Impact

Geometrical and anatomical modelling:

After two years, PASSPORT has already reached several of its objectives. The first task (workpackage, in short WP) of the project has thus led to a new automated and fast segmentation of thoraco-abdominal CT-scan images and a new liver and surrounding organs segmentation from US. It also provided a new mesh generation algorithm providing topologically correct meshes of segmented organs. In the same way, it allowed the integration of a mesh refinement algorithm in a framework dedicated to mesh and also the automatic quality estimation thanks to dedicated metrics that will be used for instance to both estimate the quality of a mesh for visualisation and for deformation simulation.

Moreover, this first WP has defined a new anatomical segmentation of the liver that avoids current anatomical mistakes and therefore improves the postoperative success of liver surgery. This new topologically correct patient-specific anatomy is based on the result of the state of the art in liver anatomy that showed that Couinaud's anatomy was correct in only 40% of cases.

Mechanic, texture, biology and dynamic models:

The second WP that aims at developing patient-specific mechanical modelling, has allowed the development of a new MRI Elastographic (MRE) acquisition. Comparison with the Transcient US elastography clearly demonstrates greater quality of such modelling providing fewer errors in fibrosis detection. From an in vitro and experimental in vivo evaluation, we have also obtained a first Finite Model element of a liver that will be completed and linked to the MRE information during the upcoming last year of the project.

The third WP has provided an organ texture database from HD endoscopic views of real patients with and without liver pathologies. This database has been used to develop new high-speed texture-rendering techniques providing a fully realistic rendering. Moreover, an automatic texture generation from endoscopic views of patients has been developed.

The fourth WP was aiming at providing biological modelling of the liver. Results achieved offer today the first effective simulation of liver cell regeneration. Moreover, MRI perfusion and diffusion image analysis developed in this WP offer the possibility to characterize liver pathologies from a MRI medical image. These results will have to be more precisely validated during the last year of the project.

Hepatic lobule regeneration simulation: cell scale. The fifth WP is focusing on the dynamic modelling of liver movements. Several motion models have been derived from real-time MRI acquisitions, ultrasonic images, endoscopic images and external camera. First validations show a possible predictive simulation of liver movements during breathing cycles with a 2 mm precision.

Integration:

Thanks to WP6, the various models are progressively integrated in an opensource framework dedicated to surgical simulation: SOFA (www.sofa-framework.org). It allows the development of patient-specific simulation, a first educative version having already been developed.

Moreover, several softwares have been developed in the integration WP7 from a new opensource framework dedicated to computer-assisted surgical software development (code.google.com/p/fw4spl).

The first one, 3D VPM 2.0, integrates new segmentation and modelling algorithms and can be used in another medical field of the Virtual Physiological Human. The second one, VR-Render©IRCAD2010, a freeware available on the PASSPORT website, is a new DICOM viewer, integrating direct volume rendering techniques (that do not require organ modelling) and surface-rendering techniques (after organ modelling). The initial patient medical image and resulting segmented organs can thus be easily visualized with this free user friendly software. The third one, VRRender WebSurg Limited Edition ©IRCAD2010, is an educative version of VR-Render available freely on the free online virtual university WebSurg. It allows to link the patient modelling to an educative video of the surgery applied to the same patient. The last software, VR-Planning ©IRCAD2010, offers the opportunity to resect a liver with several topological components allowing thus multi segmentectomy. It then automatically computes the future liver remain rate and volumetry of each resected part.

Validation:

From our results, we have created an online service providing free patient modelling to several associated European clinical teams. First results clearly demonstrate the benefits of the resulting patient-specific surgical planning based on a more precise Future Liver remain volumetry.

From the more than 300 models realized, an open anonymous database of real clinical cases has been created. This database containing medical images in DICOM and 3D models of organs is freely available to the scientific community on PASSPORT's website.

Track record of sharing

PASSPORT project results have been shared to the scientific community through the development of two open source frameworks, one dedicated to computer-assisted surgery (FW4SPL) and one for real-time surgical simulation (SOFA).  Moreover, the PASSPORT project has opened a free access to anonymous patient databases integrating medical images, their associated 3D modelling and also intraoperative video illustrating the use of software (freely available on www.websurg.com/softwares). The extension of PASSPORT results to other specialities is thus open and several examples of such applications have been successfully realized in pancreatic surgery, spleen surgery, colorectal surgery, endocrine surgery and paediatric surgery.

Lessons learnt

"The anatomy has to be based on ontology": The modern anatomy that has been described and used for five centuries is based on the comparison with an average anatomy extracted from living human or having lived. All patients being different, such a definition leads to mistakes or approximations that are called exceptions. In liver anatomy, these "exceptions" represent 60% of cases. The PASSPORT project clearly demonstrates that replacing such a definition of the anatomy by labelling rules applied to each patient modelled in 3D from their medical image, allows the correction of such mistakes. Couinaud's anatomy of the liver defined in 1957 will thus have to be replaced by the resulting new definition, allowing to avoid diagnostic mistakes and to improve surgical eligibility.

"Preoperative patient-specific modelling is mandatory in liver surgery": Eligibility and definition of a hepatic surgical procedure have to be defined from patient-specific modelling of the patient including geometry, topology, mechanics, biology and functionality. Without such modelling, approximation is too important and can lead to a wrong estimation of the optimal therapy to apply.

"Surgeons are not only waiting for preoperative assistance but also for an intraoperative one":  Augmented reality consists in the superimposition of preoperative data on the live intraoperative vision of the patient. The resulting virtual transparency can be used to guide the surgeon intraoperatively. Such guidance is one of the biggest dreams of surgeons who wait for such an improvement of results. Future European projects will have to target such a problem, since it is today not available for surgery on soft tissues due to many limitations.