Anderson Alvarenga de Moura Meneses

Artificial neural networks applied to bone recognition in X-Ray computer microtomography imaging for histomorphometric analysis

Bone Histomorphometry is an important analysis in preventing and treatment of cancer and osteoporosis patients, providing quantitative information about the bone structure. X-Ray Micro-Computer Tomography is a non-invasive and non-destructive imaging technique, with a high space resolution that enables magnified images. In the histomorphometric analysis of such images, it is possible to use filters and binarization, nevertheless these techniques may cause loss of information. In this paper we describe the usage of Artificial Neural Networks (ANNs) in Microtomography X-Ray imaging bone recognition as a part of a histomorphometric analysis research with raw images obtained at the Synchrotron Radiation for Medical Physics (SYRMEP) beamline of the ELETTRA Laboratory at Trieste, Italy. A Multilayer Perceptron Model for the ANNs with Error Back-Propagation and supervised learning has been used in the recognition task. The classification of bone subimages yielded a Receiver Operating Characteristic Curve with an area under curve of 1.000, which means that the ANN is able to distinguish successfully the bone mass. The images obtained are also depicted herein. The quality and characteristics of the X-Ray Computer Microtomography are compatible with the ANN-based proposed methodology, avoiding the loss of information due to image manipulation.

Particle swarm optimisation applied to nuclear engineering problems

Evolutionary computation (EC) techniques, and more specifically genetic algorithms (GA) and their variations, have been efficiently applied to many complex problems found in the nuclear engineering field. Such methods have been shown to be robust and efficient, but highly time consuming. Other population-based methods have been proposed as alternatives to these traditional EC techniques. The Particle Swarm Optimisation (PSO) technique has been shown to be faster and many times more efficient than GA. Motivated by that, investigations concerning applications of PSO to nuclear engineering problems have started in the Brazilian Nuclear Engineering Institute (IEN/CNEN) and Federal University of Rio de Janeiro (UFRJ). This paper describes applications of PSO to four classical nuclear engineering problems: (i) nuclear fuel reload, (ii) core design optimisation, (iii) surveillance tests planning and (iv) accident classification. Computational experiments demonstrate that PSO can be efficiently applied to the problems studied. Moreover, the results described are comparable with, or even better than, some good results (obtained by GA) found in the literature.

Neural computing for quantitative analysis of human bone trabecular structures in synchrotron radiation X-Ray μCT images

Prevention and treatment of Osteoporosis in elderly patients is critical and important since this disease became a major public health problem. It is well known the fact that osteoporotic fractures may occur as a result of a combination of the degeneration of trabecular structures and low bone mass. Therefore, the quantitative analysis of human bone trabecular architecture might be useful for treatment and diagnosis of this disease. Synchrotron Radiation X-Ray micro-Computed Tomography (μCT) enables magnified images with a high space resolution that allows detailed analysis of the trabecular structure. In the quantitative analysis of medical images of human bone, it is necessary to use filters and binarization, nevertheless these techniques may cause loss of information. This paper describes the alternative application of Neural Computing (Artificial Neural Networks) to the pixel classification in order perform the quantitative analysis of human bone trabecular structure in Synchrotron Radiation μCT images obtained at the Synchrotron Radiation for Medical Physics (SYRMEP) beam line of the ELETTRA Laboratory at Trieste, Italy. Results demonstrate that, despite the complexity of the trabecular architecture, the ANNs have considerable success in the recognition of bone pixels for the quantitative analysis and that its use is compatible to the characteristics of Synchrotron Radiation images.

Optimization of Nuclear Reactors Loading Patterns with Computational Intelligence Methods

The goal of the Loading Pattern (LP) optimization problem is to determine an optimal (or near-optimal) distribution of Fuel Assemblies of a Nuclear Reactor for producing full power within adequate safety margins. Also known as In-Core Fuel Management Optimization, the LP optimization is a prominent real-world problem in Nuclear Engineering with high complexity due to its combinatorial formulation with a large number of feasible solutions, a large number of sub-optimal solutions, disconnected feasible regions, high dimensionality, complex and time-consuming evaluation functions with Reactor Physics calculations. In the present chapter, we discuss LP optimization problem and four computational intelligence optimization methods, also known as optimization metaheuristics or generic heuristic methods, namely the Cross-Entropy algorithm, the Particle Swarm Optimization, Artificial Bee Colonies, and Population-Based Incremental Learning. Results using actual models are described and also discussed.

Study of confinements in the particle swarm optimisation for application to the nuclear reactor reload problem

Particle Swarm Optimisation (PSO) is a metaheuristic technique based on the social aspects of intelligence. Some PSO models have been developed for combinatorial optimisation, although none of them presented satisfactory results to optimise the combinatorial problem of the Nuclear Reactor Reload Problem (NRRP). The Particle Swarm Optimisation with Random Keys (PSORK) model is a variant of the PSO applied to combinatorial problems such as the NRRP. In this paper, we present the results of the in-core fuel optimisation of the Angra 1 Nuclear Power Plant (NPP) located at the southeast of Brazil and a survey on the confinement of particles in the PSORK for the NRRP. A confinement analysis is interesting in a continuous function optimisation since it may influence the search, resulting in biases that favour particular regions of the search space. Nevertheless, there are no similar studies for confinements applied to a combinatorial optimisation. We have submitted the PSORK to a confinement analysis when applied to the Travelling Salesman Problem (TSP) Rykel48 (ry48p), a benchmark for a combinatorial optimisation, in order to study the consequences of the confined PSORK in this type of combinatorial problem. Finally, the Confinement for the Combinatorial Optimisation (CCO) of the NRRP is proposed and the results are presented.