Abstract
Breast Cancer Conservative Treatment (BCCT) is considered nowadays to be the most widespread form of locor-regional breast cancer treatment. However, aesthetic results are heterogeneous and difficult to evaluate in a standardized way. The limited reproducibility of subjective aesthetic evaluation in BCCT motivated the research towards objective methods. A recent computer system (BCCT. core) was developed to objectively and automatically evaluate the aesthetic result of BCCT. The system is centered on a support vector machine (SVM) classifier with a radial basis function (RBF) used to predict the overall cosmetic result from features computed on a digital photograph of the patient. However, this classifier is not ideal for the interpretation of the factors being used in the prediction. Therefore, an often suggested improvement is the interpretability of the model being used to assess the overall aesthetic result. In the current work we investigate the accuracy of different interpretable methods against the model currently deployed in the BCCT. core software. We compare the performance of decision trees and linear classifiers with the RBF SVM currently in BCCT. core. In the experimental study, these interpretable models shown a similar accuracy to the currently used RBF SVM, suggesting that the later can be replaced without sacrificing the performance of the BCCT.core.

Abstract

Abstract
The correct estimation of the friction coefficient in automotive applications is of paramount importance in the design of effective vehicle safety systems. In this article a new parametrization for estimating the peak friction coefficient, in the tire-road interface, is presented. The proposed parametrization is based on a feedforward neural network (FFNN), trained by the Extreme Learning Machine (ELM) method. Unlike traditional learning techniques for FFNN, typically based on backpropagation and inappropriate for real time implementation, the ELM provides a learning process based on random assignment in the weights between input and the hidden layer. With this approach, the network training becomes much faster, and the unknown parameters can be identified through simple and robust regression methods, such as the Recursive Least Squares. Simulation results, obtained with the CarSim program, demonstrate a good performance of the proposed parametrization; compared with previous methods described in the literature, the proposed method reduces the estimation errors using a model with a lower number of parameters.

Abstract
Breast Cancer Conservative Treatment (BCCT) is considered the gold standard of breast cancer treatment. However, the aesthetic outcome is diverse and very difficult to evaluate in a consistent way partly due to the weak reproducibility of the subjective methods in use. T his motivated the research on the objective methods. BCCT.core is a very recent software that objectively and automatically evaluates the aesthetic outcome of BCCT. However, as in other approaches, the system only uses frontal patient information, disregarding volumetric perception on lateral measurements. In the current work we investigate the improvement of the BCCT.core model by introducing lateral information extracted from patients images. We compare the performance of the model currently used on BCCT.core with the model developed in this study. Experimental results suggest that with lateral measurements the model presents better performance, however improvements are not significant. We can conclude that is essential to use robust models on the BCCT, and the input of 3D models will probably help to obtain better results. © 2010 IEEE.

Abstract
Automatic image annotation or image classification can be an important step when searching for images from a database. Common approaches to medical image annotation with the Image Retrieval for Medical Applications (IRMA) code make poor or no use of its hierarchical nature, where different dense sampled pixel based information methods outperform global image descriptors. In this work we address the problem of hierarchical medical image annotation by building a Content Based Image Retrieval (CBIR) system aiming to explore the combination of three different methods using Support Vector Machines (SVMs): first we concatenate global image descriptors with an interest points Bag-of-Words (BoW) to build a feature vector; second, we perform an initial annotation of the data using two known methods, disregarding the hierarchy of the IRMA code, and a third that takes the hierarchy into consideration by classifying consecutively its instances; finally, we make use of pairwise majority voting between methods by simply summing strings in order to produce a final annotation. Our results show that although almost all fusion methods result in an improvement over standalone classifications, none clearly outperforms each other. Nevertheless, these are quite competitive when compared with related works using an identical database. © 2010 IEEE.

Abstract
A simple and efficient hybrid framework for evaluating algorithms for tracking objects in video sequences is presented. The framework unifies state-of-the-art evaluation metrics with diverse requirements in terms of reference information, thus overcoming weaknesses of individual approaches. With foundations on already demonstrated and well known metrics, this framework assumes the role of a flexible and powerful tool for the research community to assess and compare algorithms.