Abstract
Beamforming is an essential step in ultrasound image reconstruction that can alter both image quality and framerate. Adaptive methods estimate a set of data-dependent apodization weights among which Minimum Variance Beamforming (MVB) is one of the most powerful approaches performing well regardless of the imaging settings. MVB, however, is not applicable online as it is computationally expensive. Recently, in order to speed up MVB, we took advantages of state-of-the-art methods in deep learning and adapted the MobileNetV2 structure to train and test a model that mimics MVB. In terms of image quality, our method was ranked first in the Challenge on Ultrasound Beamforming with Deep Learning (CUBDL). However, considering both image quality and network size, our method was jointly ranked first with another submission which had a smaller number of parameters. The number of parameters and processing time are important especially for the point-of-care ultrasound machines, which have limited size and computational power. Herein, we propose an approach to prune the trained MobileNetV2 to reduce the number of parameters and computational complexity to further speed-up beamforming. Results confirm that there is no discernible reduction in the network performance, in terms of either visual or quantitative comparison after pruning. In terms of the memory footprint, the post-pruned networks contain 0.3 million parameters as compared to the 2.3 million pre-pruned networks, a reduction by a factor of 7.67. The run-times of MVB, pre- and post-pruned models are 4.05, 0.67 and 0.29 min, respectively.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
References
Afrakhteh, S., Behnam, H.: Low-complexity adaptive minimum variance ultrasound beam-former based on diagonalization. Biomed. Sig. Process. Control 62, 102110 (2020). https://doi.org/10.1016/j.bspc.2020.102110
Bell, M.A.L., Huang, J., Hyun, D., Eldar, Y.C., van Sloun, R., Mischi, M.: Challenge on ultrasound beamforming with deep learning (CUBDL). In: 2020 IEEE International Ultrasonics Symposium (IUS), pp. 1–5 (2020). https://doi.org/10.1109/IUS46767.2020.9251434
Blalock, D., Ortiz, J.J.G., Frankle, J., Guttag, J.: What is the state of neural network pruning? arXiv preprint arXiv:2003.03033 (2020)
Chen, J., Chen, J., Zhuang, R., Min, H.: Multi-operator minimum variance adaptive beamforming algorithms accelerated with GPU. IEEE Trans. Med. Imag. 39(9), 2941–2953 (2020). https://doi.org/10.1109/TMI.2020.2982239
Goudarzi, S., Asif, A., Rivaz, H.: Fast multi-focus ultrasound image recovery using generative adversarial networks. IEEE Trans. Comput. Imag. 6, 1272–1284 (2020). https://doi.org/10.1109/TCI.2020.3019137
Goudarzi, S., Asif, A., Rivaz, H.: Ultrasound beamforming using mobilenetv2. In: 2020 IEEE International Ultrasonics Symposium (IUS), pp. 1–4 (2020). https://doi.org/10.1109/IUS46767.2020.9251565
Han, S., Pool, J., Tran, J., Dally, W.: Learning both weights and connections for efficient neural network. In: Advances in Neural Information Processing Systems, vol. 28. Curran Associates, Inc. (2015)
Hyun, D., et al.: Deep learning for ultrasound image formation: CUBDL evaluation framework amp;amp; open datasets. IEEE Trans. Ultrasonics Ferroelectrics Frequency Control 1 (2021). https://doi.org/10.1109/TUFFC.2021.3094849
Luchies, A.C., Byram, B.C.: Assessing the robustness of frequency-domain ultrasound beamforming using deep neural networks. IEEE Trans. Ultrasonics Ferroelectrics Frequency Control 67(11), 2321–2335 (2020). https://doi.org/10.1109/TUFFC.2020.3002256
Luijten, B., et al.: Adaptive ultrasound beamforming using deep learning. IEEE Trans. Med. Imag. 39(12), 3967–3978 (2020). https://doi.org/10.1109/TMI.2020.3008537
Rodriguez-Molares, A., et al.: The ultrasound toolbox. In: 2017 IEEE International Ultrasonics Symposium (IUS), pp. 1–4 (2017). https://doi.org/10.1109/ULTSYM.2017.8092389
Rothlübbers, S., et al.: Improving image quality of single plane wave ultrasound via deep learning based channel compounding. In: 2020 IEEE International Ultrasonics Symposium (IUS), pp. 1–4 (2020). https://doi.org/10.1109/IUS46767.2020.9251322
Sandler, M., Howard, A., Zhu, M., Zhmoginov, A., Chen, L.C.: Mobilenetv 2: Inverted residuals and linear bottlenecks. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 4510–4520 (2018)
Sandler, M., Howard, A., Zhu, M., Zhmoginov, A., Chen, L.C.: Mobilenetv 2: Inverted residuals and linear bottlenecks. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2018
Synnevag, J.f., Austeng, A., Holm, S.: Benefits of minimum-variance beamforming in medical ultrasound imaging. IEEE Trans. Ultrasonics Ferroelectrics Frequency Control 56(9), 1868–1879 (2009). https://doi.org/10.1109/TUFFC.2009.1263
Vaidya, A.S., Srinivas, M.: A low-complexity and robust minimum variance beamformer for ultrasound imaging systems using beamspace dominant mode rejection. Ultrasonics 101, 105979 (2020). https://doi.org/10.1016/j.ultras.2019.105979
Acknowledgment
We acknowledge the support of the Natural Sciences and Engineering Research Council of Canada (NSERC) RGPIN-2020-04612 and RGPIN-2017-01539. We would like to thank NVIDIA for the donation of the Titan Xp GPU.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2021 Springer Nature Switzerland AG
About this paper
Cite this paper
Goudarzi, S., Asif, A., Rivaz, H. (2021). Pruning MobileNetV2 for Efficient Implementation of Minimum Variance Beamforming. In: Noble, J.A., Aylward, S., Grimwood, A., Min, Z., Lee, SL., Hu, Y. (eds) Simplifying Medical Ultrasound. ASMUS 2021. Lecture Notes in Computer Science(), vol 12967. Springer, Cham. https://doi.org/10.1007/978-3-030-87583-1_21
Download citation
DOI: https://doi.org/10.1007/978-3-030-87583-1_21
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-87582-4
Online ISBN: 978-3-030-87583-1
eBook Packages: Computer ScienceComputer Science (R0)