Annonce

Les commentaires sont clos.

M2 Internship: Ultrasound liver image segmentation with AI

5 December 2022


Catégorie : Stagiaire


M2 Internship

Ultrasound liver image segmentation with AI:

 

Application to image-guided percutaneous procedure

 

Context

Image-guided percutaneous methods have been progressively recognized as an efficient alternative for treating Hepatocellular Carcinoma (HCC). Access to organs is ensured by using needles puncturing the skin, which represents the least invasive surgical technique to access deep internal structures into the organs [VLSP18]. However, the effectiveness of needle-based intervention treatment depends on the accuracy of the needle positioning (≈ 3 mm [JBTM18]). Reaching this accuracy is particularly challenging because, contrary to laparoscopic surgery, no direct visual access is permitted.

Non-invasive imaging techniques are required to control the needle's placement efficiently. The most spread imaging modality is the US due to its low cost, harmlessness (radiation-free), and real-time capabilities. However, the US raises the difficulty of aligning the image's acquisition plan with the needle and the tumor [KSHH15].

Our project, in collaboration with ICube (CNRS Strasbourg), aims at developing a novel solution for needle steering using intra-operative US images and non-rigid registration of a biomechanical model. The biomechanical models will be used to extrapolate the 3D displacement of the volume, even where no imaging data are available. Such an approach can then be used to display with Augmented Reality (AR) 3D information of the organ on top of medical images and automatic needle steering [BCDB20, BaCB21].

 

Main objective

Intraoperative images are traditionally used to close the loop-control, but such data are usually incomplete, sparse, and of poor quality. The proposed image-guided percutaneous method requires segmentation of the internal structures of the liver. Deep learning (DL) methods represent a robust solution for image segmentation. We will train a network using generated data from synthetic data provided by our partner ICube. Then, we will evaluate the performance of our solution with the actual US B-mode images obtained in real-time. The network's output will be the segmentation of internal structures (we envision the outline of the organ, the vessels, or the tumor). In addition, using the localization of the probe, these data will be reconstructed in 3D and serve as the basis of the biomechanical registration.

 

Profile

Master2 in computer science, image processing, AI, applied mathematics, data scientist. Autonomy, open-mindedness and motivation. Some experience in deep learning is appreciated.

 

Position context

The trainee will join the INSERM UMR1101 Laboratory of Medical Information Processing (LaTIM, Brest, France, https://latim.univ- brest.fr). We will work in collaboration with different academic, hospital and company partnerswithin a national project. Access will be given to the computer cluster PLACIS (http://placis.univ-brest.fr/english) and to clinical data from our partners.

 

Contact and additional information

For application, a CV must be sent to the following e-mails:

pierre-henri.conze@imt-atlantique.fr

julien.bert@univ-brest.fr

 

References

[VLSP18] L. Viganò, A. Laurenzi, L. Solbiati, F. Procopio, D. Cherqui, and G. Torzilli, “Open Liver Resection, Laparoscopic Liver Resection, and Percutaneous Thermal Ablation for Patients with Solitary Small Hepatocellular Carcinoma (≤30 mm): Review of the Literature and Proposal for a Therapeutic Strategy,” Dig. Surg., vol. 35, no. 4, pp. 359–371, Jul. 2018.

[JBTM18] T. L. De Jong, N. J. van de Berg, L. Tas, A. Moelker, J. Dankelman, and J. J. van den Dobbelsteen, “Needle placement errors: Do we need steerable needles in interventional radiology?,” Med. Devices Evid. Res., vol. 11, pp. 259–265, 2018.

[KSHH15] J. W. Kim, S. S. Shin, S. H. Heo, J. H. Hong, H. S. Lim, H. J. Seon, Y. H. Hur, C. H. Park, Y. Y. Jeong, and H. K. Kang, “Ultrasound-Guided Percutaneous Radiofrequency Ablation of Liver Tumors: How We Do It Safely and Completely,” Korean J. Radiol., vol. 16, no. 6, p. 1226, Nov. 2015.

[BCDB20] P. Baksic, H. Courtecuisse, C. Duriez, and B. Bayle, “Robotic needle insertion in moving soft tissues using constraint-based inverse Finite Element simulation,” Proc. - IEEE Int. Conf. Robot. Autom., pp. 2407–2413, 2020.

[BaCB21] P. Baksic, H. Courtecuisse, and B. Bayle, “Shared control strategy for needle insertion into deformable tissue using inverse Finite Element simulation,” in IEEE International Conference on Robotics and Automation, 2021, pp. 12442–12448.