, 22-24 February 2023, Part 1 of 2. 2023: 328-335. https://doi.org/10.5220/0011667100003411
