Schedule - Access map - Program - Video - Slides

Workshop 1-2 December 2008 - Amphi A207

Ecole Nationale Supérieure de Physique de Strasbourg ENSPS

Bd S. Brant, Boulevard Sébastien Brant  - Pôle d'Innovation API, BP 10413
F-67412 ILLKIRCH
Tél     : (+33) (0)3 90 24 44 90
Fax    : (+33) (0)3 90 24 43 42 
Email :  Christophe.Collet(at)ensps.u-strasbg.fr

. Organizing committee : Professor Barbara Bass (Methodist Hospital, Houston) Professor Marc Garbey (Computer Science Univ. Houton) Professor Christophe Collet (LSIIT, Strasbourg University) Professor Michel de Mathelin (LSIIT, Strasbourg University) Professor Roger Tran Son Tay (University of Florida)

•  Barbara Bass (Dept. of Surgery – The Methodist Hospital) and Marc Garbey – CS@UH: "Can We Talk the Same Language? Barriers and Opportunities in Cross Disciplinary Research"

The value of multidisciplinary research is often exalted; cross fertilization of ideas, knowledge and skills. However, building cohesive research programs between scientists of various backgrounds and with clinicians can prove challenging at best. This session will identify potential barriers to success and offer some preliminary debate on methods to foster collaborative research based.

•  Barbara Bass (Dept. of Surgery – The Methodist Hospital) and Marc Garbey – CS@UH: “Breast Conserving Therapy for Breast Cancer: Targets for Inverstigation to Improve Results" (part1 ; part 2)

Breast conserving therapy, which requires complete removal of the malignant tumor and treatment of the remaining breast with radiation therapy, is feasible in up to 70 per cent of women with breast cancer. The benefits of the procedure are that the woman is spared the surgical risks of total mastectomy with or without reconstructive surgery and the disfigurement of loss of the breast. However, cosmetic results in BCT are inconsistent and there is no current technology that allows prediction of breast contour or deformity after lumpectomy and radiation. Reliable preoperative modeling and prediction would support surgical approaches to optimize tumor resection and cosmetic results.

•  B.Bayle (LSIIT Univ. of Strasbourg), “Robot-assisted procedures in interventional radiology”

Interventional Radiology is a medical field in which some of the most promising procedures are performed. It consists in the guidance of small tools, like needles or catheters, to a zone in which a treatment or a diagnosis has to be done. These minimally invasive techniques allow limited trauma to the surrounding organs together with good results for indicated treatments, like liver cancer or spine fracture. A drawback of these procedures regardless to conventional surgery is due to the interaction with the medical imaging devices. Depending on the modality, there may have different limitations, for instance because of images acquisition rate, radiations or tools registration.

In this talk we will discuss how medical robotic assistance can help radiologists in performing needle insertions efficiently and comfortably, i.e. with automatic guidance tools and perceptual feedback. We will describe the general specifications of needle insertion procedures and we will survey already existing systems. We will point out the more relevant problems in the field of robotic assistance for interventional radiology. Finally, we will focus on the particular case of interventional radiology with CT imaging, for which the robotic assistance is strongly desirable in order to protect medical staff from X-radiations.

•  Scott Berceli (Dept. of Surgery University of Florida), “Emerging Mechanisms of Vein Graft Failure”

Although the last decade has witnessed significant advances in cardiovascular health, few durable treatment options are available for patients with symptomatic peripheral arterial disease. Vein bypass grafting stands as a mainstay of arterial occlusive disease treatment, yet the durability of this therapy is frequently measured in the time frame of months. Therapies aimed at improving vein graft durability have been limited by a paucity of information regarding the local and systemic factors that control this vascular response to injury. Work over the last decade confirm that physical forces serve as the primary regulator of local vein graft behavior. Based on animal experimentation by our group and others, the standard paradigm of vein graft remodeling has been established, where reductions in wall shear stress induce an accelerated intimal hyperplastic response with narrowing of the lumen. While attractive in its simplicity and likely valid within a uniform flow-field, this relationship fails to explain the substantial heterogeneity in lesion development observed in a complex, three-dimensional geometry. Recent work in our laboratory redefines these concepts, demonstrating that the dynamics of the complex flow-field are critical in understanding the remodeling process. Using a rabbit vein graft, focal stenosis model, we identified a biphasic relationship between shear stress and the resulting hyperplastic response. Despite evolving evidence in animal models, the application of these concepts to the understanding of human vein graft failure remains poorly defined. Parallel to the basic science and computational efforts, our laboratory has been active in development of a translational research program. Critical to these efforts is imaging of in vivo human vein grafts, with particular emphasis on evolving vein graft stenotic lesions and their response to the hemodynamic environment, where computed tomography (CT) scanning is used to obtain accurate geometries along the length of the vein graft. Coupled with the hemodynamic information obtained from Doppler spectral analysis, dynamic shear stress profiles are calculated and mapped to the three-dimensional graft geometry. Sequential examination provides a powerful tool to understand the influence of physical forces on lesion development.

•  Jean Bismuth (Dept. of Cardiovascular Surgery – The Methodist Hospital) “New Dimensions in Diagnostic Imaging of the Aorta”

Digital subtraction angiography (DSA) remains the gold standard for arterial imaging. Recent advances in software and resolution have expanded the use of computed tomographic (CTA) and magnetic resonance angiography (MRA). Even with 3-D reconstruction using these techniques, there remains a major limitation. Arteries are dynamic organs and as such imaging properties should include a fourth dimension (4-D), namely motion. Due to its anatomical position, the thoracic aorta is particularly predisposed to have wall motion throughout the cardiac cycle, and this plays a significant role in endovascular management of aortic pathology. The management of aortic dissection is particularly challenging and understanding this disease process and its treatment have remained an enigma, which we aim to resolve by combining dynamic phase contrast MRA with computational fluid dynamics (CFD). We have prospectively analyzed patients referred to The Methodist Hospital for management of Stanford Type B dissections. All patients had evaluation by phase contrast MRA and CFD. We developed a mathematical model, which evaluates wall motion of the aorta. We use this model to predict which patients would benefit from endovascular treatment of their dissection versus medical management. We have observed that wall motion is much more pronounced in the acute aortic dissection, and given appropriate indications these patients have recovered well when treated with a stentgraft. Wall shear stress, dynamic pressures, and flow vectors with CFD have also allowed us to understand the conformational and adaptive changes the aorta goes through. We have successfully managed patients based on 4-D.

•  I Bloch (Telecom ParisTech), “Knowledge-driven recognition and segmentation of internal brain structures in 3D MRI”

In this presentation, I will stress the importance of structural knowledge for anatomical structures recognition in brain images. This knowledge can be expressed using spatial relations between structures. An ontology of spatial relations has been built and used to enriched the FMA. This was further enriched by fuzzy models of the spatial relations, as a mean to fill the semantic gap. This framework was applied to the recognition on internal brain structures in 3D MRI images, in both normal and pathological cases.

•  Thierry Colin (University Bordeaux 1 and Inria) – “Multi-Scale Model of Tumor Growth”

•  Christophe Collet (LSIIT Univ.of Strasbourg), “Segmentation using atlas information for MS lesion detection”

Brain imaging techniques provide each day an increasing quantity of multimodal data. To deal with such data set, data cube analysis are nowadays developed to assist experts in the qualitative and quantitative analysis of multimodal images and to make easier their interpretation. Automatic segmentation has become a fundamental step for quantitative analysis in many brain diseases such as Multiple Sclerosis (MS). In this talk the study is focused on the segmentation of multimodal brain MRI and MS lesion detection. Firstly, a method for brain tissue segmentation based on hidden Markov chains models taking into account the neighborhood information is presented. This approach includes prior information provided by a probabilistic atlas, takes into account the heterogeneity of the intensity on the MRI images and partial volume e?ects to produce segmentation maps of each tissue on each voxel. In a second part, this original method is coupled to detect MS lesions by using a robust estimator. Based on Bayesian inference and statistical behavior model, lesions are detected as outliers to a statistical model of normal brain images. All algorithms have been validated on synthetic and real images. The results were compared with other existing methods of segmentation, and with manual segmentations carried out by experts. This approach has been in competition in MICCAI'08 (New York, USA, Sept 2008) “A Grand Challenge II : MS lesion segmentation” and obtained the best score among nine teams for Pre vs Onsite competition.

•  Mark Davies (Dept. of Cardiovascular Surgery – The Methodist Hospital) “Changing Paradigms in the Management of Peripheral Vascular Disease”

The prevalence of lower extremity arterial occlusive disease is increasing and interventions in this area are becoming a principal component of the management of atherosclerosis. The symptoms of atherosclerosis in the superficial femoral artery and in the tibial vessels lead to immobility and limb loss. While conservative therapy for lower extremity atherosclerosis was the norm in past decades, aggressive endovascular intervention with angioplasty and stenting is now ubiquitous in the current decade. The vessels of the lower extremity have unique anatomic biomechanical and biological characteristics. The superficial femoral artery (SFA) is the longest artery in the body and experiences a unique array of biomechanical forces during daily life. While the modeling and imaging of these vessels has been restricted to healthy volunteers, there is only limited data on the biological and mechanical features of these vessels as they develop atherosclerosis, as they stenose and produce symptoms and as they respond to intervention (drug therapy, balloon angioplasty or stent placement). There is are several potential areas of research:

• Modeling vessel biomechanics and blood flow in atherosclerosis

• Modeling the impact of angioplasty and stenting on the vessel

• Modeling the patterns of restenosis, dissection and occlusion after intervention

• Modeling the changes in bypass grafts (vein and prosthetic).

Successful modeling and simulations of the SFA and tibial vessels is crucial to further therapeutic advances in the treatment of lower extremity disease and has the potential to garner significant industry collaboration and support.

•  Michel de Mathelin (LSIIT Univ.of Strasbourg), “Natural Orifice Transluminal Surgery - NOTES, a new field for medical robotics”

Nutural Orifice Endoscopic Transluminal Surgery (NOTES) is a new approach to minimally invasive surgery that uses a flexible endoscope to reach the abdominal cavity through a natural orifice and an internal lumen like, e.g., the mouth and the gastric wall. The surgery is performed using flexible instruments going through working channels of the flexible endoscope. The procedure requires up to three surgeons to manipulate the endoscope and the instruments. Robotics may become an answer to increase the manipulability of the instruments and to reduce the number of required hands. Current developments made in Strasbourg towards robotization of flexible endoscopes for NOTES will presented. In particular, active physiological motion cancellation is demonstrated with a prototype motorized gastroscope.

•  Craig Fisher (Dept. of Surgery – The Methodist Hospital) "Surgical Data Management - Improving Situational Awareness with an Improved Data Interface"

Current electronic medical records (EMR) are complex systems of data capture for diverse data types, including alphanumeric text, numbers and images. Integration of these diverse data types has been a challenge. Furthermore, design of the graphical user interface (GUI) has received little attention. Current data design and management includes a nested folder environment and spreadsheet format for user interface with data. Medical data is complex – the user is often bombarded with huge quantities of data, most of while is not relevant to the users current task. For example, if a physician/user is seeing a patient for appendicitis a key group of laboratory and radiologic tests is required and other data may not be germane, such a immunization status. However, the user is required by current EMR user interfaces to search and find the relevant information. No artificial intelligence is used and the navigation process is cumbersome. The linkage of relevant medical data to parts of the body (organized anatomically and by system) would create a visual EMR that would allow users to click on a body part (the appendix) and get information relevant to that organ (studies, test relevant to appendicitis). A system of automatically identifying organs, using standard computed tomography, will be described. The database creation and intelligent ‘tagging' of data relevant to particular diagnosis will be described.

•  Edgar Gabriel (CS@UH), “Image Analysis of FNAC Images for Automated Cancer Detection on Multi-Core Processors and PC Clusters”

With the upcoming of multi-core processors and the omni-presence of clustered computational resources, image analysis of large scale images such as occurring in medical image computing for automated cancer detection, becomes a viable option. However, exploring parallel computing for image analysis requires non-trivial modifications to the source and cautious performance tuning. In this talk we present an OpenMP parallelization of a code used for image analysis of FNAC images targeting muli-core desktop systems, as well as an MPI parallel version of the code for clustered resources.

•  Jacques Gangloff (LSIIT Univ. of Strasbourg), “Computer assisted minimally invasive heart surgery”

Off-pump Coronary Artery Bypass Grafting (CABG) is still a technically difficult procedure. The mechanical stabilizers used for local suppression of the heart excursion have been demonstrated to exhibit significant residual motion, which could lead to a lack of accuracy in performing the surgical task, particularly when using a minimally invasive surgery (MIS) approach. We therefore propose a novel active stabilizer to compensate for the residual motion whose architecture is compatible with MIS. An experimental evaluation of a commercially available totally endoscopic stabilizer is first presented to demonstrate the unsatisfactory behavior of this device. Then, the interaction between the heart and a mechanical stabilizer is assessed in vivo using an animal model. Finally, the principle of active stabilization, based on the high-speed vision-based control of a piezoactuated compliant mechanism, is presented, along with in vivo experimental results obtained using a prototype to demonstrate its efficiency.

•  Marc Garbey and Victoria Hilford (CS@UH) “A Computational Desk for Surgeons”

Our Computational Desk for Surgery (CDS) is a system which combines numerical simulation with high-definition visualization and large data storage into a single desk. These three sub-systems are interdependent. They are connected to the network to get external information from various imaging systems, and to provide data mining and data processing. Our target application for the CDS is endovascular surgery but the concept is general and might be applied to other fields of surgery. The challenge is to provide to surgeons an interface to access all the data of interest at once on multiple displays. The interface must be friendly and very intuitive to the surgeon. We will present the design and software integration of our prototype.

•  Ioannis Kakadiaris (CS@UH), “Quo Vadis, Cardiovascular Informatics ?”

Approximately 1.5 million heart attacks are suffered annually by Americans, and about half of them prove fatal, despite a host of new public health initiatives targeting heart disease and its aggravating factors such as obesity. The case of Former US President Bill Clinton, who underwent quadruple bypass surgery, demonstrates that even a former president with access to the best medical care available can have undiagnosed heart disease. Clinton himself blamed “insufficient vigilance” and stressed the importance of repeated testing as a means of heart disease prevention. Considering the large amounts of data that a comprehensive vascular health screening will produce, there is an urgent need for biomedical image analysis tools (segmentation, shape and motion estimation) to assist in screening for the conditions that underlie sudden cardiac events. In this talk, we present biomedical image analysis tools for the mining of information from cardiovascular imaging for the detection of persons with a high likelihood of developing a heart attack in the near future (vulnerable patients).

•  Christof Karmonik (Dept. of Radiology – The Methodist Hospital): “Multi-Modality Imaging for the Simulation of Cerebral Aneurysm Blood Flow Dynamics”

Computational simulations of blood flow in cerebral aneurysm is an active field of research (1-10) motivated by the potential of assessing aneurysmal rupture risk through knowledge of hemodynamic parameters, in particular wall shear stresses (WSS)(11-13). It has been suggested that WSS modulates the endothelial cell phenotype and that a WSS magnitude of less than 1.5 Pascal (Pa) may lead to the degeneration of the arterial wall (14).

Due to the lack of a reliable method to measure wall shear stress in-vivo, computational fluid dynamics (CFD) are increasingly being used in these simulations to obtain patient-specific results. An accurate knowledge of the vulnerability of a particular aneurysms may have impact on the decision to treat the aneurysms either by endovascular techniques or through open surgery or to postpone treatment.

Initially, geometrical data derived from rotational 3D digital subtraction angiographic (DSA) images of the aneurysm and the surrounding vasculature were used in the CFD simulations. Recent work however suggested, that accurate patient-specific results can only be obtained if also considering patient-specific physiological flow information (6,15) Therefore, we have developed a multi-modality approach combining X-ray fluoroscopy, 3D DSA, time-of-flight magnetic resonance angiography and phase contrast magnetic resonance imaging for obtaining geometrical and physiological boundary conditions for CFD simulations of cerebral aneurysms.

•  Tirimachos Bourlai (CS@UH) : “Thermal Imaging of the Superficial Temporal Vessels: Advances in Design and Algorithmic Steps of a Pulse Measurement System"

The measurement of the arterial pulse from the extended frontal branch of the Superficial Temporal Vessels (STV) using passive thermal infra red sensors in a semi-automatic manner is a very complicated process with many key factors to consider. It is based on the information contained in the thermal signal emitted from the superficial vessel under study. The temperature on the frontal STV is modulated by the pulsative nature of the blood flow.To compute the frequency of the modulation and the arterial pulse waveform, we employ our two enhanced Along (ALM) and Across (ACM) models. They both have a physical and physiological basis and are capable of extracting a line-based region either along or across a selected vessel. Both use a Fourier-based, vessel registration dependent, signal analysis approach to capitalize on the thermal propagation effect of pulse. At the final stage, pulse is quantified by the use of an adaptive estimation function applied on the average FFT outcome. In this work we present the advances in design, and algorithmic steps of our pulse measurement system employing our two models and concentrating on an extensive sensitivity analysis. The main design issues are to prove the performance advantages of using our new tracking noise cleaning algorithm (TNCA) developed to optimize the parameters of both of our models, and finally to identify the motion tracking methodology (single tracking, sequential tracking, or automatic micro-tracking) that provides the optimum performance versus computational complexity trade-off.

The adopted methodology of optimization of the relevant parameters is using the Paired Student's T-Test (PSTT) and the Cumulative Sum Error (CUSUM) as our basic criterions of optimality. However, we also evaluate the performance of the method regarding the mean pulse computation by calculating the normalized mean pulse difference (NMPD) against the mean ground-truth measurements over the whole thermal clip. We conducted experiments on a data-set of 12 subjects acquired from a high quality thermal camera and validated the results using a standard contact sensor (piezo-electric transducer) as a ground truth. Our efforts achieved highly statistically significant results and improved the performance of our methodology from 10.5% to 7.5% in terms of CUSUM, and from 88.6% to 97.7% in terms of NMPD, depending on the model used and the clarity of the arterial's thermal imprint. The best overall results in terms of performance and computational time are achieved by using the sequential tracker and the optimized ALM model. To the best of our knowledge, an analysis of that extent on a semi-automatic pulse measurement system has never been reported in the literature.

•  Shishir Shah (CS@UH): “Automated Cell Segmentation and Computer-Aided Tool for Fast Screening of Thyroid Fine Needle Aspiration Cytology Smears”

This talk presents a multispectral microscopy system for quantitative cytology. The automated system aims to classify cytology samples of thyroid follicular lesions for the purpose of disease diagnosis. While conventional practices rely on the analysis of grey scale or RGB color images, presented system uses thirty one spectral bands for analysis. A novel method that leverages information across multiple spectral bands to detect and segment cells of interest is developed and integrated into a tool to aid rapid screening of digitized fine needle aspiration smears. Results of cell segmentation are compared to a traditional watershed-based color image segmentation approach. Experiences in the use of the developed tool in a clinical workflow are also discussed.

•  Luc Soler (Ircad), “Computed assisted digestive surgery”

•  Roger Tran Son Tay (dept. of Mechanical and Aerospace Engineering University of Florida, USA) Modeling and Role of Leukocytes in Inflammation

Inflammation is a protective response of the body to infection or injury. In some diseases such as atherosclerosis and rheumatoid arthritis, the inflammatory response is triggered inappropriately and the inflammatory cells damage the normal tissues. This suggests using therapeutic strategies in order to interfere with the inflammatory response. In this review, the role of leukocytes and the modeling of the leukocyte behavior in inflammation are presented. Leukocytes play a central role in our immune system response. Upon infection or injury, inflammatory cells are recruited to the inflammation site in order to remove the foreign invaders or start the healing process. Leukocytes go through a series of steps before reaching the inflammation site from inside the blood vessel: initial capture by the endothelium, rolling, firm adhesion, and transendothelial migration. Various adhesion molecules on the surfaces of the leukocyte and endothelium interact with each other in this process. The fluid force imposed by the blood flow is another significant factor in the movement of the leukocyte. There have been numerous theoretical and computational models of leukocyte behavior to complement experimental observations. These models involve a combination of factors including fluid flow, bond force, and the rheological properties of the leukocyte. Modeling techniques used in these models and their results are presented.

•  Nikolaos V. Tsekos (CS@UH) : “MRI-Guided Robotic Interventions: A Multidisciplinary Challenge and Opportunity”

As one of the frontiers of modern medicine, the field of Image Guided Interventions (IGI) is rapidly evolving with an ever-widening range of clinical applications. This evolution is stemmed by a necessity to incorporate new technologies for improved patient care and cost effectiveness. Among such technological advancements are multimodality imaging, from the molecular to the organ level, assistive robotic manipulators, image and vital signs signal processing, haptic sensing, information fusion and comprehensive visualization. One of modalities under development for use with IGI is magnetic resonance imaging (MRI), due to its wide range of soft tissue contrast mechanisms, lack of ionizing radiation, true 3D nature, and on-the-fly adjustment of imaging parameters. Areas of major engineering challenges and potential growth in interventional MRI include those of dynamic imaging and visualization, assistive (robotic) manipulators, the implementation of multi-modality guidance, and image processing and visualization.

We will present dynamic MR imaging approaches that include strategies for the fast selective acquisition of raw MR data, fast MRI preparations for the manipulation of tissue contrast, and multiplanar volumetric imaging. In addition, two MR-compatible robotic manipulators will be presented that offer direct access to the patient inside MRI scanners for real-time guidance. We will also present approaches in the integration of the robotic manipulators with the MR scanner, and current work on image segmentation. Finally, we will overview current needs in multi-modality IGI, real-time image processing, human-machine interfacing and augmented reality visualization of procedures in real-time.

•  Barbara Bass (Dept. of Surgery – The Methodist Hospital) "The Surgical Privileging Process: Is that Surgeon Qualified to do that Operation?"

Surgical privileging is the process by which surgeons are granted the “privilege” of doing a particular surgical procedure in an operating room. Privileges are granted on the basis of years of surgical residency training, and ongoing experience. In recent years, the pace of change in surgical technologies has accelerated at an astonishing pace making it difficult for surgeons to stay fully trained in the latest and best technologies. When is a surgeon adequately skilled to add a new technology or procedure to his or her privileges? This presentation will address the potential role of computational surgery to enhance surgical technological training and as an aid in the privileging process.

..... ......

Dernière mise à jour : 26-Jan-2009 12:35 | webmaster : Ch Collet | Contact | ©2008 Université de Strasbourg, FRANCE |