Skip to Main content Skip to Navigation

Un système de réalité augmentée pour guider les opérations du foie en radiologie interventionnelle

Abstract : In the interventional radiology field, the medical expert needs several CT-scan acquisitions (2D slices) to reach the target during a percutaneous puncture in the abdominal area. We aim at designing a guidance system to reduce the number of CT acquisition and thus, the intervention duration. We propose an augmented reality system that superimposes preoperative 3D reconstructions of the patient abdominal structures in external video images of his/her body. This system will not only have to be accurate, quick and reliable, but will also have to be easily introduced in the operative room.
In our case, the patient is intubated and the breathing is controlled. Therefore, we can neglect breathing motion: a rigid 3D/2D registration of radio-opaque markers is sufficient to obtain the necessary accuracy. Since statistical assumptions of existing 3D/2D registration criteria do not fit our application, we derived a new criterion generalizing classical methods. A rigorous evaluation of the performances showed the superiority of our approach in terms of accuracy and robustness.
To reach real-time processing in the operative room, we developed and validated several algorithms to automatically extract and match radio-opaque markers. Since the system accuracy depends on numerous factors (camera position, marker number...), it is not possible to calculate it beforehand. To provide a reliable system, we propose a technique of covariance propagation that enables to estimate the superimposition error of reconstructed models. A rigorous validation step on synthetic and real data showed that our prediction is correct in our conditions.
All the different parts of the system being validated, we perform an evaluation of the whole system on an abdominal phantom. Results proved the feasibility and the interest of the system: four surgeons managed to reach emboddied targets with an average accuracy under 2 mm. Moreover, the time needed for the needle positioning was ten times under classical interventional duration. Finally, several clinical experiments on patients showed that our system can be used in the operating room and suggest that this system could be used routinely in a near future.
Complete list of metadata
Contributor : Stephane Nicolau <>
Submitted on : Tuesday, May 3, 2005 - 11:28:31 AM
Last modification on : Monday, October 19, 2020 - 11:08:46 AM
Long-term archiving on: : Thursday, April 1, 2010 - 9:24:45 PM


  • HAL Id : tel-00000006, version 1



Stephane Nicolau. Un système de réalité augmentée pour guider les opérations du foie en radiologie interventionnelle. Vision par ordinateur et reconnaissance de formes [cs.CV]. Université Nice Sophia Antipolis, 2004. Français. ⟨tel-00000006⟩



Record views


Files downloads