Modeling and Mapping Agroforestry Aboveground Biomass in the Brazilian Amazon Using Airborne Lidar Data.

Agroforestry has large potential for carbon (C) sequestration while providing many economical, social, and ecological benefits via its diversified products. Airborne lidar is considered as the most accurate technology for mapping aboveground biomass (AGB) over landscape levels. However, little research in the past has been done to study AGB of agroforestry systems using airborne lidar data. Focusing on an agroforestry system in the Brazilian Amazon, this study first predicted plot-level AGB using fixed-effects regression models that assumed the regression coefficients to be constants. The model prediction errors were then analyzed from the perspectives of tree DBH (diameter at breast height)?height relationships and plot-level wood density, which suggested the need for stratifying agroforestry fields to improve plot-level AGB modeling. We separated teak plantations from other agroforestry types and predicted AGB using mixed-effects models that can incorporate the variation of AGB-height relationship across agroforestry types. We found that, at the plot scale, mixed-effects models led to better model prediction performance (based on leave-one-out cross-validation) than the fixed-effects models, with the coefficient of determination (R2) increasing from 0.38 to 0.64. At the landscape level, the difference between AGB densities from the two types of models was ~10% on average and up to ~30% at the pixel level. This study suggested the importance of stratification based on tree AGB allometry and the utility of mixed-effects models in modeling and mapping AGB of agroforestry systems.

Analytical approaches for the measurement of sulfide in biological and pharmaceutical fields

Hydrogen sulfide (H2S) is a widely recognized gasotransmitter, with key roles in physiological and pathological processes. The accurate quantification of H2S and reactive sulfur species (RSS) may hold important implications for the diagnosis and prognosis of various diseases. However, H2S species quantification in biological matrices is still a challenge. Among the sulfide detection methods, monobromobimane (MBB) derivatization coupled with reversed phase high-performance liquid chromatography (RP-HPLC) is one of the most reported. However, it is characterized by a complex preparation and time-consuming process, which may alter the actual H2S level. Moreover, quantitative validation has still not been described based on a survey of previously published works. In this study, we developed and validated an improved analytical protocol for the MBB RP-HPLC method. Main parameters like MBB concentration, temperature, reaction time, and sample handling were optimized, and the calibration method was further validated using leave-one-out cross-validation (CV) and tested in a clinical setting. The method shows high sensitivity and allows the quantification of H2S species, with a limit of detection (LOD) of 0.5 µM and a limit of quantification (LOQ) of 0.9 µM. Additionally, this model was successfully applied in measurements of H2S levels in the serum of patients subjected to inhalation with vapors rich in H2S. In addition, a properly procedure was established for H2S release with the modified MBB HPLC-FLD method. The proposed analytical approach demonstrated the slow-release kinetics of H2S from the multilayer Silk-Fibroin scaffolds with the combination of different H2S donor’s concentration with respect to the weight of PLGA nanofiber. In the end, some efforts were made on sulfide measurements by using size exclusion chromatography fluorescence/ultraviolet detection and inductively coupled plasma-mass spectrometry (SEC-FLD/UV-ICP/MS). It’s intended as a preliminary study in order to define the feasibility of a separation-detection-quantification platform to analyze biological samples and quantify sulfur species.

Developing a radiomic based machine learning model to identify patients with resected non-small-cell lung cancer at high risk of relapse

Background There is a wide variation of recurrence risk of Non-small-cell lung cancer (NSCLC) within the same Tumor Node Metastasis (TNM) stage, suggesting that other parameters are involved in determining this probability. Radiomics allows extraction of quantitative information from images that can be used for clinical purposes. The primary objective of this study is to develop a radiomic prognostic model that predicts a 3 year disease free-survival (DFS) of resected Early Stage (ES) NSCLC patients. Material and Methods 56 pre-surgery non contrast Computed Tomography (CT) scans were retrieved from the PACS of our institution and anonymized. Then they were automatically segmented with an open access deep learning pipeline and reviewed by an experienced radiologist to obtain 3D masks of the NSCLC. Images and masks underwent to resampling normalization and discretization. From the masks hundreds Radiomic Features (RF) were extracted using Py-Radiomics. Hence, RF were reduced to select the most representative features. The remaining RF were used in combination with Clinical parameters to build a DFS prediction model using Leave-one-out cross-validation (LOOCV) with Random Forest. Results and Conclusion A poor agreement between the radiologist and the automatic segmentation algorithm (DICE score of 0.37) was found. Therefore, another experienced radiologist manually segmented the lesions and only stable and reproducible RF were kept. 50 RF demonstrated a high correlation with the DFS but only one was confirmed when clinicopathological covariates were added: Busyness a Neighbouring Gray Tone Difference Matrix (HR 9.610). 16 clinical variables (which comprised TNM) were used to build the LOOCV model demonstrating a higher Area Under the Curve (AUC) when RF were included in the analysis (0.67 vs 0.60) but the difference was not statistically significant (p=0,5147).

Selection bias and the cross-validation of regression models for prediction

Strategies are compared for the development of a linear regression model with stochastic (multivariate normal) regressor variables and the subsequent assessment of its predictive ability. Bias and mean squared error of four estimators of predictive performance are evaluated in simulated samples of 32 population correlation matrices. Models including all of the available predictors are compared with those obtained using selected subsets. The subset selection procedures investigated include two stopping rules, C$\sb{\rm p}$ and S$\sb{\rm p}$, each combined with an 'all possible subsets' or 'forward selection' of variables. The estimators of performance utilized include parametric (MSEP$\sb{\rm m}$) and non-parametric (PRESS) assessments in the entire sample, and two data splitting estimates restricted to a random or balanced (Snee's DUPLEX) 'validation' half sample. The simulations were performed as a designed experiment, with population correlation matrices representing a broad range of data structures.^ The techniques examined for subset selection do not generally result in improved predictions relative to the full model. Approaches using 'forward selection' result in slightly smaller prediction errors and less biased estimators of predictive accuracy than 'all possible subsets' approaches but no differences are detected between the performances of C$\sb{\rm p}$ and S$\sb{\rm p}$. In every case, prediction errors of models obtained by subset selection in either of the half splits exceed those obtained using all predictors and the entire sample.^ Only the random split estimator is conditionally (on $\\beta$) unbiased, however MSEP$\sb{\rm m}$ is unbiased on average and PRESS is nearly so in unselected (fixed form) models. When subset selection techniques are used, MSEP$\sb{\rm m}$ and PRESS always underestimate prediction errors, by as much as 27 percent (on average) in small samples. Despite their bias, the mean squared errors (MSE) of these estimators are at least 30 percent less than that of the unbiased random split estimator. The DUPLEX split estimator suffers from large MSE as well as bias, and seems of little value within the context of stochastic regressor variables.^ To maximize predictive accuracy while retaining a reliable estimate of that accuracy, it is recommended that the entire sample be used for model development, and a leave-one-out statistic (e.g. PRESS) be used for assessment. ^

Prediction of cross-recognition of peptide-HLA A2 by Melan-A-specific cytotoxic T lymphocytes using three-dimensional quantitative structure-activity relationships.

The cross-recognition of peptides by cytotoxic T lymphocytes is a key element in immunology and in particular in peptide based immunotherapy. Here we develop three-dimensional (3D) quantitative structure-activity relationships (QSARs) to predict cross-recognition by Melan-A-specific cytotoxic T lymphocytes of peptides bound to HLA A*0201 (hereafter referred to as HLA A2). First, we predict the structure of a set of self- and pathogen-derived peptides bound to HLA A2 using a previously developed ab initio structure prediction approach [Fagerberg et al., J. Mol. Biol., 521-46 (2006)]. Second, shape and electrostatic energy calculations are performed on a 3D grid to produce similarity matrices which are combined with a genetic neural network method [So et al., J. Med. Chem., 4347-59 (1997)] to generate 3D-QSAR models. The models are extensively validated using several different approaches. During the model generation, the leave-one-out cross-validated correlation coefficient (q (2)) is used as the fitness criterion and all obtained models are evaluated based on their q (2) values. Moreover, the best model obtained for a partitioned data set is evaluated by its correlation coefficient (r = 0.92 for the external test set). The physical relevance of all models is tested using a functional dependence analysis and the robustness of the models obtained for the entire data set is confirmed using y-randomization. Finally, the validated models are tested for their utility in the setting of rational peptide design: their ability to discriminate between peptides that only contain side chain substitutions in a single secondary anchor position is evaluated. In addition, the predicted cross-recognition of the mono-substituted peptides is confirmed experimentally in chromium-release assays. These results underline the utility of 3D-QSARs in peptide mimetic design and suggest that the properties of the unbound epitope are sufficient to capture most of the information to determine the cross-recognition.

Kernel-Based Ranking. Methods for Learning and Performance Estimation

Machine learning provides tools for automated construction of predictive models in data intensive areas of engineering and science. The family of regularized kernel methods have in the recent years become one of the mainstream approaches to machine learning, due to a number of advantages the methods share. The approach provides theoretically well-founded solutions to the problems of under- and overfitting, allows learning from structured data, and has been empirically demonstrated to yield high predictive performance on a wide range of application domains. Historically, the problems of classification and regression have gained the majority of attention in the field. In this thesis we focus on another type of learning problem, that of learning to rank. In learning to rank, the aim is from a set of past observations to learn a ranking function that can order new objects according to how well they match some underlying criterion of goodness. As an important special case of the setting, we can recover the bipartite ranking problem, corresponding to maximizing the area under the ROC curve (AUC) in binary classification. Ranking applications appear in a large variety of settings, examples encountered in this thesis include document retrieval in web search, recommender systems, information extraction and automated parsing of natural language. We consider the pairwise approach to learning to rank, where ranking models are learned by minimizing the expected probability of ranking any two randomly drawn test examples incorrectly. The development of computationally efficient kernel methods, based on this approach, has in the past proven to be challenging. Moreover, it is not clear what techniques for estimating the predictive performance of learned models are the most reliable in the ranking setting, and how the techniques can be implemented efficiently. The contributions of this thesis are as follows. First, we develop RankRLS, a computationally efficient kernel method for learning to rank, that is based on minimizing a regularized pairwise least-squares loss. In addition to training methods, we introduce a variety of algorithms for tasks such as model selection, multi-output learning, and cross-validation, based on computational shortcuts from matrix algebra. Second, we improve the fastest known training method for the linear version of the RankSVM algorithm, which is one of the most well established methods for learning to rank. Third, we study the combination of the empirical kernel map and reduced set approximation, which allows the large-scale training of kernel machines using linear solvers, and propose computationally efficient solutions to cross-validation when using the approach. Next, we explore the problem of reliable cross-validation when using AUC as a performance criterion, through an extensive simulation study. We demonstrate that the proposed leave-pair-out cross-validation approach leads to more reliable performance estimation than commonly used alternative approaches. Finally, we present a case study on applying machine learning to information extraction from biomedical literature, which combines several of the approaches considered in the thesis. The thesis is divided into two parts. Part I provides the background for the research work and summarizes the most central results, Part II consists of the five original research articles that are the main contribution of this thesis.

Structure- and Ligand-Based Structure-Activity Relationships for a Series of Inhibitors of Aldolase

Aldolase has emerged as a promising molecular target for the treatment of human African trypanosomiasis. Over the last years, due to the increasing number of patients infected with Trypanosoma brucei, there is an urgent need for new drugs to treat this neglected disease. In the present study, two-dimensional fragment-based quantitative-structure activity relationship (QSAR) models were generated for a series of inhibitors of aldolase. Through the application of leave-one-out and leave-many-out cross-validation procedures, significant correlation coefficients were obtained (r(2) = 0.98 and q(2) = 0.77) as an indication of the statistical internal and external consistency of the models. The best model was employed to predict pK(i) values for a series of test set compounds, and the predicted values were in good agreement with the experimental results, showing the power of the model for untested compounds. Moreover, structure-based molecular modeling studies were performed to investigate the binding mode of the inhibitors in the active site of the parasitic target enzyme. The structural and QSAR results provided useful molecular information for the design of new aldolase inhibitors within this structural class.

Méthodes de Bootstrap pour les modèles à facteurs

Cette thèse développe des méthodes bootstrap pour les modèles à facteurs qui sont couram- ment utilisés pour générer des prévisions depuis l'article pionnier de Stock et Watson (2002) sur les indices de diffusion. Ces modèles tolèrent l'inclusion d'un grand nombre de variables macroéconomiques et financières comme prédicteurs, une caractéristique utile pour inclure di- verses informations disponibles aux agents économiques. Ma thèse propose donc des outils éco- nométriques qui améliorent l'inférence dans les modèles à facteurs utilisant des facteurs latents extraits d'un large panel de prédicteurs observés. Il est subdivisé en trois chapitres complémen- taires dont les deux premiers en collaboration avec Sílvia Gonçalves et Benoit Perron. Dans le premier article, nous étudions comment les méthodes bootstrap peuvent être utilisées pour faire de l'inférence dans les modèles de prévision pour un horizon de h périodes dans le futur. Pour ce faire, il examine l'inférence bootstrap dans un contexte de régression augmentée de facteurs où les erreurs pourraient être autocorrélées. Il généralise les résultats de Gonçalves et Perron (2014) et propose puis justifie deux approches basées sur les résidus : le block wild bootstrap et le dependent wild bootstrap. Nos simulations montrent une amélioration des taux de couverture des intervalles de confiance des coefficients estimés en utilisant ces approches comparativement à la théorie asymptotique et au wild bootstrap en présence de corrélation sérielle dans les erreurs de régression. Le deuxième chapitre propose des méthodes bootstrap pour la construction des intervalles de prévision permettant de relâcher l'hypothèse de normalité des innovations. Nous y propo- sons des intervalles de prédiction bootstrap pour une observation h périodes dans le futur et sa moyenne conditionnelle. Nous supposons que ces prévisions sont faites en utilisant un ensemble de facteurs extraits d'un large panel de variables. Parce que nous traitons ces facteurs comme latents, nos prévisions dépendent à la fois des facteurs estimés et les coefficients de régres- sion estimés. Sous des conditions de régularité, Bai et Ng (2006) ont proposé la construction d'intervalles asymptotiques sous l'hypothèse de Gaussianité des innovations. Le bootstrap nous permet de relâcher cette hypothèse et de construire des intervalles de prédiction valides sous des hypothèses plus générales. En outre, même en supposant la Gaussianité, le bootstrap conduit à des intervalles plus précis dans les cas où la dimension transversale est relativement faible car il prend en considération le biais de l'estimateur des moindres carrés ordinaires comme le montre une étude récente de Gonçalves et Perron (2014). Dans le troisième chapitre, nous suggérons des procédures de sélection convergentes pour les regressions augmentées de facteurs en échantillons finis. Nous démontrons premièrement que la méthode de validation croisée usuelle est non-convergente mais que sa généralisation, la validation croisée «leave-d-out» sélectionne le plus petit ensemble de facteurs estimés pour l'espace généré par les vraies facteurs. Le deuxième critère dont nous montrons également la validité généralise l'approximation bootstrap de Shao (1996) pour les regressions augmentées de facteurs. Les simulations montrent une amélioration de la probabilité de sélectionner par- cimonieusement les facteurs estimés comparativement aux méthodes de sélection disponibles. L'application empirique revisite la relation entre les facteurs macroéconomiques et financiers, et l'excès de rendement sur le marché boursier américain. Parmi les facteurs estimés à partir d'un large panel de données macroéconomiques et financières des États Unis, les facteurs fortement correlés aux écarts de taux d'intérêt et les facteurs de Fama-French ont un bon pouvoir prédictif pour les excès de rendement.

Méthodes de Bootstrap pour les modèles à facteurs

Cette thèse développe des méthodes bootstrap pour les modèles à facteurs qui sont couram- ment utilisés pour générer des prévisions depuis l'article pionnier de Stock et Watson (2002) sur les indices de diffusion. Ces modèles tolèrent l'inclusion d'un grand nombre de variables macroéconomiques et financières comme prédicteurs, une caractéristique utile pour inclure di- verses informations disponibles aux agents économiques. Ma thèse propose donc des outils éco- nométriques qui améliorent l'inférence dans les modèles à facteurs utilisant des facteurs latents extraits d'un large panel de prédicteurs observés. Il est subdivisé en trois chapitres complémen- taires dont les deux premiers en collaboration avec Sílvia Gonçalves et Benoit Perron. Dans le premier article, nous étudions comment les méthodes bootstrap peuvent être utilisées pour faire de l'inférence dans les modèles de prévision pour un horizon de h périodes dans le futur. Pour ce faire, il examine l'inférence bootstrap dans un contexte de régression augmentée de facteurs où les erreurs pourraient être autocorrélées. Il généralise les résultats de Gonçalves et Perron (2014) et propose puis justifie deux approches basées sur les résidus : le block wild bootstrap et le dependent wild bootstrap. Nos simulations montrent une amélioration des taux de couverture des intervalles de confiance des coefficients estimés en utilisant ces approches comparativement à la théorie asymptotique et au wild bootstrap en présence de corrélation sérielle dans les erreurs de régression. Le deuxième chapitre propose des méthodes bootstrap pour la construction des intervalles de prévision permettant de relâcher l'hypothèse de normalité des innovations. Nous y propo- sons des intervalles de prédiction bootstrap pour une observation h périodes dans le futur et sa moyenne conditionnelle. Nous supposons que ces prévisions sont faites en utilisant un ensemble de facteurs extraits d'un large panel de variables. Parce que nous traitons ces facteurs comme latents, nos prévisions dépendent à la fois des facteurs estimés et les coefficients de régres- sion estimés. Sous des conditions de régularité, Bai et Ng (2006) ont proposé la construction d'intervalles asymptotiques sous l'hypothèse de Gaussianité des innovations. Le bootstrap nous permet de relâcher cette hypothèse et de construire des intervalles de prédiction valides sous des hypothèses plus générales. En outre, même en supposant la Gaussianité, le bootstrap conduit à des intervalles plus précis dans les cas où la dimension transversale est relativement faible car il prend en considération le biais de l'estimateur des moindres carrés ordinaires comme le montre une étude récente de Gonçalves et Perron (2014). Dans le troisième chapitre, nous suggérons des procédures de sélection convergentes pour les regressions augmentées de facteurs en échantillons finis. Nous démontrons premièrement que la méthode de validation croisée usuelle est non-convergente mais que sa généralisation, la validation croisée «leave-d-out» sélectionne le plus petit ensemble de facteurs estimés pour l'espace généré par les vraies facteurs. Le deuxième critère dont nous montrons également la validité généralise l'approximation bootstrap de Shao (1996) pour les regressions augmentées de facteurs. Les simulations montrent une amélioration de la probabilité de sélectionner par- cimonieusement les facteurs estimés comparativement aux méthodes de sélection disponibles. L'application empirique revisite la relation entre les facteurs macroéconomiques et financiers, et l'excès de rendement sur le marché boursier américain. Parmi les facteurs estimés à partir d'un large panel de données macroéconomiques et financières des États Unis, les facteurs fortement correlés aux écarts de taux d'intérêt et les facteurs de Fama-French ont un bon pouvoir prédictif pour les excès de rendement.

QSAR Modeling of a Set of Pyrazinoate Esters as Antituberculosis Prodrugs

Tuberculosis is an infection caused mainly by Mycobacterium tuberculosis. A first-line antimycobacterial drug is pyrazinamide (PZA), which acts partially as a prodrug activated by a pyrazinamidase releasing the active agent, pyrazinoic acid (POA). As pyrazinoic acid presents some difficulty to cross the mycobacterial cell wall, and also the pyrazinamide-resistant strains do not express the pyrazinamidase, a set of pyrazinoic acid esters have been evaluated as antimycobacterial agents. In this work, a QSAR approach was applied to a set of forty-three pyrazinoates against M. tuberculosis ATCC 27294, using genetic algorithm function and partial least squares regression (WOLF 5.5 program). The independent variables selected were the Balaban index (I), calculated n-octanol/water partition coefficient (ClogP), van-der-Waals surface area, dipole moment, and stretching-energy contribution. The final QSAR model (N = 32, r(2) = 0.68, q(2) = 0.59, LOF = 0.25, and LSE = 0.19) was fully validated employing leave-N-out cross-validation and y-scrambling techniques. The test set (N = 11) presented an external prediction power of 73%. In conclusion, the QSAR model generated can be used as a valuable tool to optimize the activity of future pyrazinoic acid esters in the designing of new antituberculosis agents.

Application of GA-PLS and GA-KPLS calculations for the prediction of the retention indices of essential oils

Genetic algorithm and partial least square (GA-PLS) and kernel PLS (GA-KPLS) techniques were used to investigate the correlation between retention indices (RI) and descriptors for 117 diverse compounds in essential oils from 5 Pimpinella species gathered from central Turkey which were obtained by gas chromatography and gas chromatography-mass spectrometry. The square correlation coefficient leave-group-out cross validation (LGO-CV) (Q²) between experimental and predicted RI for training set by GA-PLS and GA-KPLS was 0.940 and 0.963, respectively. This indicates that GA-KPLS can be used as an alternative modeling tool for quantitative structure-retention relationship (QSRR) studies.

QSAR modeling: um novo pacote computacional open source para gerar e validar modelos QSAR

QSAR modeling is a novel computer program developed to generate and validate QSAR or QSPR (quantitative structure- activity or property relationships) models. With QSAR modeling, users can build partial least squares (PLS) regression models, perform variable selection with the ordered predictors selection (OPS) algorithm, and validate models by using y-randomization and leave-N-out cross validation. An additional new feature is outlier detection carried out by simultaneous comparison of sample leverage with the respective Studentized residuals. The program was developed using Java version 6, and runs on any operating system that supports Java Runtime Environment version 6. The use of the program is illustrated. This program is available for download at lqta.iqm.unicamp.br.

Validation procedures in radiological diagnostic models. Neural network and logistic regression

The objective of this paper is to compare the performance of twopredictive radiological models, logistic regression (LR) and neural network (NN), with five different resampling methods. One hundred and sixty-seven patients with proven calvarial lesions as the only known disease were enrolled. Clinical and CT data were used for LR and NN models. Both models were developed with cross validation, leave-one-out and three different bootstrap algorithms. The final results of each model were compared with error rate and the area under receiver operating characteristic curves (Az). The neural network obtained statistically higher Az than LR with cross validation. The remaining resampling validation methods did not reveal statistically significant differences between LR and NN rules. The neural network classifier performs better than the one based on logistic regression. This advantage is well detected by three-fold cross-validation, but remains unnoticed when leave-one-out or bootstrap algorithms are used.

Development, validation and uncertainty analysis of quantitative structure and activity relationship models for Log P of disinfection by -products

Hydrophobicity as measured by Log P is an important molecular property related to toxicity and carcinogenicity. With increasing public health concerns for the effects of Disinfection By-Products (DBPs), there are considerable benefits in developing Quantitative Structure and Activity Relationship (QSAR) models capable of accurately predicting Log P. In this research, Log P values of 173 DBP compounds in 6 functional classes were used to develop QSAR models, by applying 3 molecular descriptors, namely, Energy of the Lowest Unoccupied Molecular Orbital (ELUMO), Number of Chlorine (NCl) and Number of Carbon (NC) by Multiple Linear Regression (MLR) analysis. The QSAR models developed were validated based on the Organization for Economic Co-operation and Development (OECD) principles. The model Applicability Domain (AD) and mechanistic interpretation were explored. Considering the very complex nature of DBPs, the established QSAR models performed very well with respect to goodness-of-fit, robustness and predictability. The predicted values of Log P of DBPs by the QSAR models were found to be significant with a correlation coefficient R2 from 81% to 98%. The Leverage Approach by Williams Plot was applied to detect and remove outliers, consequently increasing R 2 by approximately 2% to 13% for different DBP classes. The developed QSAR models were statistically validated for their predictive power by the Leave-One-Out (LOO) and Leave-Many-Out (LMO) cross validation methods. Finally, Monte Carlo simulation was used to assess the variations and inherent uncertainties in the QSAR models of Log P and determine the most influential parameters in connection with Log P prediction. The developed QSAR models in this dissertation will have a broad applicability domain because the research data set covered six out of eight common DBP classes, including halogenated alkane, halogenated alkene, halogenated aromatic, halogenated aldehyde, halogenated ketone, and halogenated carboxylic acid, which have been brought to the attention of regulatory agencies in recent years. Furthermore, the QSAR models are suitable to be used for prediction of similar DBP compounds within the same applicability domain. The selection and integration of various methodologies developed in this research may also benefit future research in similar fields.

Comparison of different devices for predicting the lean meat percentage of pig carcasses

Lean meat percentage (LMP) is the criterion for carcass classification and it must be measured on line objectively. The aim of this work was to compare the error of the prediction (RMSEP) of the LMP measured with the following different devices: Fat-O-Meat’er (FOM), UltraFOM (UFOM), AUTOFOM and -VCS2000. For this reason the same 99 carcasses were measured using all 4 apparatus and dissected according to the European Reference Method. Moreover a subsample of the carcasses (n=77) were fully scanned with a X-ray Computed Tomography equipment (CT). The RMSEP calculated with cross validation leave-one-out was lower for FOM and AUTOFOM (1.8% and 1.9%, respectively) and higher for UFOM and VCS2000 (2.3% for both devices). The error obtained with CT was the lowest (0.96%) in accordance with previous results, but CT cannot be used on line. It can be concluded that FOM and AUTOFOM presented better accuracy than UFOM and VCS2000.