Salah Maouche

Decision support system for urban transportation networks

This paper deals with the real-time regulation of traffic within a disrupted transportation system. We outline the necessity of a decision support system that detects, analyzes, and resolves the unpredicted disturbances. Due to the distributed aspects of transportation systems, we present a multi-agent approach for the regulation process. Moreover, this approach also includes an evolutionary algorithm that is based on an original genetic coding representing the decisions on a set of vehicles and stops affected by the disturbance. This set constitutes, in fact, the space-time horizon of the regulation process. The evolutionary algorithm then treats the regulation problem as an optimization and provides the regulator with relevant decisions that can result in a partial reconfiguration of the network.

Local fractal and multifractal features for volumic texture characterization

For texture analysis, several features such as co-occurrence matrices, Gabor filters and the wavelet transform are used. Recently, fractal geometry appeared to be an effective feature to analyze texture. But it is often restricted to 2D images, while 3D information can be very important especially in medical image processing. Moreover applications are limited to the use of fractal dimension. This study focuses on the benefits of fractal geometry in a classification method based on volumic texture analysis. The proposed methods make use of fractal and multifractal features for a 3D texture analysis of a voxel neighborhood. They are validated with synthetic data before being applied on real images. Their efficiencies are proved by comparison to some other texture features in supervised classification processes (AdaBoost and support vector machine classifiers). The results showed that features based on fractal geometry (by combining fractal and multifractal features) contributed to new texture characterization. Information on new features was useful and complementary for a classification method. This study suggests that fractal geometry can provide a new useful information in 3D texture analysis, especially in medical imaging.

Conformal radiotherapy optimization with micromultileaf collimators: comparison with radiosurgery techniques

PURPOSE Conformal radiotherapy (CRT) consists of irradiating the target volume while avoiding the healthy peripheral tissues and organs at risk as far as possible. One technique used to treat intracranial tumors consists of using micromultileaf collimators (MMLCs). Given the dose constraints involved, it is of interest to optimize MMLC irradiation parameters and compare the results of this technique with those of conventional radiosurgery (RT) techniques (Gamma Knife and linear accelerator stereotactic RT). METHODS AND MATERIALS MMLC protocols are optimized in two stages. The orientation of the fields, delimited by a beams eye view technique, is determined using a genetic algorithm method. The weighting of the fields and subfields when using intensity modulation and the position of the leaves are optimized using a simulated annealing method. We compared the results obtained for 8 clinical cases using 5 intensity-modulated fields with those obtained using the two radiosurgery techniques. The comparison indexes are those defined by the Radiation Therapy Oncology Group (RTOG). RESULTS The results of this study demonstrated the advantages of using intensity modulation and the improvement obtained for the RTOG indexes in the case of CRT with MMLC, although the healthy peripheral tissues were less exposed to radiation with the radiosurgery techniques. The results also highlight the difficulty encountered with radiosurgery techniques in obtaining satisfactory dose homogeneity when the protocol is defined with numerous iosocenters. CONCLUSION In CRT with MMLC, intensity modulation makes it possible to reduce the number of fields used. It is especially useful to optimize the orientations in the case of target volumes of complex shape or when volumes at risk are in the vicinity of the target. If used correctly, MMLC can be a valuable alternative to conventional radiosurgery techniques.

AUDyC Neural Network using a new Gaussian Densities Merge Mechanism

In the context of evolutionary data classification, dynamical modeling techniques are useful to continuously learn clusters models. Dedicated to on-line clustering, the AUDyC (Auto-adaptive and Dynamical Clustering) algorithm is an unsupervised neural network with auto-adaptive abilities in nonstationary environment. These particular abilities are based on specific learning rules that are developed into three stages: “Classification”, “Evaluation” and “Fusion”. In this paper, we propose a new densities merge mechanism to improve the “Fusion” stage in order to avoid some local optima drawbacks of Gaussian fitting. The novelty of our approach is to use an ambiguity rule of fuzzy modelling with new merge acceptance criteria. Our approach can be generalized to any type of fuzzy classification method using Gaussian models. Some experiments are presented to show the efficiency of our approach to circumvent to AUDyC NN local optima problems.

Needle positioning in interventional MRI procedure: real time optical localisation and accordance with the roadmap

This study presents a system designed to assist the surgeon during interventional procedures performed by magnetic resonance imaging (MRI). In order to reach the target during guidance in a double obliquity trajectory, this system provides accurate information about both the entry point and the orientation of the needle.

SAKM: Self-adaptive kernel machine A kernel-based algorithm for online clustering

This paper presents a new online clustering algorithm called SAKM (Self-Adaptive Kernel Machine) which is developed to learn continuously evolving clusters from non-stationary data. Based on SVM and kernel methods, the SAKM algorithm uses a fast adaptive learning procedure to take into account variations over time. Dedicated to online clustering in a multi-class environment, the algorithm designs an unsupervised neural architecture with self-adaptive abilities. Based on a specific kernel-induced similarity measure, the SAKM learning procedures consist of four main stages: Creation, Adaptation, Fusion and Elimination. In addition to these properties, the SAKM algorithm is attractive to be computationally efficient in online learning of real-drifting targets. After a theoretical study of the error convergence bound of the SAKM local learning, a comparison with NORMA and ALMA algorithms is made. In the end, some experiments conducted on simulation data, UCI benchmarks and real data are given to illustrate the capacities of the SAKM algorithm for online clustering in non-stationary and multi-class environment.

Urban bus traffic regulation by evolutionary algorithms

The most important disadvantage of urban bus networks (UBN) is their close dependence on diverse random phenomena that can alter the initial preestablished schedules of the buses. In fact, any changes occurring in the global urban traffic, in the demand, in the equipment or in the stuff, may certainly cause disturbances within the network traffic. Moreover, these disturbances affect the waiting time of the customers at the stops, the transit operations and also the duration of the different trips on the buses. This article deals with the traffic regulation in an UBN, especially with the transit operations. After introducing the considered disturbances, we state the different criteria that have to be optimized. The regulation approach by evolutionary algorithms is then exposed and some simulation results are presented.

Fractal features for localization of temporal lobe epileptic foci using SPECT imaging

Single photon emission computed tomography (SPECT) is an accurate imaging method for the diagnosis of refractory partial epilepsy. Two scans are carried out: interictal and ictal. The interest of this method is to provide an image in the ictal period, which allows hyperperfused areas linked to the seizure to be localized. The epileptic foci localization is improved by subtracting the two acquisitions (subtracted ictal SPECT: SIS). In some cases, the SIS method is not effective and does not isolate the seizure foci. In this article, we investigate a new method based on texture analysis using fractal geometry features. Fractal geometry features were extracted from each scan in order to quantify the heterogeneity change resulting from the hyperperfusion. A support vector machine (SVM) classification algorithm was used to classify the voxels into two classes: focal and healthy. Quantitative evaluation was performed on simulated images and clinical images from 22 patients with temporal lobe epilepsy. Results on both experiments showed that the proposed method is more specific and more sensitive than the SIS method.

Feature Selection using an SVM learning machine

In this paper we suggest an approach to select features for the Support Vector Machines (SVM). Feature selection is efficient in searching the most descriptive features which would contribute in increasing the effectiveness of the classifier algorithm. The process described here consists in backward elimination strategy based on the criterion of the rate of misclassification. We used the tabu algorithm to guide the search of the optimal set of features; each set of features is assessed according to its goodness of fit. This procedure is exploited in the regulation of urban transport network systems. It was first applied in a binary case and then it was extended to the multiclass case thanks to the MSVM technique: Binary Tree.

Automatic segmentation of prostate boundaries from abdominal ultrasound images using priori knowledge

This article presents a method for automatic segmentation of prostate from abdominal freehand ultrasound images. A statistical model of prostate is estimated from manually delineated images. The segmentation starts by smoothing the image to enhance edges by applying a modified version of the adaptive filter which detects individual speckles and remove them, while it preserves valuable details. Then the boundary is initialized starting from the model and a simulated annealing optimization algorithm seeks the final form. The performances of the algorithm were compared with manual segmentation, the average distance was 3.7 pixels with a standard deviation of 2.3.