New kernel functions and learning methods for text and data mining

Recent advances in machine learning methods enable increasingly the automatic construction of various types of computer assisted methods that have been difficult or laborious to program by human experts. The tasks for which this kind of tools are needed arise in many areas, here especially in the fields of bioinformatics and natural language processing. The machine learning methods may not work satisfactorily if they are not appropriately tailored to the task in question. However, their learning performance can often be improved by taking advantage of deeper insight of the application domain or the learning problem at hand. This thesis considers developing kernel-based learning algorithms incorporating this kind of prior knowledge of the task in question in an advantageous way. Moreover, computationally efficient algorithms for training the learning machines for specific tasks are presented. In the context of kernel-based learning methods, the incorporation of prior knowledge is often done by designing appropriate kernel functions. Another well-known way is to develop cost functions that fit to the task under consideration. For disambiguation tasks in natural language, we develop kernel functions that take account of the positional information and the mutual similarities of words. It is shown that the use of this information significantly improves the disambiguation performance of the learning machine. Further, we design a new cost function that is better suitable for the task of information retrieval and for more general ranking problems than the cost functions designed for regression and classification. We also consider other applications of the kernel-based learning algorithms such as text categorization, and pattern recognition in differential display. We develop computationally efficient algorithms for training the considered learning machines with the proposed kernel functions. We also design a fast cross-validation algorithm for regularized least-squares type of learning algorithm. Further, an efficient version of the regularized least-squares algorithm that can be used together with the new cost function for preference learning and ranking tasks is proposed. In summary, we demonstrate that the incorporation of prior knowledge is possible and beneficial, and novel advanced kernels and cost functions can be used in algorithms efficiently.

L'informatique: un outil remarquable pour l'enseignement de la médecine interne générale [The informatics: a remarkable tool for teaching general internal medicine].

INTERMED training implies a three week course, integrated in the "primary care module" for medical students in the first master year at the school of medicine in Lausanne. INTERMED uses an innovative teaching method based on repetitive sequences of e-learning-based individual learning followed by collaborative learning activities in teams, named Team-based learning (TBL). The e-learning takes place in a web-based virtual learning environment using a series of interactive multimedia virtual patients. By using INTERMED students go through a complete medical encounter applying clinical reasoning and choosing the diagnostic and therapeutic approach. INTERMED offers an authentic experience in an engaging and safe environment where errors are allowed and without consequences.

Diagnostic methods and treatment options for focal cortical dysplasia.

Our inability to adequately treat many patients with refractory epilepsy caused by focal cortical dysplasia (FCD), surgical inaccessibility and failures are significant clinical drawbacks. The targeting of physiologic features of epileptogenesis in FCD and colocalizing functionality has enhanced completeness of surgical resection, the main determinant of outcome. Electroencephalography (EEG)-functional magnetic resonance imaging (fMRI) and magnetoencephalography are helpful in guiding electrode implantation and surgical treatment, and high-frequency oscillations help defining the extent of the epileptogenic dysplasia. Ultra high-field MRI has a role in understanding the laminar organization of the cortex, and fluorodeoxyglucose-positron emission tomography (FDG-PET) is highly sensitive for detecting FCD in MRI-negative cases. Multimodal imaging is clinically valuable, either by improving the rate of postoperative seizure freedom or by reducing postoperative deficits. However, there is no level 1 evidence that it improves outcomes. Proof for a specific effect of antiepileptic drugs (AEDs) in FCD is lacking. Pathogenic mutations recently described in mammalian target of rapamycin (mTOR) genes in FCD have yielded important insights into novel treatment options with mTOR inhibitors, which might represent an example of personalized treatment of epilepsy based on the known mechanisms of disease. The ketogenic diet (KD) has been demonstrated to be particularly effective in children with epilepsy caused by structural abnormalities, especially FCD. It attenuates epigenetic chromatin modifications, a master regulator for gene expression and functional adaptation of the cell, thereby modifying disease progression. This could imply lasting benefit of dietary manipulation. Neurostimulation techniques have produced variable clinical outcomes in FCD. In widespread dysplasias, vagus nerve stimulation (VNS) has achieved responder rates >50%; however, the efficacy of noninvasive cranial nerve stimulation modalities such as transcutaneous VNS (tVNS) and noninvasive (nVNS) requires further study. Although review of current strategies underscores the serious shortcomings of treatment-resistant cases, initial evidence from novel approaches suggests that future success is possible.

Verifique sus conocimientos sobre medicina nuclear (2): tomografía por emisión de positrones (PET) y PET con tomografía computerizada (PET-TC)

La presente entrega de la serie de Nursing sobre las pruebas complementarias está dedicada a la tomografía por emisión de positrones o PET, acrónimo de positron emission tomography. La PET es una técnica de diagnóstico por la imagen de medicina nuclear en la cual se administra al paciente un radiofármaco emisor de positrones. Este radiofármaco se incorpora a los tejidos adecuados siguiendo una vía metabólica determinada. La radiactividad emitida por esos tejidos del paciente es detectable por los equipos PET y se obtienen imágenes que proporcionan una información funcional in vivo. El radiofármaco PET más habitual es un análogo de la glucosa que se llama F-18-fluordesoxiglucosa, conocido como FDG, el cual permite estudiar la actividad metabólica. La incorporación de la tomografía computarizada (TC) en el mismo equipo híbrido PET-TC permite obtener además la información anatómica del paciente. En el presente artículo se describen los fundamentos físicos y fisiológicos básicos de las exploraciones PET-TC con FDG en oncología, así como los procedimientos de enfermería necesarios para el cuidado del paciente y la correcta obtención de las imágenes.

Efficient total variation algorithm for fetal MRI reconstruction

Fetal MRI reconstruction aims at finding a high-resolution image given a small set of low-resolution images. It is usually modeled as an inverse problem where the regularization term plays a central role in the reconstruction quality. Literature has considered several regularization terms s.a. Dirichlet/Laplacian energy [1], Total Variation (TV)based energies [2,3] and more recently non-local means [4]. Although TV energies are quite attractive because of their ability in edge preservation, standard explicit steepest gradient techniques have been applied to optimize fetal-based TV energies. The main contribution of this work lies in the introduction of a well-posed TV algorithm from the point of view of convex optimization. Specifically, our proposed TV optimization algorithm for fetal reconstruction is optimal w.r.t. the asymptotic and iterative convergence speeds O(1/n(2)) and O(1/root epsilon), while existing techniques are in O(1/n) and O(1/epsilon). We apply our algorithm to (1) clinical newborn data, considered as ground truth, and (2) clinical fetal acquisitions. Our algorithm compares favorably with the literature in terms of speed and accuracy.

Verifique sus conocimientos sobre medicina nuclear (2): tomografía por emisión de positrones (PET) y PET con tomografía computerizada (PET-TC)

La presente entrega de la serie de Nursing sobre las pruebas complementarias está dedicada a la tomografía por emisión de positrones o PET, acrónimo de positron emission tomography. La PET es una técnica de diagnóstico por la imagen de medicina nuclear en la cual se administra al paciente un radiofármaco emisor de positrones. Este radiofármaco se incorpora a los tejidos adecuados siguiendo una vía metabólica determinada. La radiactividad emitida por esos tejidos del paciente es detectable por los equipos PET y se obtienen imágenes que proporcionan una información funcional in vivo. El radiofármaco PET más habitual es un análogo de la glucosa que se llama F-18-fluordesoxiglucosa, conocido como FDG, el cual permite estudiar la actividad metabólica. La incorporación de la tomografía computarizada (TC) en el mismo equipo híbrido PET-TC permite obtener además la información anatómica del paciente. En el presente artículo se describen los fundamentos físicos y fisiológicos básicos de las exploraciones PET-TC con FDG en oncología, así como los procedimientos de enfermería necesarios para el cuidado del paciente y la correcta obtención de las imágenes.

Impact of 18F-FDG PET scan on the prevalence of benign thoracic lesions at surgical resection

OBJECTIVE: The main utility of 18-fluorodeoxyglucose positron emission tomography (FDG-PET) lies in the staging of lung cancer. However, it can also be used to differentiate indeterminate pulmonary lesions, but its impact on the resection of benign lesions at surgery is unknown. The aim of this study was to compare the prevalence of benign lesions at thoracotomy carried out for suspected lung cancer, before and after the introduction of PET scanning in a large thoracic surgical centre. MATERIALS AND METHODS: We reviewed our prospectively recorded surgical database for all consecutive patients undergoing thoracotomy for suspected or proven lung cancer and compared the prevalence of benign lesions in 2 consecutive 2-year groups, before (group I) and after (group II) the introduction of FDG-PET scan respectively. RESULTS: Surgical resection was performed on 1233 patients during the study period. The prevalence of benign lesions at surgery in groups I and II was similar (44/626 and 41/607, both 7%), and also in group II between those who underwent FDG-PET scan and the remainder (21/301 and 20/306 respectively, both 7%). In group II, of the 21 patients with benign lesions, who underwent FDG-PET, 19 had a false positive scan (mean standardised uptake value 5.3 [range 2.6-12.7]). Of these, 13 and 4 patients respectively had non-diagnostic bronchoscopy and percutaneous transthoracic lung biopsy pre thoracotomy. There was no difference in the proportion of different benign lesions resected between group I and those with FDG-PET in group II. CONCLUSION: The introduction of FDG-PET scanning has not altered the proportion of patients undergoing thoracotomy for ultimately benign lesions, mainly due to the avidity for the isotope of some non-malignant lesions. Such false positive results need to be considered when patients with unconfirmed lung cancer are contemplated for surgical resection.

Cerebral Lactate Metabolism After Traumatic Brain Injury.

Cerebral energy dysfunction has emerged as an important determinant of prognosis following traumatic brain injury (TBI). A number of studies using cerebral microdialysis, positron emission tomography, and jugular bulb oximetry to explore cerebral metabolism in patients with TBI have demonstrated a critical decrease in the availability of the main energy substrate of brain cells (i.e., glucose). Energy dysfunction induces adaptations of cerebral metabolism that include the utilization of alternative energy resources that the brain constitutively has, such as lactate. Two decades of experimental and human investigations have convincingly shown that lactate stands as a major actor of cerebral metabolism. Glutamate-induced activation of glycolysis stimulates lactate production from glucose in astrocytes, with subsequent lactate transfer to neurons (astrocyte-neuron lactate shuttle). Lactate is not only used as an extra energy substrate but also acts as a signaling molecule and regulator of systemic and brain glucose use in the cerebral circulation. In animal models of brain injury (e.g., TBI, stroke), supplementation with exogenous lactate exerts significant neuroprotection. Here, we summarize the main clinical studies showing the pivotal role of lactate and cerebral lactate metabolism after TBI. We also review pilot interventional studies that examined exogenous lactate supplementation in patients with TBI and found hypertonic lactate infusions had several beneficial properties on the injured brain, including decrease of brain edema, improvement of neuroenergetics via a "cerebral glucose-sparing effect," and increase of cerebral blood flow. Hypertonic lactate represents a promising area of therapeutic investigation; however, larger studies are needed to further examine mechanisms of action and impact on outcome.

Connectivity and tissue microstructural alterations in right and left temporal lobe epilepsy revealed by diffusion spectrum imaging.

Focal epilepsy is increasingly recognized as the result of an altered brain network, both on the structural and functional levels and the characterization of these widespread brain alterations is crucial for our understanding of the clinical manifestation of seizure and cognitive deficits as well as for the management of candidates to epilepsy surgery. Tractography based on Diffusion Tensor Imaging allows non-invasive mapping of white matter tracts in vivo. Recently, diffusion spectrum imaging (DSI), based on an increased number of diffusion directions and intensities, has improved the sensitivity of tractography, notably with respect to the problem of fiber crossing and recent developments allow acquisition times compatible with clinical application. We used DSI and parcellation of the gray matter in regions of interest to build whole-brain connectivity matrices describing the mutual connections between cortical and subcortical regions in patients with focal epilepsy and healthy controls. In addition, the high angular and radial resolution of DSI allowed us to evaluate also some of the biophysical compartment models, to better understand the cause of the changes in diffusion anisotropy. Global connectivity, hub architecture and regional connectivity patterns were altered in TLE patients and showed different characteristics in RTLE vs LTLE with stronger abnormalities in RTLE. The microstructural analysis suggested that disturbed axonal density contributed more than fiber orientation to the connectivity changes affecting the temporal lobes whereas fiber orientation changes were more involved in extratemporal lobe changes. Our study provides further structural evidence that RTLE and LTLE are not symmetrical entities and DSI-based imaging could help investigate the microstructural correlate of these imaging abnormalities.

Noves funcionalitats per desenvolupar el treball en grup a dins de l'ACME

La plataforma ACME (Avaluació Continuada i Millora de l’Ensenyament) va ser creada l’any 1998 per un grup de professors del departament d’Informàtica i Matemàtica Aplicada. L’ACME es va concebre com una plataforma d’e-learning, és a dir, un sistema que mitjançant l’ús d’Internet afavorís l’aprenentatge, permeten la interactivitat entre l’alumne i el professor. La creació de la plataforma ACME tenia com a objectiu reduir el fracàs dels alumnes en les assignatures de matemàtiques, però degut a l’èxit que va suposar en aquestes, es va decidir incorporar la metodologia de treball ACME a altres disciplines com la programació, les bases de dades, la química, l’economia, etc. de manera que actualment es poden desenvolupar activitats ACME en moltes disciplines. Actualment l’ACME s’utilitza com a complement a les classes presencials, on el professor exposa de manera magistral els conceptes i resol algun exercici a mode d’exemple, per a que després l’alumne, utilitzant la plataforma ACME, intenti resoldre els exercicis proposats pel professor.L’objectiu d’aquest projecte és desenvolupar l’anàlisi, disseny i implementació de les modificacions necessàries a incorporar a la plataforma ACME per tal de millorar el gestor de grups, els exercicis Excel i finalment permetre el treball en grup. El projecte consta de tres parts: millorar les interfícies del professor i de l’alumne, la millora del exercicis Excel i la resolució d’exercicis en grup

Módulo para gestionar el modelo de usuario y permisos, el sistema de archivos y el de noticias en la plataforma CoSpace

CoSpace és una plataforma web dissenyada per proporcionar un espai virtual d’interacció i col•laboració entre formadors en comunitats virtuals. Es va originar com a resultat de les necessitats addicionals que van sorgir quan els professors que treballen en temes educatius en context de diversitat (en àrees de llenguatge, matemàtiques i ciències) van necessitar una eina per afavorir o recolzar la interacció i la col•laboració entre ells per compartir idees, experiències, objectes virtuals d’aprenentatge, entre d’altres, des d’una perspectiva actual de l’ús de la tecnologia i des d’una visió més propera a l’usuari final. Aquest paradigma promou la idea que les aplicacions han de ser concebudes per a usuaris poc experts i tenint en compte els principis d’usabilitat i accessibilitat.L’abast del projecte està definit per la necessitat de gestió per part dels professors, de la informació del seu perfil personal, permetent als usuaris la creació, edició, consulta i eliminació d’aquesta informació. També la necessitat de compartir arxius, publicar notícies entre comunitats de pràctica i, finalment, de la necessitat de gestionar els permisos d’usuari per tal que gestionin mòduls de la plataforma

Suggestion of a national diagnostic reference level for 18F-FDG/PET scans in adult cancer patients in Brazil

Objective To suggest a national value for the diagnostic reference level (DRL) in terms of activity in MBq.kg–1, for nuclear medicine procedures with fluorodeoxyglucose (18F-FDG) in whole body positron emission tomography (PET) scans of adult patients. Materials and Methods A survey on values of 18F-FDG activity administered in Brazilian clinics was undertaken by means of a questionnaire including questions about number and manufacturer of the installed equipment, model and detector type. The suggested DRL value was based on the calculation of the third quartile of the activity values distribution reported by the clinics. Results Among the surveyed Brazilian clinics, 58% responded completely or partially the questionnaire; and the results demonstrated variation of up to 100% in the reported radiopharmaceutical activity. The suggested DRL for 18F-FDG/PET activity was 5.54 MBq.kg–1 (0.149 mCi.kg–1). Conclusion The present study has demonstrated the lack of standardization in administered radiopharmaceutical activities for PET procedures in Brazil, corroborating the necessity of an official DRL value to be adopted in the country. The suggested DLR value demonstrates that there is room for optimization of the procedures and 18F-FDG/PET activities administered in Brazilian clinics to reduce the doses delivered to patients. It is important to highlight that this value should be continually revised and optimized at least every five years.

Prognostic value of baseline total metabolic tumor volume (TMTV0) measured on FDG-PET/CT in patients with peripheral T-cell lymphoma (PTCL).

BACKGROUND: Most peripheral T-cell lymphoma (PTCL) patients have a poor outcome and the identification of prognostic factors at diagnosis is needed. PATIENTS AND METHODS: The prognostic impact of total metabolic tumor volume (TMTV0), measured on baseline [(18)F]2-fluoro-2-deoxy-d-glucose positron emission tomography/computed tomography, was evaluated in a retrospective study including 108 PTCL patients (27 PTCL not otherwise specified, 43 angioimmunoblastic T-cell lymphomas and 38 anaplastic large-cell lymphomas). All received anthracycline-based chemotherapy. TMTV0 was computed with the 41% maximum standardized uptake value threshold method and an optimal cut-off point for binary outcomes was determined and compared with others prognostic factors. RESULTS: With a median follow-up of 23 months, 2-year progression-free survival (PFS) was 49% and 2-year overall survival (OS) was 67%. High TMTV0 was significantly associated with a worse prognosis. At 2 years, PFS was 26% in patients with a high TMTV0 (>230 cm(3), n = 53) versus 71% for those with a low TMTV0, [P < 0.0001, hazard ratio (HR) = 4], whereas OS was 50% versus 80%, respectively, (P = 0.0005, HR = 3.1). In multivariate analysis, TMTV0 was the only significant independent parameter for both PFS and OS. TMTV0, combined with PIT, discriminated even better than TMTV0 alone, patients with an adverse outcome (TMTV0 >230 cm(3) and PIT >1, n = 33,) from those with good prognosis (TMTV0 ≤230 cm(3) and PIT ≤1, n = 40): 19% versus 73% 2-year PFS (P < 0.0001) and 43% versus 81% 2-year OS, respectively (P = 0.0002). Thirty-one patients (other TMTV0-PIT combinations) had an intermediate outcome, 50% 2-year PFS and 68% 2-year OS. CONCLUSION: TMTV0 appears as an independent predictor of PTCL outcome. Combined with PIT, it could identify different risk categories at diagnosis and warrants further validation as a prognostic marker.

Correlation between PET/CT results and histological and immunohistochemical findings in breast carcinomas

Objective To correlate the results of 18F-fluoro-2-deoxy-D-glucose (18F-FDG) positron emission tomography/computed tomography (PET/CT) performed with a specific protocol for assessment of breasts with histological/immunohistochemical findings in breast carcinoma patients. Materials and Methods Cross-sectional study with prospective data collection, where patients with biopsy-confirmed breast carcinomas were studied. The patients underwent PET/CT examination in prone position, with a specific protocol for assessment of breasts. PET/CT findings were compared with histological and immunohistochemical data. Results The authors identified 59 malignant breast lesions in 50 patients. The maximum diameter of the lesions ranged from 6 to 80 mm (mean: 32.2 mm). Invasive ductal carcinoma was the most common histological type (n = 47; 79.7%). At PET/CT, 53 (89.8%) of the lesions demonstrated anomalous concentrations of 18F-FDG, with maximum SUV ranging from 0.8 to 23.1 (mean: 5.5). A statistically significant association was observed between higher values of maximum SUV and histological type, histological grade, molecular subtype, tumor diameter, mitotic index and Ki-67 expression. Conclusion PET/CT performed with specific protocol for assessment of breasts has demonstrated good sensitivity and was associated with relevant histological/immunohistochemical factors related to aggressiveness and prognosis of breast carcinomas.

Integración de recursos digitales en las tareas de aprendizaje de lenguas: portafolios electrónicos y traducción asistida por ordenador

Las herramientas informáticas abren un amplio campo de posibilidades pedagógicas a las asignaturas de lengua. En el presente artículo se propone un modelo de combinación de recursos digitales (portafolios electrónicos y traducción asistida por ordenador) que refuerzan proyectos docentes del ámbito de las lenguas desde un enfoque pedagógico socioconstructivista. En algunos casos, las actividades se pueden integrar en proyectos reales. Por otra parte, los proyectos relacionados con el uso de estas herramientas pueden tener un enfoque multidisciplinar que implique tanto a los departamentos de las lenguas extranjeras y como a los de las lenguas maternas.