Catálogo de publicaciones - libros

Purpose of this work is the development of an automatic system which can be useful for radiologists in the investigation of breast and lung cancer. A breast neoplasia is often marked by the presence of microcalcifications and massive lesions in the mammogram. The first are a very small object in a noise background and the second are large object with particular shape. The need for tools able to recognize such lesions at an early stage is therefore apparent. In this article is shown an application of artificial neural network on the imaging analysis in mammography. The results obtained in terms of sensitivity and specificity when it has been tested alone and then used as second reader will be presented. We present also an overview about the methods developed for pulmonary nodule detection in CT images and the preliminary results obtained with a pre-processing filter will be also presented.

This paper proposes a new neural network approach to the classification of vehicles in image sequences recorded by a stationary camera. The novelty consists in organizing the tracking data into directed acyclic graphs and in the use of recursive neural networks to discriminate which vehicle is represented in each graph. Some preliminary experimental results from real-world traffic scenes prove the viability of the method.

In this paper we propose a neural network identification of a mathematical model called MINMOD, which describes the interactions between glucose and insulin in human subjects, in order to realize an adequate model for patients suffering from Type 2. The model has been tested on the basis of clinical data and it can correctly reproduce glucose and insulin reply and temporal evolution, according to experimental data test. Using neural networks, we can predict the glucose temporal evolution without invasive technique for patients, with the aim to determine the clinical effects to be made in case of pathological behaviors.

The reliability of communication networks is assessed by employing two machine learning algorithms, Support Vector Machines (SVM) and Hamming Clustering (HC), acting on a subset of possible system configurations, generated by a Monte Carlo simulation and an appropriate Evaluation Function. The experiments performed with two different reliability measures show that both methods yield excellent predictions, though the performances of models generated by HC are significantly better than those of SVM.

Disruption is a sudden loss of magnetic confinement that can cause a damage of the machine walls and support structures. For this reason is of practical interest to be able to early detect the onset of the event. This paper presents a novel technique of early prediction of plasma disruption in Tokamak reactors which uses Neural Networks and Chaos theory. In particular, dynamical reconstruction and chaos theory have been considered for choosing the time window of prediction and to select the inputs set for the prediction system. Multi-Layer-Perceptron nets have been exploited for predicting the incoming of disruption.