Pascal Haigron

Construction of a complete set of orthogonal Fourier-Mellin moment invariants for pattern recognition applications

The completeness property of a set of invariant descriptors is of fundamental importance from the theoretical as well as the practical points of view. In this paper, we propose a general approach to construct a complete set of orthogonal Fourier-Mellin moment (OFMM) invariants. By establishing a relationship between the OFMMs of the original image and those of the image having the same shape but distinct orientation and scale, a complete set of scale and rotation invariants is derived. The efficiency and the robustness to noise of the method for recognition tasks are shown by comparing it with some existing methods on several data sets.

Image description with generalized pseudo-Zernike moments

A new set, to our knowledge, of orthogonal moment functions for describing images is proposed. It is based on the generalized pseudo-Zernike polynomials that are orthogonal on the unit circle. The generalized pseudo-Zernike polynomials are scaled to ensure numerical stability, and some properties are discussed. The performance of the proposed moments is analyzed in terms of image reconstruction capability and invariant character recognition accuracy. Experimental results demonstrate the superiority of generalized pseudo-Zernike moments compared with pseudo-Zernike and Chebyshev-Fourier moments in both noise-free and noisy conditions.

Transvenous path finding in cardiac resynchronization therapy

Cardiovascular diseases are a major health concern all over the world and, especially, heart failure has gained more importance in the recent years. Improving diagnosis and therapy is therefore critical and among the several resources at our disposal, implantable devices is expected to have a better rate of success. This paper is focused on two topics: (i) our views of the main challenges to face in order to reach these objectives and (ii) a specific target regarding the pose of leads for multisite pacemakers by means of virtual endoscopy pre-operative planning and path finding throughout the coronary venous tree.

Issues in image-guided therapy [A Look at . . .]

M edical robotics, computer- assisted surgery (CAS), image-guided therapy (IGT), and the like emerged more than 20 years ago, and many advances have been made since. Conferences and workshops have been organized; scientific contributions, position papers, and patents have been published; new academic societies have been launched; and companies were created all over the world to propose methods, devices, and systems in the area. Researchers in robotics, computer vision and graphics, electronics, mechanics, biomedical engineers, physicians, and surgeons have been involved, thus demonstrating the enthusiasm for this emerging field. Their commitments emphasize the transdisciplinary nature of the efforts to be made. However, the effective dissemination of CAS-IGT systems in medical disciplines remains limited. There are several reasons that may explain this situation, among which the effective demonstration of patient benefits and cost savings, the reluctance of surgeons or therapists to use them, and, of course, the technological breakthroughs are still expected. This series of articles attempts to point out some of them and will also show that many opportunities in computer-assisted interventions are open in the near future.

Spatial characterization and classification of rectal bleeding in prostate cancer radiotherapy with a voxel-based principal components analysis model for 3D dose distribution

Although external beam radiotherapy is one of the most commonly prescribed treatments for prostate cancer, severe complications may arise as a result of high delivered doses to the neighboring organs at risk, namely the bladder and the rectum. The prediction of this toxicity events are commonly based on the planned dose distribution using the dose-volume histograms within predictive models. However, as different spatial dose distributions may produce similar dose-volume histograms, these models may not be accurate in revealing the subtleties of the dose-effect relationships. Using the prescribed dose, we propose a voxel-based principal component analysis method for characterizing and classifying those individuals at risk of rectal bleeding. Sixty-five cases of patients treated for prostate cancer were reviewed; fifteen of them presented rectal bleeding within two years after the treatment. The method was able to classify rectal bleeding with 0.8 specificity and 0.73 sensitivity. In addition, eigenimages with the most discriminant features suggest that some specific dose patterns are related to rectal bleeding.

Depth map based scene analysis for active navigation

Three dimensional volume data sets provided by CT or MRI allow the user to move virtually around within anatomic structure and to observe them. We propose a new method to guide the path planning based on the local image interpretation during the navigation inside the anatomic structures with cavities, without preprocessing for 3D volume data sets. Thanks to the scene analysis process, the virtual sensor constructs by itself a model of the unknown scene. Qualitative and quantitative characterization of the anatomic structures, derived from this exploration, is of main interest in such a new application of virtual endoscopy that is vascular surgery. In this paper we especially describe the scene analysis process which is based on the processing of the depth map directly produced by the ray casting procedure. Application to vascular surgery, i.e. virtual angioscopy, is then depicted.

Patient-Specific Finite-Element Simulation of the Insertion of Guidewire During an EVAR Procedure: Guidewire Position Prediction Validation on 28 Cases

Objective: Validation of a numerical method to compute arterial deformations under the insertion of an “extra-siff” guidewire during Endovascular Repair of Abdominal Aortic Aneurysm. Methods: We propose the validation of a previously developed simulation method. The model is calibrated using anatomical hypothesis and intraoperative observations. Simulation results are blindly evaluated against 3-D imaging data acquired during the surgical procedure on 28 patients, based on the predicted position of the intraoperative guidewire. Results: Simulation was successfully conducted on the 28 patients. The mean position error given by the Modified Hausdorff Distance for the 28 cases was 3.8 ± 1.9 mm, which demonstrates very good results for most of the cases. Conclusion: The work reported here shows that numerical simulation can predict some rather large variations in the vascular geometry due to tools insertion, for a wide variety of aorto-iliac morphologies. This is a new step toward clinically applicable mechanical simulation. Significance: Validation on 3-D intraoperative data on a large number of cases—robustness on adverse anatomies.

Interindividual registration and dose mapping for voxelwise population analysis of rectal toxicity in prostate cancer radiotherapy.

PURPOSE Recent studies revealed a trend toward voxelwise population analysis in order to understand the local dose/toxicity relationships in prostate cancer radiotherapy. Such approaches require, however, an accurate interindividual mapping of the anatomies and 3D dose distributions toward a common coordinate system. This step is challenging due to the high interindividual variability. In this paper, the authors propose a method designed for interindividual nonrigid registration of the rectum and dose mapping for population analysis. METHODS The method is based on the computation of a normalized structural description of the rectum using a Laplacian-based model. This description takes advantage of the tubular structure of the rectum and its centerline to be embedded in a nonrigid registration-based scheme. The performances of the method were evaluated on 30 individuals treated for prostate cancer in a leave-one-out cross validation. RESULTS Performance was measured using classical metrics (Dice score and Hausdorff distance), along with new metrics devised to better assess dose mapping in relation with structural deformation (dose-organ overlap). Considering these scores, the proposed method outperforms intensity-based and distance maps-based registration methods. CONCLUSIONS The proposed method allows for accurately mapping interindividual 3D dose distributions toward a single anatomical template, opening the way for further voxelwise statistical analysis.

Virtual implantation and patient-specific simulation for optimization of outcomes in ventricular assist device recipients

BACKGROUND Left ventricular assist devices (LVAD) are increasingly used for long-term mechanical circulatory support and are effective in improving survival and quality-of-life of patient with advanced heart failure. Nonetheless, they are associated with significant early and late morbidity rates (including pump thrombosis, thromboembolic events, and pump dysfunction). These complications are at least partially associated with suboptimal pump positioning. Currently, we are missing tools to further improve the positioning of LVAD devices in a patient-specific fashion. METHODS AND EVALUATION OF THE HYPOTHESIS We hypothesized that the analysis of the implanted device in patients presenting selected LVAD-related complications through segmentation and three-dimensional reconstruction of CT scans may provide patient-specific information into mechanical factors contributing to pump dysfunction and thromboembolic events, with potential to guide preventive interventions against development of new complications. We also hypothesized that preoperative virtual implantation and computer-assisted surgery in candidates to LVAD implantation may help in the customization of device positioning, with potential to minimize severe complications. The hypothesis was evaluated in a multidisciplinary fashion (cardiac surgeons, biomedical engineers and biomedical images processing experts). CT scans of 14 LVAD recipients were reconstructed through semi-automatic segmentation (including the whole heart, the implanted device and the chest wall). A coordinate system was built to quantify the coaxiality of the LVAD apical cannula with the mitral annulus. Patients were stratified into Group 1 (presenting complications such as thromboembolic events, pump dysfunction or thrombosis) and Group 2 (no complications). Group 1 patients presented significantly greater average rotation of the apical cannula towards the interventricular septum (p=0.015), although no difference was observed in terms of average rotation towards the anterior or posterior left ventricular wall. CONCLUSIONS Several patient-specific factors (including left ventricular morphology and chest wall conflict with the device after wound closure) may influence the effectiveness and safety of LVAD therapy, but they are difficultly managed through the current implantation techniques. We suggest that the clinical results of LVAD treatment can be improved through preoperative virtual implantation and computer-assisted surgery (in order to guide device selection, exact site of left ventricular wall coring site). Given these preliminary results, we are examining larger patient datasets in order to further test the hypothesis. Dedicated tools for virtual implantation are currently under development.

Spatial nonparametric mixed-effects model with spatial-varying coefficients for analysis of population

Voxel-wise comparisons have been largely used in the analysis of populations to identify biomarkers for pathologies, disease progression, or to predict a treatment outcome. On the basis of a good interindividual spatial alignment, 3D maps are produced, allowing to localise regions where significant differences between groups exist. However, these techniques have received some criticism as they rely on conditions wich are not always met. Firstly, the results may be affected by misregistrations; secondly, the statistics behind the models assumes normally distributed data; finally, because of the size of the images, some strategies must be used to control for the rate of false detection. In this paper, we propose a spatial (3D) nonparametric mixed-effects model for population analysis. It overcomes some of the issues of classical voxel-based approaches, namely robustness to false positive rates, misregistrations and large variances between groups. Examples on numerical phantoms and real clinical data illustrate the feasiblity of the approach. An example of application within the development of voxel-wise predictive models of rectal toxicity in prostate cancer radiotherapy is presented. Results demonstrate an improved sensitivity and reliability for group analysis compared with standard voxel-wise methods and open the way for potential applications in computational anatomy.