Advanced Kernel Methods For Remote Sensing Image Classification

Résumé Suite aux recentes avancées technologiques, les archives d'images digitales ont connu une croissance qualitative et quantitative sans précédent. Malgré les énormes possibilités qu'elles offrent, ces avancées posent de nouvelles questions quant au traitement des masses de données saisies. Cette question est à la base de cette Thèse: les problèmes de traitement d'information digitale à très haute résolution spatiale et/ou spectrale y sont considérés en recourant à des approches d'apprentissage statistique, les méthodes à noyau. Cette Thèse étudie des problèmes de classification d'images, c'est à dire de catégorisation de pixels en un nombre réduit de classes refletant les propriétés spectrales et contextuelles des objets qu'elles représentent. L'accent est mis sur l'efficience des algorithmes, ainsi que sur leur simplicité, de manière à augmenter leur potentiel d'implementation pour les utilisateurs. De plus, le défi de cette Thèse est de rester proche des problèmes concrets des utilisateurs d'images satellite sans pour autant perdre de vue l'intéret des méthodes proposées pour le milieu du machine learning dont elles sont issues. En ce sens, ce travail joue la carte de la transdisciplinarité en maintenant un lien fort entre les deux sciences dans tous les développements proposés. Quatre modèles sont proposés: le premier répond au problème de la haute dimensionalité et de la redondance des données par un modèle optimisant les performances en classification en s'adaptant aux particularités de l'image. Ceci est rendu possible par un système de ranking des variables (les bandes) qui est optimisé en même temps que le modèle de base: ce faisant, seules les variables importantes pour résoudre le problème sont utilisées par le classifieur. Le manque d'information étiquétée et l'incertitude quant à sa pertinence pour le problème sont à la source des deux modèles suivants, basés respectivement sur l'apprentissage actif et les méthodes semi-supervisées: le premier permet d'améliorer la qualité d'un ensemble d'entraînement par interaction directe entre l'utilisateur et la machine, alors que le deuxième utilise les pixels non étiquetés pour améliorer la description des données disponibles et la robustesse du modèle. Enfin, le dernier modèle proposé considère la question plus théorique de la structure entre les outputs: l'intègration de cette source d'information, jusqu'à présent jamais considérée en télédétection, ouvre des nouveaux défis de recherche. Advanced kernel methods for remote sensing image classification Devis Tuia Institut de Géomatique et d'Analyse du Risque September 2009 Abstract The technical developments in recent years have brought the quantity and quality of digital information to an unprecedented level, as enormous archives of satellite images are available to the users. However, even if these advances open more and more possibilities in the use of digital imagery, they also rise several problems of storage and treatment. The latter is considered in this Thesis: the processing of very high spatial and spectral resolution images is treated with approaches based on data-driven algorithms relying on kernel methods. In particular, the problem of image classification, i.e. the categorization of the image's pixels into a reduced number of classes reflecting spectral and contextual properties, is studied through the different models presented. The accent is put on algorithmic efficiency and the simplicity of the approaches proposed, to avoid too complex models that would not be used by users. The major challenge of the Thesis is to remain close to concrete remote sensing problems, without losing the methodological interest from the machine learning viewpoint: in this sense, this work aims at building a bridge between the machine learning and remote sensing communities and all the models proposed have been developed keeping in mind the need for such a synergy. Four models are proposed: first, an adaptive model learning the relevant image features has been proposed to solve the problem of high dimensionality and collinearity of the image features. This model provides automatically an accurate classifier and a ranking of the relevance of the single features. The scarcity and unreliability of labeled. information were the common root of the second and third models proposed: when confronted to such problems, the user can either construct the labeled set iteratively by direct interaction with the machine or use the unlabeled data to increase robustness and quality of the description of data. Both solutions have been explored resulting into two methodological contributions, based respectively on active learning and semisupervised learning. Finally, the more theoretical issue of structured outputs has been considered in the last model, which, by integrating outputs similarity into a model, opens new challenges and opportunities for remote sensing image processing.

Identification of different heart tissues from MRI C-SENC images using an unsupervised multi-stage fuzzy clustering technique.

PURPOSE: To objectively characterize different heart tissues from functional and viability images provided by composite-strain-encoding (C-SENC) MRI. MATERIALS AND METHODS: C-SENC is a new MRI technique for simultaneously acquiring cardiac functional and viability images. In this work, an unsupervised multi-stage fuzzy clustering method is proposed to identify different heart tissues in the C-SENC images. The method is based on sequential application of the fuzzy c-means (FCM) and iterative self-organizing data (ISODATA) clustering algorithms. The proposed method is tested on simulated heart images and on images from nine patients with and without myocardial infarction (MI). The resulting clustered images are compared with MRI delayed-enhancement (DE) viability images for determining MI. Also, Bland-Altman analysis is conducted between the two methods. RESULTS: Normal myocardium, infarcted myocardium, and blood are correctly identified using the proposed method. The clustered images correctly identified 90 +/- 4% of the pixels defined as infarct in the DE images. In addition, 89 +/- 5% of the pixels defined as infarct in the clustered images were also defined as infarct in DE images. The Bland-Altman results show no bias between the two methods in identifying MI. CONCLUSION: The proposed technique allows for objectively identifying divergent heart tissues, which would be potentially important for clinical decision-making in patients with MI.

Indicators of therapeutic effect in FIT-06, a Phase II trial of a DNA vaccine, GTU(®)-Multi-HIVB, in untreated HIV-1 infected subjects.

BACKGROUND: Combination highly active antiretroviral therapy (HAART) has significantly decreased HIV-1 related morbidity and mortality globally transforming HIV into a controllable condition. HAART has a number of limitations though, including limited access in resource constrained countries, which have driven the search for simpler, affordable HIV-1 treatment modalities. Therapeutic HIV-1 vaccines aim to provide immunological support to slow disease progression and decrease transmission. We evaluated the safety, immunogenicity and clinical effect of a novel recombinant plasmid DNA therapeutic HIV-1 vaccine, GTU(®)-multi-HIVB, containing 6 different genes derived from an HIV-1 subtype B isolate. METHODS: 63 untreated, healthy, HIV-1 infected, adults between 18 and 40 years were enrolled in a single-blinded, placebo-controlled Phase II trial in South Africa. Subjects were HIV-1 subtype C infected, had never received antiretrovirals, with CD4 ≥ 350 cells/mm(3) and pHIV-RNA ≥ 50 copies/mL at screening. Subjects were allocated to vaccine or placebo groups in a 2:1 ratio either administered intradermally (ID) (0.5mg/dose) or intramuscularly (IM) (1mg/dose) at 0, 4 and 12 weeks boosted at 76 and 80 weeks with 1mg/dose (ID) and 2mg/dose (IM), respectively. Safety was assessed by adverse event monitoring and immunogenicity by HIV-1-specific CD4+ and CD8+ T-cells using intracellular cytokine staining (ICS), pHIV-RNA and CD4 counts. RESULTS: Vaccine was safe and well tolerated with no vaccine related serious adverse events. Significant declines in log pHIV-RNA (p=0.012) and increases in CD4+ T cell counts (p=0.066) were observed in the vaccine group compared to placebo, more pronounced after IM administration and in some HLA haplotypes (B*5703) maintained for 17 months after the final immunisation. CONCLUSIONS: The GTU(®)-multi-HIVB plasmid recombinant DNA therapeutic HIV-1 vaccine is safe, well tolerated and favourably affects pHIV-RNA and CD4 counts in untreated HIV-1 infected individuals after IM administration in subjects with HLA B*57, B*8101 and B*5801 haplotypes.

Detection of coronary thrombosis after multi-phase postmortem CT-angiography.

The aim of this study was to compare postmortem angiography-based, autopsy-based and histology-based diagnoses of acute coronary thrombosis in a series of medicolegal cases that underwent postmortem angiographies according to multiphase CT-angiography protocol. Our study included 150 medicolegal cases. All cases underwent native CT-scan, postmortem angiography, complete conventional autopsy and histological examination of the main organs and coronary arteries. In 10 out of the 150 investigated cases, postmortem angiographies revealed coronary arterial luminal filling defects and the absence of collateral vessels, suggesting acute coronary thromboses. Radiological findings were confirmed by autopsy and histological examinations in all cases. In 40 out of 150 cases, angiograms revealed complete or incomplete coronary arterial luminal filling defects and the presence of collateral vessels. Histological examinations did not reveal free-floating or non-adherent thrombi in the coronary arteries in any of these cases. Though postmortem angiography examination has not been well-established for the diagnosis of acute coronary thrombosis, luminal filling defects in coronary arteries suggesting acute thromboses can be observed through angiography and subsequently confirmed by autopsy and histological examinations.

Classification of urban multi-angular image sequences by aligning their manifolds

When dealing with multi-angular image sequences, problems of reflectance changes due either to illumination and acquisition geometry, or to interactions with the atmosphere, naturally arise. These phenomena interplay with the scene and lead to a modification of the measured radiance: for example, according to the angle of acquisition, tall objects may be seen from top or from the side and different light scatterings may affect the surfaces. This results in shifts in the acquired radiance, that make the problem of multi-angular classification harder and might lead to catastrophic results, since surfaces with the same reflectance return significantly different signals. In this paper, rather than performing atmospheric or bi-directional reflection distribution function (BRDF) correction, a non-linear manifold learning approach is used to align data structures. This method maximizes the similarity between the different acquisitions by deforming their manifold, thus enhancing the transferability of classification models among the images of the sequence.

CT dose optimization when changing to CT multi-detector row technology

The purpose of this article was to review the strategies to control patient dose in adult and pediatric computed tomography (CT), taking into account the change of technology from single-detector row CT to multi-detector row CT. First the relationships between computed tomography dose index, dose length product, and effective dose in adult and pediatric CT are revised, along with the diagnostic reference level concept. Then the effect of image noise as a function of volume computed tomography dose index, reconstructed slice thickness, and the size of the patient are described. Finally, the potential of tube current modulation CT is discussed.

Identity [R]Evolution: Multi-Disciplinary Perpectives

The identity [r]evolution is happening. Who are you, who am I in the information society? In recent years, the convergence of several factors - technological, political, economic - has accelerated a fundamental change in our networked world. On a technological level, information becomes easier to gather, to store, to exchange and to process. The belief that more information brings more security has been a strong political driver to promote information gathering since September 11. Profiling intends to transform information into knowledge in order to anticipate one's behaviour, or needs, or preferences. It can lead to categorizations according to some specific risk criteria, for example, or to direct and personalized marketing. As a consequence, new forms of identities appear. They are not necessarily related to our names anymore. They are based on information, on traces that we leave when we act or interact, when we go somewhere or just stay in one place, or even sometimes when we make a choice. They are related to the SIM cards of our mobile phones, to our credit card numbers, to the pseudonyms that we use on the Internet, to our email addresses, to the IP addresses of our computers, to our profiles... Like traditional identities, these new forms of identities can allow us to distinguish an individual within a group of people, or describe this person as belonging to a community or a category. How far have we moved through this process? The identity [r]evolution is already becoming part of our daily lives. People are eager to share information with their "friends" in social networks like Facebook, in chat rooms, or in Second Life. Customers take advantage of the numerous bonus cards that are made available. Video surveillance is becoming the rule. In several countries, traditional ID documents are being replaced by biometric passports with RFID technologies. This raises several privacy issues and might actually even result in changing the perception of the concept of privacy itself, in particular by the younger generation. In the information society, our (partial) identities become the illusory masks that we choose -or that we are assigned- to interplay and communicate with each other. Rights, obligations, responsibilities, even reputation are increasingly associated with these masks. On the one hand, these masks become the key to access restricted information and to use services. On the other hand, in case of a fraud or negative reputation, the owner of such a mask can be penalized: doors remain closed, access to services is denied. Hence the current preoccupying growth of impersonation, identity-theft and other identity-related crimes. Where is the path of the identity [r]evolution leading us? The booklet is giving a glance on possible scenarios in the field of identity.

What drives invasibility? A multi-model inference test and spatial modelling of alien plant species richness patterns in northern Portugal

Understanding and anticipating biological invasions can focus either on traits that favour species invasiveness or on features of the receiving communities, habitats or landscapes that promote their invasibility. Here, we address invasibility at the regional scale, testing whether some habitats and landscapes are more invasible than others by fitting models that relate alien plant species richness to various environmental predictors. We use a multi-model information-theoretic approach to assess invasibility by modelling spatial and ecological patterns of alien invasion in landscape mosaics and testing competing hypotheses of environmental factors that may control invasibility. Because invasibility may be mediated by particular characteristics of invasiveness, we classified alien species according to their C-S-R plant strategies. We illustrate this approach with a set of 86 alien species in Northern Portugal. We first focus on predictors influencing species richness and expressing invasibility and then evaluate whether distinct plant strategies respond to the same or different groups of environmental predictors. We confirmed climate as a primary determinant of alien invasions and as a primary environmental gradient determining landscape invasibility. The effects of secondary gradients were detected only when the area was sub-sampled according to predictions based on the primary gradient. Then, multiple predictor types influenced patterns of alien species richness, with some types (landscape composition, topography and fire regime) prevailing over others. Alien species richness responded most strongly to extreme land management regimes, suggesting that intermediate disturbance induces biotic resistance by favouring native species richness. Land-use intensification facilitated alien invasion, whereas conservation areas hosted few invaders, highlighting the importance of ecosystem stability in preventing invasions. Plants with different strategies exhibited different responses to environmental gradients, particularly when the variations of the primary gradient were narrowed by sub-sampling. Such differential responses of plant strategies suggest using distinct control and eradication approaches for different areas and alien plant groups.

Constraints on the permeability structure of alluvial aquifers from the poro-elastic inversion of multi-frequency P-wave velocity logs

We have explored the possibility of obtaining first-order permeability estimates for saturated alluvial sediments based on the poro-elastic interpretation of the P-wave velocity dispersion inferred from sonic logs. Modern sonic logging tools designed for environmental and engineering applications allow one for P-wave velocity measurements at multiple emitter frequencies over a bandwidth covering 5 to 10 octaves. Methodological considerations indicate that, for saturated unconsolidated sediments in the silt to sand range and typical emitter frequencies ranging from approximately 1 to 30 kHz, the observable velocity dispersion should be sufficiently pronounced to allow one for reliable first-order estimations of the permeability structure. The corresponding predictions have been tested on and verified for a borehole penetrating a typical surficial alluvial aquifer. In addition to multifrequency sonic logs, a comprehensive suite of nuclear and electrical logs, an S-wave log, a litholog, and a limited number laboratory measurements of the permeability from retrieved core material were also available. This complementary information was found to be essential for parameterizing the poro-elastic inversion procedure and for assessing the uncertainty and internal consistency of corresponding permeability estimates. Our results indicate that the thus obtained permeability estimates are largely consistent with those expected based on the corresponding granulometric characteristics, as well as with the available evidence form laboratory measurements. These findings are also consistent with evidence from ocean acoustics, which indicate that, over a frequency range of several orders-of-magnitude, the classical theory of poro-elasticity is generally capable of explaining the observed P-wave velocity dispersion in medium- to fine-grained seabed sediments