Deep learning-based ultrasonic image reconstruction for ultrafast echographic imaging
Recrutement: 
Recrutement en cours/passé: 
Recrutement en cours
Periode: 
2019
Contact: 
Denis Friboulet et Fabien Millioz : denis.friboulet@creatis.insa-lyon.fr, fabien.millioz@univ-lyon1.fr
  1. Context:

Ultrafast imaging based on plane wave (PW) or diverging wave (DW) is currently a very active area of research in ultrasound because of its capacity of reaching very high frame rate (several thousands/s) in 2D or 3D. Obtaining ultrafast images using plane or diverging wave imaging however still remains a challenge in situations where extremely complex and fast phenomena happen, e.g. in the heart, due to the trade-off between image quality and frame rate: PW/DW imaging indeed relies on compounding in order to preserve a good image quality, usually using multiple successive emissions, which in turn yields a decrease of the frame rate. One possible strategy to overcome this problem consists in reducing the number of acquired PW and developing a reconstruction method yielding high quality images despite the missing data.

  1. Objective:

Compressed sensing techniques are well suited to this type of reconstruction and have been recently applied to this problem at Creatis [1, 2]. However, compressed sensing-based reconstruction involves solving a minimization problem of very large dimensions, and the associated algorithms currently do not yield computing speeds compatible with ultrafast acquisitions.

In this context, the objective of this internship is to develop an alternative method of reconstruction based on a deep neural network (DNN), which will reduce the computation time by orders of magnitude. The application of DNN to image reconstruction problems is an emerging topic [3, 4]. The feasibility of its application to PW has been very recently demonstrated at Creatis [5, 6, 7].

Consequently, the key points to be addressed in this work will be the following:

  • Adaptation of the method for DW imaging

  • Selection of the neural network type. In particular, the formulations based on convolutional networks (CNN), Generative adversarial networks (GAN) and hypernetworks networks (CPPN) will have to be examined.

  • Once a type of network will have been selected, optimization of the network structure.

  • Selection of input/output US data (i.e. raw radiofrequency, beamformed or envelope data) yielding the best performances

The perspectives of the developed approach concern two major issues in ultrasound: i) improving the rate of acquisition in 2D cardiac imaging, by reducing the number of DW emissions required to achieve high image quality ii) performing ultra-fast 3D acquisitions under routine clinical conditions, reducing the number of active elements of the probe needed to record the volume of data.

  1. Methodology:

    The development of DNNs adapted to the representation and reconstruction of ultrasound data will be based on PyTorch python library. The developed approach will be optimized and evaluated in terms of computing time/accuracy trade-off on numerical simulations, experimental phantom data and in vivo data acquired with the research ultrasound scanners available at Creatis.

  2. Applicant’s profile:

Good knowledge in Machine Learning, with a good experience of deep neural networks, as well as excellent skills in programming, training and testing such networks.

Prior knowledge in ultrasound imaging is not a prerequisite, since the candidate will be trained in this field at Creatis. Some background and interest in medical imaging and in ultrasound imaging in particular will nevertheless be an asset.

  1. References:

    1. Lorintiu, O., Liebgott, H., Alessandrini, M., Bernard, O., Friboulet, D., 2015. Compressed sensing reconstruction of 3D ultrasound data using dictionary learning and line-wise subsampling. IEEE Trans. Med. Imaging 34, 2467‑2477.

    2. Lorintiu, O., Liebgott, H., Friboulet, D., 2016. Compressed sensing Doppler ultrasound reconstruction using block sparse Bayesian learning. IEEE Trans. Med. Imaging 35, 978-987.

    3. Kulkarni, K., Lohit, S., Turaga, P., Kerviche, R., Ashok, A., 2016. ReconNet: Non-Iterative Reconstruction of Images from Compressively Sensed Random Measurements, IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 449-458.

    4. Mousavi, A., Baraniuk, R.G., 2017. Learning to invert: signal recovery via deep convolutional networks. ArXiv, 1‑5. https://arxiv.org/abs/1701.03891

    5. Gasse, M., Millioz, F., Roux, E., Garcia, D., Liebgott, H., Friboulet, D., 2017. High-Quality Plane Wave Compounding using Convolutional Neural Networks. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control. PP. 1-1. 10.1109/TUFFC.2017.2736890

    6. Gasse, M. ,Millioz, F., Roux, E., Liebgott, H., Friboulet, D., 2017. Accelerating Plane Wave Imaging through Deep Learning-based Reconstruction: An Experimental Study. IEEE International Ultrasonics Symposium.

    7. Gasse, M., 2018. Deep Learning for ultrasound imaging IEEE International Ultrasonics Symposium, Invited talk, accepted.