Development of a visual object tracking system using a compressive sensing framework for modeling the appearance

En el presente trabajo se aborda el problema del seguimiento de objetos, cuyo objetivo es encontrar la trayectoria de un objeto en una secuencia de video. Para ello, se ha desarrollado un método de seguimiento-por-detección que construye un modelo de apariencia en un dominio comprimido usando una nueva e innovadora técnica: “compressive sensing”. La única información necesaria es la situación del objeto a seguir en la primera imagen de la secuencia. El seguimiento de objetos es una aplicación típica del área de visión artificial con un desarrollo de bastantes años. Aun así, sigue siendo una tarea desafiante debido a varios factores: cambios de iluminación, oclusión parcial o total de los objetos y complejidad del fondo de la escena, los cuales deben ser considerados para conseguir un seguimiento robusto. Para lidiar lo más eficazmente posible con estos factores, hemos propuesto un algoritmo de tracking que entrena un clasificador Máquina Vector Soporte (“Support Vector Machine” o SVM en sus siglas en inglés) en modo online para separar los objetos del fondo de la escena. Con este fin, hemos generado nuestro modelo de apariencia por medio de un descriptor de características muy robusto que describe los objetos y el fondo devolviendo un vector de dimensiones muy altas. Por ello, se ha implementado seguidamente un paso para reducir la dimensionalidad de dichos vectores y así poder entrenar nuestro clasificador en un dominio mucho menor, al que denominamos domino comprimido. La reducción de la dimensionalidad de los vectores de características se basa en la teoría de “compressive sensing”, que dice que una señal con poca dispersión (pocos componentes distintos de cero) puede estar bien representada, e incluso puede ser reconstruida, a partir de un conjunto muy pequeño de muestras. La teoría de “compressive sensing” se ha aplicado satisfactoriamente en este trabajo y diferentes técnicas de medida y reconstrucción han sido probadas para evaluar nuestros vectores reducidos, de tal forma que se ha verificado que son capaces de preservar la información de los vectores originales. También incluimos una actualización del modelo de apariencia del objeto a seguir, mediante el reentrenamiento de nuestro clasificador en cada cuadro de la secuencia con muestras positivas y negativas, las cuales han sido obtenidas a partir de la posición predicha por el algoritmo de seguimiento en cada instante temporal. El algoritmo propuesto ha sido evaluado en distintas secuencias y comparado con otros algoritmos del estado del arte de seguimiento, para así demostrar el éxito de nuestro método.

Implementación de una Support Vector Machine en RVC – CAL para imágenes hiperespectrales

El análisis de imágenes hiperespectrales permite obtener información con una gran resolución espectral: cientos de bandas repartidas desde el espectro infrarrojo hasta el ultravioleta. El uso de dichas imágenes está teniendo un gran impacto en el campo de la medicina y, en concreto, destaca su utilización en la detección de distintos tipos de cáncer. Dentro de este campo, uno de los principales problemas que existen actualmente es el análisis de dichas imágenes en tiempo real ya que, debido al gran volumen de datos que componen estas imágenes, la capacidad de cómputo requerida es muy elevada. Una de las principales líneas de investigación acerca de la reducción de dicho tiempo de procesado se basa en la idea de repartir su análisis en diversos núcleos trabajando en paralelo. En relación a esta línea de investigación, en el presente trabajo se desarrolla una librería para el lenguaje RVC – CAL – lenguaje que está especialmente pensado para aplicaciones multimedia y que permite realizar la paralelización de una manera intuitiva – donde se recogen las funciones necesarias para implementar el clasificador conocido como Support Vector Machine – SVM. Cabe mencionar que este trabajo complementa el realizado en [1] y [2] donde se desarrollaron las funciones necesarias para implementar una cadena de procesado que utiliza el método unmixing para procesar la imagen hiperespectral. En concreto, este trabajo se encuentra dividido en varias partes. La primera de ellas expone razonadamente los motivos que han llevado a comenzar este Trabajo de Investigación y los objetivos que se pretenden conseguir con él. Tras esto, se hace un amplio estudio del estado del arte actual y, en él, se explican tanto las imágenes hiperespectrales como sus métodos de procesado y, en concreto, se detallará el método que utiliza el clasificador SVM. Una vez expuesta la base teórica, nos centraremos en la explicación del método seguido para convertir una versión en Matlab del clasificador SVM optimizado para analizar imágenes hiperespectrales; un punto importante en este apartado es que se desarrolla la versión secuencial del algoritmo y se asientan las bases para una futura paralelización del clasificador. Tras explicar el método utilizado, se exponen los resultados obtenidos primero comparando ambas versiones y, posteriormente, analizando por etapas la versión adaptada al lenguaje RVC – CAL. Por último, se aportan una serie de conclusiones obtenidas tras analizar las dos versiones del clasificador SVM en cuanto a bondad de resultados y tiempos de procesado y se proponen una serie de posibles líneas de actuación futuras relacionadas con dichos resultados. ABSTRACT. Hyperspectral imaging allows us to collect high resolution spectral information: hundred of bands covering from infrared to ultraviolet spectrum. These images have had strong repercussions in the medical field; in particular, we must highlight its use in cancer detection. In this field, the main problem we have to deal with is the real time analysis, because these images have a great data volume and they require a high computational power. One of the main research lines that deals with this problem is related with the analysis of these images using several cores working at the same time. According to this investigation line, this document describes the development of a RVC – CAL library – this language has been widely used for working with multimedia applications and allows an optimized system parallelization –, which joins all the functions needed to implement the Support Vector Machine – SVM - classifier. This research complements the research conducted in [1] and [2] where the necessary functions to implement the unmixing method to analyze hyperspectral images were developed. The document is divided in several chapters. The first of them introduces the motivation of the Master Thesis and the main objectives to achieve. After that, we study the state of the art of some technologies related with this work, like hyperspectral images, their processing methods and, concretely, the SVM classifier. Once we have exposed the theoretical bases, we will explain the followed methodology to translate a Matlab version of the SVM classifier optimized to process an hyperspectral image to RVC – CAL language; one of the most important issues in this chapter is that a sequential implementation is developed and the bases of a future parallelization of the SVM classifier are set. At this point, we will expose the results obtained in the comparative between versions and then, the results of the different steps that compose the SVM in its RVC – CAL version. Finally, we will extract some conclusions related with algorithm behavior and time processing. In the same way, we propose some future research lines according to the results obtained in this document.

Intrastriatal injection of an adenoviral vector expressing glial-cell-line-derived neurotrophic factor prevents dopaminergic neuron degeneration and behavioral impairment in a rat model of Parkinson disease

Glial-cell-line-derived neurotrophic factor (GDNF) is a potent neurotrophic factor for adult nigral dopamine neurons in vivo. GDNF has both protective and restorative effects on the nigro-striatal dopaminergic (DA) system in animal models of Parkinson disease. Appropriate administration of this factor is essential for the success of its clinical application. Since it cannot cross the blood–brain barrier, a gene transfer method may be appropriate for delivery of the trophic factor to DA cells. We have constructed a recombinant adenovirus (Ad) encoding GDNF and injected it into rat striatum to make use of its ability to infect neurons and to be retrogradely transported by DA neurons. Ad-GDNF was found to drive production of large amounts of GDNF, as quantified by ELISA. The GDNF produced after gene transfer was biologically active: it increased the survival and differentiation of DA neurons in vitro. To test the efficacy of the Ad-mediated GDNF gene transfer in vivo, we used a progressive lesion model of Parkinson disease. Rats received injections unilaterally into their striatum first of Ad and then 6 days later of 6-hydroxydopamine. We found that mesencephalic nigral dopamine neurons of animals treated with the Ad-GDNF were protected, whereas those of animals treated with the Ad-β-galactosidase were not. This protection was associated with a difference in motor function: amphetamine-induced turning was much lower in animals that received the Ad-GDNF than in the animals that received Ad-β-galactosidase. This finding may have implications for the development of a treatment for Parkinson disease based on the use of neurotrophic factors.

Stable and efficient gene transfer into the retina using an HIV-based lentiviral vector

The development of methods for efficient gene transfer to terminally differentiated retinal cells is important to study the function of the retina as well as for gene therapy of retinal diseases. We have developed a lentiviral vector system based on the HIV that can transduce terminally differentiated neurons of the brain in vivo. In this study, we have evaluated the ability of HIV vectors to transfer genes into retinal cells. An HIV vector containing a gene encoding the green fluorescent protein (GFP) was injected into the subretinal space of rat eyes. The GFP gene under the control of the cytomegalovirus promoter was efficiently expressed in both photoreceptor cells and retinal pigment epithelium. However, the use of the rhodopsin promoter resulted in expression predominantly in photoreceptor cells. Most successfully transduced eyes showed that photoreceptor cells in >80% of the area of whole retina expressed the GFP. The GFP expression persisted for at least 12 weeks with no apparent decrease. The efficient gene transfer into photoreceptor cells by HIV vectors will be useful for gene therapy of retinal diseases such as retinitis pigmentosa.

Identification and characterization of differentially expressed cDNAs of the vector mosquito, Anopheles gambiae

The isolation and study of Anopheles gambiae genes that are differentially expressed in development, notably in tissues associated with the maturation and transmission of the malaria parasite, is important for the elucidation of basic molecular mechanisms underlying vector–parasite interactions. We have used the differential display technique to screen for mRNAs specifically expressed in adult males, females, and midgut tissues of blood-fed and unfed females. We also screened for mRNAs specifically induced upon bacterial infection of larval stage mosquitoes. We have characterized 19 distinct cDNAs, most of which show developmentally regulated expression specificity during the mosquito life cycle. The most interesting are six new sequences that are midgut-specific in the adult, three of which are also modulated by blood-feeding. The gut-specific sequences encode a maltase, a V-ATPase subunit, a GTP binding protein, two different lectins, and a nontrypsin serine protease. The latter sequence is also induced in larvae subjected to bacterial challenge. With the exception of a mitochondrial DNA fragment, the other 18 sequences constitute expressed genomic sequence tags, 4 of which have been mapped cytogenetically.

Inactivation of the α C protein antigen gene, bca, by a novel shuttle/suicide vector results in attenuation of virulence and immunity in group B Streptococcus

The α C protein of group B Streptococcus (GBS) is a major surface-associated antigen. Although its role in the biology and virulence of GBS has not been defined, it is opsonic and capable of eliciting protective immunity. The α C protein is widely distributed among clinical isolates and is a potential protein carrier and antigen in conjugate vaccines to prevent GBS infections. The structural gene for the α C protein, bca, has been cloned and sequenced. The protein encoded by bca is related to a class of surface-associated proteins of Gram-positive cocci involved in virulence and immunity. To investigate the potential roles of the α C protein, bca null mutants were generated in which the bca gene was replaced with a kanamycin resistance cassette via homologous recombination using a novel shuttle/suicide vector. Studies of lethality in neonatal mice showed that the virulence of the bca null mutants was attenuated 5- to 7-fold when compared with the isogenic wild-type strain A909. Significant differences in mortality occurred in the first 24 h, suggesting that the role of the α antigen is important in the initial stages of the infection. In contrast to A909, bca mutants were no longer killed by polymorphonuclear leukocytes in the presence of α-specific antibodies in an in vitro opsonophagocytic assay. In contrast to previous studies, α antigen expression does not appear to play a role in resistance to opsonophagocytosis in the absence of α-specific antibodies. In addition, antibodies to the α C protein did not passively protect neonatal mice from lethal challenge with bca mutants, suggesting that these epitopes are uniquely present within the α antigen as expressed from the bca gene. Therefore, the α C protein is important in the pathogenesis of GBS infection and is a target for protective immunity in the development of GBS vaccines.

Interference with gene regulation in living sea urchin embryos: Transcription factor Knock Out (TKO), a genetically controlled vector for blockade of specific transcription factors

“TKO” is an expression vector that knocks out the activity of a transcription factor in vivo under genetic control. We describe a successful test of this concept that used a sea urchin transcription factor of known function, P3A2, as the target. The TKO cassette employs modular cis-regulatory elements to express an encoded single-chain antibody that prevents the P3A2 protein from binding DNA in vivo. In normal development, one of the functions of the P3A2 transcription factor is to repress directly the expression of the CyIIIa cytoskeletal actin gene outside the aboral ectoderm of the embryo. Ectopic expression in oral ectoderm occurs if P3A2 sites are deleted from CyIIIa expression constructs, and we show here that introduction of an αP3A2⋅TKO expression cassette causes exactly the same ectopic oral expression of a coinjected wild-type CyIIIa construct. Furthermore, the αP3A2⋅TKO cassette derepresses the endogenous CyIIIa gene in the oral ectoderm and in the endoderm. αP3A2⋅TKO thus abrogates the function of the endogenous SpP3A2 transcription factor with respect to spatial repression of the CyIIIa gene. Widespread expression of αP3A2⋅TKO in the endoderm has the additional lethal effect of disrupting morphogenesis of the archenteron, revealing a previously unsuspected function of SpP3A2 in endoderm development. In principle, TKO technology could be utilized for spatially and temporally controlled blockade of any transcription factor in any biological system amenable to gene transfer.

Development of HIV vectors for anti-HIV gene therapy.

Current gene therapy protocols for HIV infection use transfection or murine retrovirus mediated transfer of antiviral genes into CD4+ T cells or CD34+ progenitor cells ex vivo, followed by infusion of the gene altered cells into autologous or syngeneic/allogeneic recipients. While these studies are essential for safety and feasibility testing, several limitations remain: long-term reconstitution of the immune system is not effected for lack of access to the macrophage reservoir or the pluripotent stem cell population, which is usually quiescent, and ex vivo manipulation of the target cells will be too expensive and impractical for global application. In these regards, the lentivirus-specific biologic properties of the HIVs, which underlie their pathogenetic mechanisms, are also advantageous as vectors for gene therapy. The ability of HIV to specifically target CD4+ cells, as well as non-cycling cells, makes it a promising candidate for in vivo gene transfer vector on one hand, and for transduction of non-cycling stem cells on the other. Here we report the use of replication-defective vectors and stable vector packaging cell lines derived from both HIV-1 and HIV-2. Both HIV envelopes and vesicular stomatitis virus glycoprotein G were effective in mediating high-titer gene transfer, and an HIV-2 vector could be cross-packaged by HIV-1. Both HIV-1 and HIV-2 vectors were able to transduce primary human macrophages, a property not shared by murine retroviruses. Vesicular stomatitis virus glycoprotein G-pseudotyped HIV vectors have the potential to mediate gene transfer into non-cycling hematopoietic stem cells. If so, HIV or other lentivirus-based vectors will have applications beyond HIV infection.

Long-term expression of erythropoietin in the systemic circulation of mice after intramuscular injection of a plasmid DNA vector.

Erythropoietin (Epo)-responsive anemia is a common and debilitating complication of chronic renal failure and human immunodeficiency virus infection. Current therapy for this condition involves repeated intravenous or subcutaneous injections of recombinant Epo. In this report, we describe the development of a novel muscle-based gene transfer approach that produces long-term expression of physiologically significant levels of Epo in the systemic circulation of mice. We have constructed a plasmid expression vector, pVRmEpo, that contains the murine Epo cDNA under the transcriptional control of the cytomegalovirus immediate early (CMV-IE) promoter, the CMV-IE 5' untranslated region, and intron A. A single intramuscular (i.m.) injection of as little as 10 micrograms of this plasmid into immunocompetent adult mice produced physiologically significant elevations in serum Epo levels and increased hematocrits from preinjection levels of 48 +/- 0.4% to levels of 64 +/- 3.3% 45 days after injection. Hematocrits in these animals remained elevated at greater than 60% for at least 90 days after a single i.m. injection of 10 micrograms of pVRmEpo. We observed a dose-response relationship between the amount of plasmid DNA injected and subsequent elevations in hematocrits. Mice injected once with 300 micrograms of pVRmEpo displayed 5-fold increased serum Epo levels and elevated hematocrits of 79 +/- 3.3% at 45 days after injection. The i.m. injected plasmid DNA remained localized to the site of injection as assayed by the PCR. We conclude that i.m. injection of plasmid DNA represents a viable nonviral gene transfer method for the treatment of acquired and inherited serum protein deficiencies.

A new adenoviral vector: Replacement of all viral coding sequences with 28 kb of DNA independently expressing both full-length dystrophin and beta-galactosidase.

Adenoviral vector-mediated gene transfer offers significant potential for gene therapy of many human diseases. However, progress has been slowed by several limitations. First, the insert capacity of currently available adenoviral vectors is limited to 8 kb of foreign DNA. Second, the expression of viral proteins in infected cells is believed to trigger a cellular immune response that results in inflammation and in only transient expression of the transferred gene. We report the development of a new adenoviral vector that has all viral coding sequences removed. Thus, large inserts are accommodated and expression of all viral proteins is eliminated. The first application of this vector system carries a dual expression cassette comprising 28.2 kb of nonviral DNA that includes the full-length murine dystrophin cDNA under control of a large muscle-specific promoter and a lacZ reporter construct. Using this vector, we demonstrate independent expression of both genes in primary mdx (dystrophin-deficient) muscle cells.

Human immunodeficiency virus tat gene transfer to the murine central nervous system using a replication-defective herpes simplex virus vector stimulates transforming growth factor beta 1 gene expression.

The high incidence of neurological disorders in patients afflicted with acquired immunodeficiency syndrome (AIDS) may result from human immunodeficiency virus type 1 (HIV-1) induction of chemotactic signals and cytokines within the brain by virus-encoded gene products. Transforming growth factor beta1 (TGF-beta1) is an immunomodulator and potent chemotactic molecule present at elevated levels in HIV-1-infected patients, and its expression may thus be induced by viral trans-activating proteins such as Tat. In this report, a replication-defective herpes simplex virus (HSV)-1 tat gene transfer vector, dSTat, was used to transiently express HIV-1 Tat in glial cells in culture and following intracerebral inoculation in mouse brain in order to directly determine whether Tat can increase TGF-beta1 mRNA expression. dSTat infection of Vero cells transiently transfected by a panel of HIV-1 long terminal repeat deletion mutants linked to the bacterial chloramphenicol acetyltransferase reporter gene demonstrated that vector-expressed Tat activated the long terminal repeat in a trans-activation response element-dependent fashion independent of the HSV-mediated induction of the HIV-1 enhancer, or NF-kappaB domain. Northern blot analysis of human astrocytic glial U87-MG cells transfected by dSTat vector DNA resulted in a substantial increase in steady-state levels of TGF-beta1 mRNA. Furthermore, intracerebral inoculation of dSTat followed by Northern blot analysis of whole mouse brain RNA revealed an increase in levels of TGF-beta1 mRNA similar to that observed in cultured glial cells transfected by dSTat DNA. These results provided direct in vivo evidence for the involvement of HIV-1 Tat in activation of TGF-beta1 gene expression in brain. Tat-mediated stimulation of TGF-beta1 expression suggests a novel pathway by which HIV-1 may alter the expression of cytokines in the central nervous system, potentially contributing to the development of AIDS-associated neurological disease.

A long-range regulatory element of Hoxc8 identified by using the pClasper vector.

Hox genes are located in highly conserved clusters. The significance of this organization is unclear, but one possibility is that regulatory regions for individual genes are dispersed throughout the cluster and shared with other Hox genes. This hypothesis is supported by studies on several Hox genes in which even large genomic regions immediately surrounding the gene fail to direct the complete expression pattern in transgenic mice. In particular, previous studies have identified proximal regulatory regions that are primarily responsible for early phases of mouse Hoxc8 expression. To locate additional regulatory regions governing expression during the later periods of development, a yeast homologous recombination-based strategy utilizing the pClasper vector was employed. Using homologous recombination into pClasper, we cloned a 27-kb region around the Hoxc8 gene from a yeast artificial chromosome. A reporter gene was introduced into the coding region of the isolated gene by homologous recombination in yeast. This large fragment recapitulates critical aspects of Hoxc8 expression in transgenic mice. We show that the regulatory elements that maintain the anterior boundaries of expression in the neural tube and paraxial mesoderm are located between 11 and 19 kb downstream of the gene.

Protective immune responses induced by secretion of a chimeric soluble protein from a recombinant Mycobacterium bovis bacillus Calmette-Guérin vector candidate vaccine for human immunodeficiency virus type 1 in small animals.

A recombinant Mycobacterium bovis bacillus Calmette-Guérin (BCG) vector-based vaccine that secretes the V3 principal neutralizing epitope of human immunodeficiency virus (HIV) could induce immune response to the epitope and prevent the viral infection. By using the Japanese consensus sequence of HIV-1, we successfully constructed chimeric protein secretion vectors by selecting an appropriate insertion site of a carrier protein and established the principal neutralizing determinant (PND)-peptide secretion system in BCG. The recombinant BCG (rBCG)-inoculated guinea pigs were initially screened by delayed-type hypersensitivity (DTH) skin reactions to the PND peptide, followed by passive transfer of the DTH by the systemic route. Further, immunization of mice with the rBCG resulted in induction of cytotoxic T lymphocytes. The guinea pig immune antisera showed elevated titers to the PND peptide and neutralized HIVMN, and administration of serum IgG from the vaccinated guinea pigs was effective in completely blocking the HIV infection in thymus/liver transplanted severe combined immunodeficiency (SCID)/hu or SCID/PBL mice. In addition, the immune serum IgG was shown to neutralize primary field isolates of HIV that match the neutralizing sequence motif by a peripheral blood mononuclear cell-based virus neutralization assay. The data support the idea that the antigen-secreting rBCG system can be used as a tool for development of HIV vaccines.

Immunophenotyping of melanomas for tyrosinase: implications for vaccine development.

Tyrosinase (EC 1.14.18.1), the key enzyme in melanin synthesis, has been shown to be one of the targets for cytotoxic T-cell recognition in melanoma patients. To develop serological reagents useful for immunophenotyping melanoma for tyrosinase, human tyrosinase cDNA was expressed in an Escherichia coli expression vector. The purified recombinant tyrosinase was used to generate mouse monoclonal and rabbit polyclonal antibodies. The prototype monoclonal antibody, T311, recognized a cluster of protein moieties ranging from 70 to 80 kDa in tyrosinase mRNA-positive melanoma cell lines and melanoma specimens as well as in L cells transfected with tyrosinase cDNA. Untransfected L cells and L cells transfected with tyrosinase-related protein 1, TRP-1(gp75), were nonreactive. Immunohistochemical analysis of melanomas with T311 showed tyrosinase in melanotic and amelanotic variants, and tyrosinase expression correlated with the presence of tyrosinase mRNA. Melanocytes in skin stained with T311, whereas other normal tissues tested were negative. The expression pattern of three melanosome-associated proteins--tyrosinase, TRP-1(gp75), and gp100--in melanoma was also compared. Tyrosinase and gp100 are expressed in a higher percentage of melanomas than TRP-1(gp75), and the expression of these three antigens was discordant. Tyrosinase expression within individual tumor specimen is usually homogenous, distinctly different from the commonly observed heterogeneous pattern of gp100 expression.

Development of exfoliated layered stannosilicate for hydrogen adsorption

The use of hydrogen as an energy vector leads to the development of materials with high hydrogen adsorption capacity. In this work, a new layered stannosilicate, UZAR-S3, is synthesized and delaminated, producing UZAR-S4. UZAR-S3, with the empirical formula Na4SnSi5O14·3.5H2O and lamellar morphology, is a layered stannosilicate built from SnO6 and SiO4 polyhedra. The delamination process used here comprises three stages: protonation with acetic acid, swelling with nonylamine and the delamination itself with an HCl/H2O/ethanol solution. UZAR-S4 is composed of sheets a few nanometers thick with a high aspect ratio and a surface area of 236 m2/g, twenty times higher than that of UZAR-S3. At −196 °C for UZAR-S4, H2 adsorption reached remarkable values of 3.7 and 4.2 wt% for 10 and 40 bar, respectively, the latter value giving a high volumetric H2 storage capacity of 26.2 g of H2/L.