Table grapes are food products of considerable commercial value for several countries (USA, Brazil, Italy, South Africa, China, Chile, India and Australia are the most important producers). In Europe, Italy ranks first place for table grape production with more than eight million tons per year (ISTAT, 2011). Recently, we developed an innovative analytical method for the characterization of various table grape cultivars. In our study, multivariate statistical analysis applied to 1H NMR data of table grapes, revealed that the inter-vineyard variability of the metabolic profile has a greater discriminating effect over the intra-vineyard one.1 This presentation deals with the effects of several agronomical practices on the metabolic profile of the table grapes during different production stages. The variation of the metabolic features of the grapes was followed by 1H NMR spectroscopy. Moreover, 1H NMR spectra of ripe table grapes were processed to be used as input for expert classification systems based on three different algorithms: J48, Random Forest and an Artificial Neural Network performed with the Error Back Propagation procedure. The performances of the three algorithms in the discrimination of grapes on the bases of some common features (variety, vintage, use of plant growth regulators, trunk girdling, vineyard location) will be shown. References: 1. V. Gallo, P. Mastrorilli, I. Cafagna, G. I. Nitti, M. Latronico, V. A. Romito, A. P. Minoja, C. Napoli, F. Longobardi, H. Schäfer, B. Schütz, M. Spraul, J. Agric. Food Chem. (2012), submitted.

DNA microarray data are used to identify genes which could be considered prognostic markers. However, due to the limited sample size of each study, the signatures are unstable in terms of the composing genes and may be limited in terms of performances. It is therefore of great interest to integrate different studies, thus increasing sample size.Results: In the past, several studies explored the issue of microarray data merging, but the arrival of new techniques and a focus on SVM based classification needed further investigation. We used distant metastasis prediction based on SVM attribute selection and classification to three breast cancer data sets.Conclusions: The results showed that breast cancer classification does not benefit from data merging, confirming the results found by other studies with different techniques.

BMC Bioinformatics 2013, 14(Suppl 8):S5

Introduction and objective: In computer aided diagnosis (CAD) tools searching for colonrectal polyps and based on three dimensions virtual colonoscopy (3DVC) using computed tomography (CT) images, the reduction of the occurrence of false-positives (FPs) still represents a challenge because they are source of unreliability. Following an encouraging previous supervised approach Bevilacqua et al., Three-dimensional Virtual Colonoscopy for Polyps Detection by Supervised Artificial Neural Networks D.-S. Huang et al. (Eds.): ICIC, LNBI 6840, Springer-Verlag, Berlin Heidelberg, (2011), pp. 596-603, the aim of this work is to discuss, in details, how the adopted strategies, designed and tested on an initial reduced data set, reveals good performance and robustness in terms of FPs reduction on an enlarged cohort of new cases. Materials and methods: At the beginning, materials consisted only in 10 different polyps, diagnosed, by expert radiologists, in 6 different patients, scanning 16 rows helical CT multi slices with a resolution of 1. mm. Moreover from those 10 polyps only 7 polyps were initially used for the analysis, excluding 2 tumors with diameter bigger than 1. cm, and one polyp hardly recognizable due to fecal stool. In this paper, thanks to a new accurate phase of collecting data, materials grow impressively and then consist in total of 43 polyps all useful for the study. The whole data set was merged by using the former data set of colonrectal exams from the clinical operative unit called "Sezione di Diagnostica per Immagini" of Di.M.I.M.P. of Policlinico of Bari and the new ones coming from two new collaborations: the Oncology department of Faculty of Medicine of University of Pisa participating, as the former, to the IMPACT study (Italian Multicenter Polyps Accuracy CTC Study) Regge, Linear and nonlinear feedforward neural network classifiers: a comprehensive understanding, J. Intell. Syst., 9 (1) 1999, 1-38 and, more recently, the operative unit of radiology of the "Istituto Tumori Giovanni Paolo II" of Bari. Starting from computed tomography colonography (CTC) images, several volumes were scanned by means of three different supervised artificial neural networks (ANNs) architectures based on error back propagation training algorithm Huang and Ma, Linear and nonlinear feedforward neural network classifiers: a comprehensive understanding, J. Intell. Syst., 9 (1) 1999, 1-38. All the training sets were built by using polyps and non-polyps sub-volume samples, whose dimensions were correlated to the volume of the polyps to be detected. Results: The performance of the best ANN architecture, trained by using a training set of 27 sessile polyps from the new 43 available dataset, were evaluated in terms of FPs and false-negatives and compared to the results shown in Bevilacqua et al., Three-dimensional Virtual Colonoscopy for Polyps Detection by Supervised Artificial Neural Networks D.-S. Huang et al. (Eds.): ICIC, LNBI 6840, Springer-Verlag, Berlin He

The paper addresses feedback control of actuated prostheses based on the Stewart platform parallel mechanism. In such a problem it is essential to apply a feasible numerical method to determine in real time the solution of the forward kinematics, which is highly nonlinear and characterized by analytical indetermination. In this paper, the forward kinematics problem for a human elbow hydraulic prosthesis developed by the research group of Polytechnic of Bari is solved using artificial neural networks as an effective and simple method to obtain in real time the solution of the problem while limiting the computational effort. We show the effectiveness of the technique by designing a PID controller that governs the arm motion thanks to the provided neural computation of the forward kinematics. © 2014 Elsevier B.V.

The paper addresses closed loop control of a hydraulic prosthesis for human elbow. In such a problem it is essential to obtain quick results of simulation in order to appreciate the dynamic behavior of the entire system. In this paper, the forward kinematics problem for a hydraulic prosthesis for human arm developed by the research group of Polytechnic of Bari is solved using artificial neural networks as an effective and simple method to obtain in real time the solution of the problem without an excessive computational effort. We show the effectiveness of the method by designing a PID closed loop control that effectively controls the arm thanks to the provided neural computation of the forward kinematics.

Assessing the shrinking of lung nodules in response to the treatment is an important task to evaluate the effectiveness of the therapy, so guidelines to minimize variability and make this task objective and observer-independent are strongly needed. The Response Evaluation Criteria in Solid Tumors (RECIST) is a set of standardized rules that defines when cancer patients improve, stay the same, or deteriorate during the treatments. An accurate evaluation of the growth rate can be performed only by means of a high resolute segmentation, followed by volumetric techniques, although the higher the resolution, the more time is consumed in the process. In this work we present a framework that carries out a fully automatic 3D segmentation of the human respiratory system in Computer Tomography images. Usually, an analysis of the lungs requires manual segmentation of nodules that is always subjective and often error-prone. So the framework encompasses a second step, which semi-automatically segments the lung nodules, extracting them from the surrounding tissues and providing an accurate data set for estimating several volumetric features. These features can be used for the classification of the malignancy of the nodule.

The occurrence of false-positives (FPs) is still an important concern and source of unreliability in computer-aided diagnosis systems developed for 3D virtual colonoscopy. This work presents three different supervised approaches, based on supervised artificial neural networks (ANNs) architectures tested on 16 rows helical multi-slice computer tomography. The performance of the best ANN architecture developed, by using the volumes belonging to only 4 of 7 available nodules diagnosed by expert radiologists as polyps and non-polyps were evaluated in terms of FPs and false-negatives. It revealed good performance in terms of generalization and FPs reduction, correctly detecting all 7 polyps.

Table grapes classification is an important task in the global market because of the interest of consumers to quality of foodstuff. Objective: an expert and innovative tool, based on several robust classifiers, was designed and implemented to achieve unequivocal criteria and support decision for the discrimination of table grapes. Materials: data are acquired by powerful analytical techniques such as Nuclear Magnetic Resonance (NMR) and are related to 5 attributes: production year, vineyard location, variety, use of plant growth regulators (PGRs) and application of trunk girdling. In particular, datasets consisting of 813 samples regarded the former 3 attributes while datasets based on 596 samples regarded the latter 2 ones. Methods: in absence of an a-priori knowledge, we addressed the problem as an inferential task and then adopted supervised approaches like error back propagation neural networks, trees and random forest classifiers able to manage information from training sets. Experimental Results and Conclusion: our study has shown that the three classifiers, especially that based on a supervised neural network, when applied to NMR data, give from good to excellent performances, depending on the attribute. Such performances pave the way to development of innovative tools for classification of table grapes.

The distance learning (DL) is a teaching system that extends the education beyond the physical barriers, providing access to remote places and disabilities. The increasing need of procedures for DL certification is now involving biometric approach. An analysis of biometric techniques is shown in order to ensure the users authentication, to verify the individual's attention level and then to certificate the learning outcomes. That is necessary to implement a system to identify uniquely the users and to track both path's carried (visited pages) and use's time, to have a secure users identification and also validation of the environments conditions in which they take place during possible tests of certification. The appropriate biometric technique is appeared the Face Recognition because it allows a real-time verification of the real presence, low implementation costs by use of webcam and reasonable degree of reliability. To avoid the influence related to environmental conditions, it has been realized a modular system that implements Detection and Recognition operations. The implemented system is able to verify the presence of learners beyond the screen during lessons or learning tests, to allow authentication and to verify the simultaneous presence of other individuals in order to start an alarm if unregistered peoples are present during learning or testing sessions. This system is also capable to recognize the attention level of users through Request Random Windows (RRW). The application opens casually a RRW in different screen position during the DL and asks learner to click upon to close it within a few seconds. When this window is closed, a new step of Face Recognitions is performed again to validate the presence of the same user. Interesting results are obtained in experimental cases employing these techniques on a individuals samples set.

Introduction and objective: the purpose of this work is to design and implement an innovative tool to recognize 16 different human gestural actions and use them to predict 7 different emotional states. The solution proposed in this paper is based on RGB and depth information of 2D/3D images acquired from a commercial RGB-D sensor called Kinect. Materials: the dataset is a collection of several human actions made by different actors. Each action is performed by each actor for three times in each video. 20 actors perform 16 different actions, both seated and upright, totalling 40 videos per actor. Methods: human gestural actions are recognized by means feature extractions as angles and distances related to joints of human skeleton from RGB and depth images. Emotions are selected according to the state-of-the-art. Experimental results: despite truly similar actions, the overall accuracy reached is approximately 80%. Conclusions and future works: the proposed work seems to be back-ground- and speed independent, and it will be used in the future as part of a multimodal emotion recognition software based on facial expressions and speech analysis as well.

The distance learning (DL) is a teaching system that extends the education beyond the physical barriers, providing access to remote places and disabilities. The increasing need of procedures for DL certification is now involving biometric approach. An analysis of biometric techniques is shown in order to ensure the users authentication, to verify the individual’s attention level and then to certificate the learning outcomes. That is necessary to implement a system to identify uniquely the users and to track both path’s carried (visited pages) and use’s time, to have a secure users identification and also validation of the environments conditions in which they take place during possible tests of certification. The appropriate biometric technique is appeared the Face Recognition because it allows a real-time verification of the real presence, low implementation costs by use of webcam and reasonable degree of reliability. To avoid the influence related to environmental conditions, it has been realized a modular system that implements Detection and Recognition operations. The implemented system is able to verify the presence of learners beyond the screen during lessons or learning tests, to allow authentication and to verify the simultaneous presence of other individuals in order to start an alarm if unregistered peoples are present during learning or testing sessions. This system is also capable to recognize the attention level of users through Request Random Windows (RRW). The application opens casually a RRW in different screen position during the DL and asks learner to click upon to close it within a few seconds. When this window is closed, a new step of Face Recognitions is performed again to validate the presence of the same user. Interesting results are obtained in experimental cases employing these techniques on a individuals samples set.