Johanne Bézy-Wendling

Physiologically based modeling of 3-D vascular networks and CT scan angiography

In this paper, a model-based approach to medical image analysis is presented. It is aimed at understanding the influence of the physiological (related to tissue) and physical (related to image modality) processes underlying the image content. This methodology is exemplified by modeling first, the liver and its vascular network, and second, the standard computed tomography (CT) scan acquisition. After a brief survey on vascular modeling literature, a new method, aimed at the generation of growing three-dimensional vascular structures perfusing the tissue, is described. A solution is proposed in order to avoid intersections among vessels belonging to arterial and/or venous trees, which are physiologically connected. Then it is shown how the propagation of contrast material leads to simulate time-dependent sequences of enhanced liver CT slices.

Toward a better understanding of texture in vascular CT scan simulated images

This paper shows the influence of computed tomography slice thickness on textural parameters by simulating realistic images issued from: (1) a 3D model of vascular tree, with structural and functional features and in which angiogenesis is related to the organ growth; (2) a projection/reconstruction process using fast Fourier transform. Texture analysis is performed by means of second-order statistics and gradient based methods.

Multiscale modeling and imaging: the challenges of biocomplexity

Computational modeling and imaging in biology and medicine are gaining more and more interest with the discovery of in-depth structural and functional knowledge at all space and time scales (molecule to proteins, cells to organs and organisms). The recursion between description levels for highly dynamical, interacting and complex systems, i.e the integrative approach, is a very challenging topic where formal models, observational tools and experimental investigations have to be closely designed, coupled and confronted together. Imaging techniques play a major role in this interdisciplinary attempt to elucidate this biocomplexity: they convey relevant information about the underlying mechanisms, depict the conformations and anatomical topologies and draw the biophysical laws they may follow. Furthermore, the basic image analysis tools (from calibration to segmentation, motion estimation and registration up to pattern recognition) are generic enough to be of value whatever the objects under consideration. The same comments apply when Computer Graphics or Virtual Reality techniques are concerned. This paper will survey the recent contributions dealing with both models, imaging data and processing frames. Examples ranging over different scales, from macro to nano, will be given in order to enhance the mutual benefits and perspectives that can be expected from this coupling.

Simulation of Biphasic CT Findings in Hepatic Cellular Carcinoma by a Two-Level Physiological Model

In this paper, we present a two-level physiological model that is able to reflect morphology and function of vascular networks, in clinical images. Our approach results from the combination of a macroscopic model, providing simulation of the growth and pathological modifications of vascular network, and a microvascular model, based on compartmental approach, which simulates blood and contrast medium transfer through capillary walls. The two-level model is applied to generate biphasic computed tomography of hepatocellular carcinoma. A contrast-enhanced sequence of simulated images is acquired, and enhancement curves extracted from normal and tumoral regions are compared to curves obtained from in vivo images. The model offers the potential of finding early indicators of disease in clinical vascular images

Fast algorithm for 3-D vascular tree modeling

In this short paper, accelerated three-dimensional computer simulations of vascular trees development, preserving physiological and haemodynamic features, are reported. The new computation schemes deal: (i). with the geometrical optimization of each newly created bifurcation; and (ii). with the recalculation of blood pressures and radii of vessels in the whole tree. A significant decrease of the computation time is obtained by replacing the global optimization by the fast updating algorithm allowing more complex structure to be simulated. A comparison between the new algorithms and the previous one is illustrated through the hepatic arterial tree.

Hepatic tumor enhancement in computed tomography: combined models of liver perfusion and dynamic imaging

The objective of this study is to show how computational modeling can be used to increase our understanding of liver enhancement in dynamic computer tomography. It relies on two models: (1). a vascular model, based on physiological rules, is used to generate the 3D hepatic vascular network; (2). the physical process of CT acquisition allows to synthesize timed-stamped series of images, aimed at tracking the propagation of a contrast material through the vessel network and the parenchyma. The coupled models are used to simulate the enhancement of a hyper-vascular tumor at different acquisition times, showing a maximum conspicuity during the arterial phase.

Texture-Based Classification of Hepatic Primary Tumors in Multiphase CT

A new approach to the hepatic primary tumor recognition from dynamic CT images is proposed. In the first step of the proposed method, texture features are extracted from manually traced ROI-s to objectively characterize lesions. The second step consists in applying decision tree classifier. For the first time, the parameters obtained in subsequent acquisition moments are analyzed simultaneously.

Computer-Aided Diagnosis of Liver Tumors Based on Multi-Image Texture Analysis of Contrast-Enhanced CT. Selection of the Most Appropriate Texture Features

Abstract In this work, a system for the classification of liver dynamic contest- enhanced CT images is presented. The system simultaneously analyzes the images with the same slice location, corresponding to three typical acquisition moments (without contrast, arterial- and portal phase of contrast propagation). At first, the texture features are extracted separately for each acquisition mo- ment. Afterwards, they are united in one “multiphase” vector, characterizing a triplet of textures. The work focuses on finding the most appropriate features that characterize a multi-image texture. At the beginning, the features which are unstable and dependent on ROI size are eliminated. Then, a small subset of remaining features is selected in order to guarantee the best possible classification accuracy. In total, 9 extraction methods were used, and 61 features were calculated for each of three acquisition moments. 1511 texture triplets, corresponding to 4 hepatic tissue classes were recognized (hepatocellular carcinoma, cholangiocarcinoma, cirrhotic, and normal). As a classifier, an adaptive boosting algorithm with a C4.5 tree was used. Experiments show that a small set of 12 features is able to ensure classification accuracy exceeding 90%, while all of the 183 features provide an accuracy rate of 88.94%.

Multi-Image Texture Analysis in Classification of Prostatic Tissues from MRI. Preliminary Results

In the work, a (semi)automatic multi-image texture analysis is applied to the characterization of prostatic tissues from Magnetic Resonance Images (MRI). The method consists in a simultaneous analysis of several images, each acquired under different conditions, but representing the same part of the organ. First, the texture of each image is characterized independently of the others, using the same techniques. Afterwards, the feature values corresponding to the different acquisition conditions are combined in one vector, characterizing a multi-image texture. Thus, in the tissue classification process different tissue properties are considered simultaneously. We analyzed three MRI sequences: contrast-enhanced T1-, T2-, and diffusion-weighted one. Two classes of tissue were recognized: cancerous and healthy. Experiments with several sets of textural features and four classification methods showed that the application of multi-image texture analysis could improve the classification accuracy in comparison to single-image texture analysis.

Vascular network modeling: improved parallel implementation on computing cluster

In this paper, an improved parallel algorithm of vascular network modeling is presented. The new solution is based on a more decentralized approach. Moreover, in order to accelerate the simulation of vascular growth process both the dynamic load balancing and periodic rebuildings of vascular trees were introduced. The presented method was implemented on a computing cluster with the use of the MPI standard. The experimental results show that the improved algorithm results in better speedup thus making it possible to introduce more physiological details and also perform simulations with a greater number of vessels and cells. Furthermore, the presented approach can bring the model closer to reality where the analogous vascular processes can be also parallel.