Detecting reliable gene interactions by a hierarchy of Bayesian network classifiers

The main purpose of a gene interaction network is to map the relationships of the genes that are out of sight when a genomic study is tackled. DNA microarrays allow the measure of gene expression of thousands of genes at the same time. These data constitute the numeric seed for the induction of the gene networks. In this paper, we propose a new approach to build gene networks by means of Bayesian classifiers, variable selection and bootstrap resampling. The interactions induced by the Bayesian classifiers are based both on the expression levels and on the phenotype information of the supervised variable. Feature selection and bootstrap resampling add reliability and robustness to the overall process removing the false positive findings. The consensus among all the induced models produces a hierarchy of dependences and, thus, of variables. Biologists can define the depth level of the model hierarchy so the set of interactions and genes involved can vary from a sparse to a dense set. Experimental results show how these networks perform well on classification tasks. The biological validation matches previous biological findings and opens new hypothesis for future studies

Conceptual framework for the design, implementation and validation of psychomotor skills’ assessment systems in minimally invasive surgery

Las técnicas de cirugía de mínima invasión (CMI) se están consolidando hoy en día como alternativa a la cirugía tradicional, debido a sus numerosos beneficios para los pacientes. Este cambio de paradigma implica que los cirujanos deben aprender una serie de habilidades distintas de aquellas requeridas en cirugía abierta. El entrenamiento y evaluación de estas habilidades se ha convertido en una de las mayores preocupaciones en los programas de formación de cirujanos, debido en gran parte a la presión de una sociedad que exige cirujanos bien preparados y una reducción en el número de errores médicos. Por tanto, se está prestando especial atención a la definición de nuevos programas que permitan el entrenamiento y la evaluación de las habilidades psicomotoras en entornos seguros antes de que los nuevos cirujanos puedan operar sobre pacientes reales. Para tal fin, hospitales y centros de formación están gradualmente incorporando instalaciones de entrenamiento donde los residentes puedan practicar y aprender sin riesgos. Es cada vez más común que estos laboratorios dispongan de simuladores virtuales o simuladores físicos capaces de registrar los movimientos del instrumental de cada residente. Estos simuladores ofrecen una gran variedad de tareas de entrenamiento y evaluación, así como la posibilidad de obtener información objetiva de los ejercicios. Los diferentes estudios de validación llevados a cabo dan muestra de su utilidad; pese a todo, los niveles de evidencia presentados son en muchas ocasiones insuficientes. Lo que es más importante, no existe un consenso claro a la hora de definir qué métricas son más útiles para caracterizar la pericia quirúrgica. El objetivo de esta tesis doctoral es diseñar y validar un marco de trabajo conceptual para la definición y validación de entornos para la evaluación de habilidades en CMI, en base a un modelo en tres fases: pedagógica (tareas y métricas a emplear), tecnológica (tecnologías de adquisición de métricas) y analítica (interpretación de la competencia en base a las métricas). Para tal fin, se describe la implementación práctica de un entorno basado en (1) un sistema de seguimiento de instrumental fundamentado en el análisis del vídeo laparoscópico; y (2) la determinación de la pericia en base a métricas de movimiento del instrumental. Para la fase pedagógica se diseñó e implementó un conjunto de tareas para la evaluación de habilidades psicomotoras básicas, así como una serie de métricas de movimiento. La validación de construcción llevada a cabo sobre ellas mostró buenos resultados para tiempo, camino recorrido, profundidad, velocidad media, aceleración media, economía de área y economía de volumen. Adicionalmente, los resultados obtenidos en la validación de apariencia fueron en general positivos en todos los grupos considerados (noveles, residentes, expertos). Para la fase tecnológica, se introdujo el EVA Tracking System, una solución para el seguimiento del instrumental quirúrgico basado en el análisis del vídeo endoscópico. La precisión del sistema se evaluó a 16,33ppRMS para el seguimiento 2D de la herramienta en la imagen; y a 13mmRMS para el seguimiento espacial de la misma. La validación de construcción con una de las tareas de evaluación mostró buenos resultados para tiempo, camino recorrido, profundidad, velocidad media, aceleración media, economía de área y economía de volumen. La validación concurrente con el TrEndo® Tracking System por su parte presentó valores altos de correlación para 8 de las 9 métricas analizadas. Finalmente, para la fase analítica se comparó el comportamiento de tres clasificadores supervisados a la hora de determinar automáticamente la pericia quirúrgica en base a la información de movimiento del instrumental, basados en aproximaciones lineales (análisis lineal discriminante, LDA), no lineales (máquinas de soporte vectorial, SVM) y difusas (sistemas adaptativos de inferencia neurodifusa, ANFIS). Los resultados muestran que en media SVM presenta un comportamiento ligeramente superior: 78,2% frente a los 71% y 71,7% obtenidos por ANFIS y LDA respectivamente. Sin embargo las diferencias estadísticas medidas entre los tres no fueron demostradas significativas. En general, esta tesis doctoral corrobora las hipótesis de investigación postuladas relativas a la definición de sistemas de evaluación de habilidades para cirugía de mínima invasión, a la utilidad del análisis de vídeo como fuente de información y a la importancia de la información de movimiento de instrumental a la hora de caracterizar la pericia quirúrgica. Basándose en estos cimientos, se han de abrir nuevos campos de investigación que contribuyan a la definición de programas de formación estructurados y objetivos, que puedan garantizar la acreditación de cirujanos sobradamente preparados y promocionen la seguridad del paciente en el quirófano. Abstract Minimally invasive surgery (MIS) techniques have become a standard in many surgical sub-specialties, due to their many benefits for patients. However, this shift in paradigm implies that surgeons must acquire a complete different set of skills than those normally attributed to open surgery. Training and assessment of these skills has become a major concern in surgical learning programmes, especially considering the social demand for better-prepared professionals and for the decrease of medical errors. Therefore, much effort is being put in the definition of structured MIS learning programmes, where practice with real patients in the operating room (OR) can be delayed until the resident can attest for a minimum level of psychomotor competence. To this end, skills’ laboratory settings are being introduced in hospitals and training centres where residents may practice and be assessed on their psychomotor skills. Technological advances in the field of tracking technologies and virtual reality (VR) have enabled the creation of new learning systems such as VR simulators or enhanced box trainers. These systems offer a wide range of tasks, as well as the capability of registering objective data on the trainees’ performance. Validation studies give proof of their usefulness; however, levels of evidence reported are in many cases low. More importantly, there is still no clear consensus on topics such as the optimal metrics that must be used to assess competence, the validity of VR simulation, the portability of tracking technologies into real surgeries (for advanced assessment) or the degree to which the skills measured and obtained in laboratory environments transfer to the OR. The purpose of this PhD is to design and validate a conceptual framework for the definition and validation of MIS assessment environments based on a three-pillared model defining three main stages: pedagogical (tasks and metrics to employ), technological (metric acquisition technologies) and analytical (interpretation of competence based on metrics). To this end, a practical implementation of the framework is presented, focused on (1) a video-based tracking system and (2) the determination of surgical competence based on the laparoscopic instruments’ motionrelated data. The pedagogical stage’s results led to the design and implementation of a set of basic tasks for MIS psychomotor skills’ assessment, as well as the definition of motion analysis parameters (MAPs) to measure performance on said tasks. Validation yielded good construct results for parameters such as time, path length, depth, average speed, average acceleration, economy of area and economy of volume. Additionally, face validation results showed positive acceptance on behalf of the experts, residents and novices. For the technological stage the EVA Tracking System is introduced. EVA provides a solution for tracking laparoscopic instruments from the analysis of the monoscopic video image. Accuracy tests for the system are presented, which yielded an average RMSE of 16.33pp for 2D tracking of the instrument on the image and of 13mm for 3D spatial tracking. A validation experiment was conducted using one of the tasks and the most relevant MAPs. Construct validation showed significant differences for time, path length, depth, average speed, average acceleration, economy of area and economy of volume; especially between novices and residents/experts. More importantly, concurrent validation with the TrEndo® Tracking System presented high correlation values (>0.7) for 8 of the 9 MAPs proposed. Finally, the analytical stage allowed comparing the performance of three different supervised classification strategies in the determination of surgical competence based on motion-related information. The three classifiers were based on linear (linear discriminant analysis, LDA), non-linear (support vector machines, SVM) and fuzzy (adaptive neuro fuzzy inference systems, ANFIS) approaches. Results for SVM show slightly better performance than the other two classifiers: on average, accuracy for LDA, SVM and ANFIS was of 71.7%, 78.2% and 71% respectively. However, when confronted, no statistical significance was found between any of the three. Overall, this PhD corroborates the investigated research hypotheses regarding the definition of MIS assessment systems, the use of endoscopic video analysis as the main source of information and the relevance of motion analysis in the determination of surgical competence. New research fields in the training and assessment of MIS surgeons can be proposed based on these foundations, in order to contribute to the definition of structured and objective learning programmes that guarantee the accreditation of well-prepared professionals and the promotion of patient safety in the OR.

Markov blanket-based approach for learning multi-dimensional Bayesian network classifiers: An application to predict the European Quality of Life-5 Dimensions (EQ-5D) from the 39-item Parkinson's Disease Questionnaire (PDQ-39)

Multi-dimensional Bayesian network classifiers (MBCs) are probabilistic graphical models recently proposed to deal with multi-dimensional classification problems, where each instance in the data set has to be assigned to more than one class variable. In this paper, we propose a Markov blanket-based approach for learning MBCs from data. Basically, it consists of determining the Markov blanket around each class variable using the HITON algorithm, then specifying the directionality over the MBC subgraphs. Our approach is applied to the prediction problem of the European Quality of Life-5 Dimensions (EQ-5D) from the 39-item Parkinson’s Disease Questionnaire (PDQ-39) in order to estimate the health-related quality of life of Parkinson’s patients. Fivefold cross-validation experiments were carried out on randomly generated synthetic data sets, Yeast data set, as well as on a real-world Parkinson’s disease data set containing 488 patients. The experimental study, including comparison with additional Bayesian network-based approaches, back propagation for multi-label learning, multi-label k-nearest neighbor, multinomial logistic regression, ordinary least squares, and censored least absolute deviations, shows encouraging results in terms of predictive accuracy as well as the identification of dependence relationships among class and feature variables.

Decision boundary for discrete Bayesian network classifiers

Bayesian network classifiers are a powerful machine learning tool. In order to evaluate the expressive power of these models, we compute families of polynomials that sign-represent decision functions induced by Bayesian network classifiers. We prove that those families are linear combinations of products of Lagrange basis polynomials. In absence of V-structures in the predictor sub-graph, we are also able to prove that this family of polynomials does in- deed characterize the specific classifier considered. We then use this representation to bound the number of decision functions representable by Bayesian network classifiers with a given structure and we compare these bounds to the ones obtained using Vapnik-Chervonenkis dimension.

Supervised classification of minimally invasive surgery psychomotor competence based on instruments motion analysis

Background Objective assessment of psychomotor skills has become an important challenge in the training of minimally invasive surgical (MIS) techniques. Currently, no gold standard defining surgical competence exists for classifying residents according to their surgical skills. Supervised classification has been proposed as a means for objectively establishing competence thresholds in psychomotor skills evaluation. This report presents a study comparing three classification methods for establishing their validity in a set of tasks for basic skills’ assessment. Methods Linear discriminant analysis (LDA), support vector machines (SVM), and adaptive neuro-fuzzy inference systems (ANFIS) were used. A total of 42 participants, divided into an experienced group (4 expert surgeons and 14 residents with >10 laparoscopic surgeries performed) and a nonexperienced group (16 students and 8 residents with <10 laparoscopic surgeries performed), performed three box trainer tasks validated for assessment of MIS psychomotor skills. Instrument movements were captured using the TrEndo tracking system, and nine motion analysis parameters (MAPs) were analyzed. The performance of the classifiers was measured by leave-one-out cross-validation using the scores obtained by the participants. Results The mean accuracy performances of the classifiers were 71 % (LDA), 78.2 % (SVM), and 71.7 % (ANFIS). No statistically significant differences in the performance were identified between the classifiers. Conclusions The three proposed classifiers showed good performance in the discrimination of skills, especially when information from all MAPs and tasks combined were considered. A correlation between the surgeons’ previous experience and their execution of the tasks could be ascertained from results. However, misclassifications across all the classifiers could imply the existence of other factors influencing psychomotor competence.

GMM-based classifiers for the automatic detection of obstructive sleep apnea

The aim of automatic pathological voice detection systems is to serve as tools, to medical specialists, for a more objective, less invasive and improved diagnosis of diseases. In this respect, the gold standard for those system include the usage of a optimized representation of the spectral envelope, either based on cepstral coefficients from the mel-scaled Fourier spectral envelope (Mel-Frequency Cepstral Coefficients) or from an all-pole estimation (Linear Prediction Coding Cepstral Coefficients) forcharacterization, and Gaussian Mixture Models for posterior classification. However, the study of recently proposed GMM-based classifiers as well as Nuisance mitigation techniques, such as those employed in speaker recognition, has not been widely considered inpathology detection labours. The present work aims at testing whether or not the employment of such speaker recognition tools might contribute to improve system performance in pathology detection systems, specifically in the automatic detection of Obstructive Sleep Apnea. The testing procedure employs an Obstructive Sleep Apnea database, in conjunction with GMM-based classifiers looking for a better performance. The results show that an improved performance might be obtained by using such approach.

Decision boundary for discrete Bayesian network classifiers

Bayesian network classifiers are a powerful machine learning tool. In order to evaluate the expressive power of these models, we compute families of polynomials that sign-represent decision functions induced by Bayesian network classifiers. We prove that those families are linear combinations of products of Lagrange basis polynomials. In absence of V-structures in the predictor sub-graph, we are also able to prove that this family of polynomials does in- deed characterize the specific classifier considered. We then use this representation to bound the number of decision functions representable by Bayesian network classifiers with a given structure and we compare these bounds to the ones obtained using Vapnik-Chervonenkis dimension.

Expressive power of binary relevance and chain classifiers based on Bayesian Networks for multi-label classification

Bayesian network classifiers are widely used in machine learning because they intuitively represent causal relations. Multi-label classification problems require each instance to be assigned a subset of a defined set of h labels. This problem is equivalent to finding a multi-valued decision function that predicts a vector of h binary classes. In this paper we obtain the decision boundaries of two widely used Bayesian network approaches for building multi-label classifiers: Multi-label Bayesian network classifiers built using the binary relevance method and Bayesian network chain classifiers. We extend our previous single-label results to multi-label chain classifiers, and we prove that, as expected, chain classifiers provide a more expressive model than the binary relevance method.

Predicting EQ-5D from the Parkinson's disease questionnaire PDQ-8 using multi-dimensional Bayesian network classifiers

The impact of the Parkinson's disease and its treatment on the patients' health-related quality of life can be estimated either by means of generic measures such as the european quality of Life-5 Dimensions (EQ-5D) or specific measures such as the 8-item Parkinson's disease questionnaire (PDQ-8). In clinical studies, PDQ-8 could be used in detriment of EQ-5D due to the lack of resources, time or clinical interest in generic measures. Nevertheless, PDQ-8 cannot be applied in cost-effectiveness analyses which require generic measures and quantitative utility scores, such as EQ-5D. To deal with this problem, a commonly used solution is the prediction of EQ-5D from PDQ-8. In this paper, we propose a new probabilistic method to predict EQ-5D from PDQ-8 using multi-dimensional Bayesian network classifiers. Our approach is evaluated using five-fold cross-validation experiments carried out on a Parkinson's data set containing 488 patients, and is compared with two additional Bayesian network-based approaches, two commonly used mapping methods namely, ordinary least squares and censored least absolute deviations, and a deterministic model. Experimental results are promising in terms of predictive performance as well as the identification of dependence relationships among EQ-5D and PDQ-8 items that the mapping approaches are unable to detect

Robust thermal face recognition using region classifiers

This paper presents a robust approach for recognition of thermal face images based on decision level fusion of 34 different region classifiers. The region classifiers concentrate on local variations. They use singular value decomposition (SVD) for feature extraction. Fusion of decisions of the region classifier is done by using majority voting technique. The algorithm is tolerant against false exclusion of thermal information produced by the presence of inconsistent distribution of temperature statistics which generally make the identification process difficult. The algorithm is extensively evaluated on UGC-JU thermal face database, and Terravic facial infrared database and the recognition performance are found to be 95.83% and 100%, respectively. A comparative study has also been made with the existing works in the literature.

Advanced background modeling with RGB-D sensors through classifiers combination and inter-frame foreground prediction

An innovative background modeling technique that is able to accurately segment foreground regions in RGB-D imagery (RGB plus depth) has been presented in this paper. The technique is based on a Bayesian framework that efficiently fuses different sources of information to segment the foreground. In particular, the final segmentation is obtained by considering a prediction of the foreground regions, carried out by a novel Bayesian Network with a depth-based dynamic model, and, by considering two independent depth and color-based mixture of Gaussians background models. The efficient Bayesian combination of all these data reduces the noise and uncertainties introduced by the color and depth features and the corresponding models. As a result, more compact segmentations, and refined foreground object silhouettes are obtained. Experimental results with different databases suggest that the proposed technique outperforms existing state-of-the-art algorithms.

Multi-sensor background subtraction by fusing multiple region-based probabilistic classifiers

In the recent years, the computer vision community has shown great interest on depth-based applications thanks to the performance and flexibility of the new generation of RGB-D imagery. In this paper, we present an efficient background subtraction algorithm based on the fusion of multiple region-based classifiers that processes depth and color data provided by RGB-D cameras. Foreground objects are detected by combining a region-based foreground prediction (based on depth data) with different background models (based on a Mixture of Gaussian algorithm) providing color and depth descriptions of the scene at pixel and region level. The information given by these modules is fused in a mixture of experts fashion to improve the foreground detection accuracy. The main contributions of the paper are the region-based models of both background and foreground, built from the depth and color data. The obtained results using different database sequences demonstrate that the proposed approach leads to a higher detection accuracy with respect to existing state-of-the-art techniques.

Background foreground segmentation with RGB-D Kinect data: An efficient combination of classifiers

Low cost RGB-D cameras such as the Microsoft’s Kinect or the Asus’s Xtion Pro are completely changing the computer vision world, as they are being successfully used in several applications and research areas. Depth data are particularly attractive and suitable for applications based on moving objects detection through foreground/background segmentation approaches; the RGB-D applications proposed in literature employ, in general, state of the art foreground/background segmentation techniques based on the depth information without taking into account the color information. The novel approach that we propose is based on a combination of classifiers that allows improving background subtraction accuracy with respect to state of the art algorithms by jointly considering color and depth data. In particular, the combination of classifiers is based on a weighted average that allows to adaptively modifying the support of each classifier in the ensemble by considering foreground detections in the previous frames and the depth and color edges. In this way, it is possible to reduce false detections due to critical issues that can not be tackled by the individual classifiers such as: shadows and illumination changes, color and depth camouflage, moved background objects and noisy depth measurements. Moreover, we propose, for the best of the author’s knowledge, the first publicly available RGB-D benchmark dataset with hand-labeled ground truth of several challenging scenarios to test background/foreground segmentation algorithms.

Decision boundary for discrete Bayesian network classifiers

Bayesian network classifiers are a powerful machine learning tool. In order to evaluate the expressive power of these models, we compute families of polynomials that sign-represent decision functions induced by Bayesian network classifiers. We prove that those families are linear combinations of products of Lagrange basis polynomials. In absence of V -structures in the predictor sub-graph, we are also able to prove that this family of polynomials does indeed characterize the specific classifier considered. We then use this representation to bound the number of decision functions representable by Bayesian network classifiers with a given structure.

Improving classification using a Confidence Matrix based on weak classifiers applied to OCR

This paper proposes a new feature representation method based on the construction of a Confidence Matrix (CM). This representation consists of posterior probability values provided by several weak classifiers, each one trained and used in different sets of features from the original sample. The CM allows the final classifier to abstract itself from discovering underlying groups of features. In this work the CM is applied to isolated character image recognition, for which several set of features can be extracted from each sample. Experimentation has shown that the use of CM permits a significant improvement in accuracy in most cases, while the others remain the same. The results were obtained after experimenting with four well-known corpora, using evolved meta-classifiers with the k-Nearest Neighbor rule as a weak classifier and by applying statistical significance tests.