Image Processing Challenges in the Creation of Spatiotemporal Gene Expression Atlases of Developing Embryos

To properly understand and model animal embryogenesis it is crucial to obtain detailed measurements, both in time and space, about their gene expression domains and cell dynamics. Such challenge has been confronted in recent years by a surge of atlases which integrate a statistically relevant number of different individuals to get robust, complete information about their spatiotemporal locations of gene patterns. This paper will discuss the fundamental image analysis strategies required to build such models and the most common problems found along the way. We also discuss the main challenges and future goals in the field.

Scalable software architecture for on-line multi-camera video processing

In this paper we present a scalable software architecture for on-line multi-camera video processing, that guarantees a good trade off between computational power, scalability and flexibility. The software system is modular and its main blocks are the Processing Units (PUs), and the Central Unit. The Central Unit works as a supervisor of the running PUs and each PU manages the acquisition phase and the processing phase. Furthermore, an approach to easily parallelize the desired processing application has been presented. In this paper, as case study, we apply the proposed software architecture to a multi-camera system in order to efficiently manage multiple 2D object detection modules in a real-time scenario. System performance has been evaluated under different load conditions such as number of cameras and image sizes. The results show that the software architecture scales well with the number of camera and can easily works with different image formats respecting the real time constraints. Moreover, the parallelization approach can be used in order to speed up the processing tasks with a low level of overhead

The ciao modular, standalone compiler and its generic program processing library

Ciao Prolog incorporates a module system which allows sepárate compilation and sensible creation of standalone executables. We describe some of the main aspects of the Ciao modular compiler, ciaoc, which takes advantage of the characteristics of the Ciao Prolog module system to automatically perform sepárate and incremental compilation and efficiently build small, standalone executables with competitive run-time performance, ciaoc can also detect statically a larger number of programming errors. We also present a generic code processing library for handling modular programs, which provides an important part of the functionality of ciaoc. This library allows the development of program analysis and transformation tools in a way that is to some extent orthogonal to the details of module system design, and has been used in the implementation of ciaoc and other Ciao system tools. We also describe the different types of executables which can be generated by the Ciao compiler, which offer different tradeoffs between executable size, startup time, and portability, depending, among other factors, on the linking regime used (static, dynamic, lazy, etc.). Finally, we provide experimental data which illustrate these tradeoffs.

High Performance FPGA-oriented mersenne twister uniform random number generator

Mersenne Twister (MT) uniform random number generators are key cores for hardware acceleration of Monte Carlo simulations. In this work, two different architectures are studied: besides the classical table-based architecture, a different architecture based on a circular buffer and especially targeting FPGAs is proposed. A 30% performance improvement has been obtained when compared to the fastest previous work. The applicability of the proposed MT architectures has been proven in a high performance Gaussian RNG.

Evolution of microstructure, macrotexture and mechanical properties of commercially pure Ti during ECAP-conform processing and drawing

Long-length ultrafine-grained (UFG) Ti rods are produced by equal-channel angular pressing via the conform scheme (ECAP-C) at 200 °C, which is followed by drawing at 200 °C. The evolution of microstructure, macrotexture, and mechanical properties (yield strength, ultimate tensile strength, failure stress, uniform elongation, elongation to failure) of pure Ti during this thermo-mechanical processing is studied. Special attention is also paid to the effect of microstructure on the mechanical behavior of the material after macrolocalization of plastic flow. The number of ECAP-C passes varies in the range of 1–10. The microstructure is more refined with increasing number of ECAP-C passes. Formation of homogeneous microstructure with a grain/subgrain size of 200 nm and its saturation after 6 ECAP-C passes are observed. Strength properties increase with increasing number of ECAP passes and saturate after 6 ECAP-C passes to a yield strength of 973 MPa, an ultimate tensile strength of 1035 MPa, and a true failure stress of 1400 MPa (from 625, 750, and 1150 MPa in the as-received condition). The true strain at failure failure decreases after ECAP-C processing. The reduction of area and true strain to failure values do not decrease after ECAP-C processing. The sample after 6 ECAP-C passes is subjected to drawing at 200¯C resulting in reduction of a grain/subgrain size to 150 nm, formation of (10 1¯0) fiber texture with respect to the rod axis, and further increase of the yield strength up to 1190 MPa, the ultimate tensile strength up to 1230 MPa and the true failure stress up to 1600 MPa. It is demonstrated that UFG CP Ti has low resistance to macrolocalization of plastic deformation and high resistance to crack formation after necking.

Multilingual number transcription for text-to-speech conversion

This paper describes the text normalization module of a text to speech fully-trainable conversion system and its application to number transcription. The main target is to generate a language independent text normalization module, based on data instead of on expert rules. This paper proposes a general architecture based on statistical machine translation techniques. This proposal is composed of three main modules: a tokenizer for splitting the text input into a token graph, a phrase-based translation module for token translation, and a post-processing module for removing some tokens. This architecture has been evaluated for number transcription in several languages: English, Spanish and Romanian. Number transcription is an important aspect in the text normalization problem.

A scalable evolvable hardware processing array

Evolvable hardware (EH) is an interesting alternative to conventional digital circuit design, since autonomous generation of solutions for a given task permits self-adaptivity of the system to changing environments, and they present inherent fault tolerance when evolution is intrinsically performed. Systems based on FPGAs that use Dynamic and Partial Reconfiguration (DPR) for evolving the circuit are an example. Also, thanks to DPR, these systems can be provided with scalability, a feature that allows a system to change the number of allocated resources at run-time in order to vary some feature, such as performance. The combination of both aspects leads to scalable evolvable hardware (SEH), which changes in size as an extra degree of freedom when trying to achieve the optimal solution by means of evolution. The main contributions of this paper are an architecture of a scalable and evolvable hardware processing array system, some preliminary evolution strategies which take scalability into consideration, and to show in the experimental results the benefits of combined evolution and scalability. A digital image filtering application is used as use case.

Tag detection for preventing unauthorized face image processing

A new technology is being proposed as a solution to the problem of unintentional facial detection and recognition in pictures in which the individuals appearing want to express their privacy preferences, through the use of different tags. The existing methods for face de-identification were mostly ad hoc solutions that only provided an absolute binary solution in a privacy context such as pixelation, or a bar mask. As the number and users of social networks are increasing, our preferences regarding our privacy may become more complex, leaving these absolute binary solutions as something obsolete. The proposed technology overcomes this problem by embedding information in a tag which will be placed close to the face without being disruptive. Through a decoding method the tag will provide the preferences that will be applied to the images in further stages.

Parallel Computer Vision Algorithms for Graphics Processing Units

La evolución de los teléfonos móviles inteligentes, dotados de cámaras digitales, está provocando una creciente demanda de aplicaciones cada vez más complejas que necesitan algoritmos de visión artificial en tiempo real; puesto que el tamaño de las señales de vídeo no hace sino aumentar y en cambio el rendimiento de los procesadores de un solo núcleo se ha estancado, los nuevos algoritmos que se diseñen para visión artificial han de ser paralelos para poder ejecutarse en múltiples procesadores y ser computacionalmente escalables. Una de las clases de procesadores más interesantes en la actualidad se encuentra en las tarjetas gráficas (GPU), que son dispositivos que ofrecen un alto grado de paralelismo, un excelente rendimiento numérico y una creciente versatilidad, lo que los hace interesantes para llevar a cabo computación científica. En esta tesis se exploran dos aplicaciones de visión artificial que revisten una gran complejidad computacional y no pueden ser ejecutadas en tiempo real empleando procesadores tradicionales. En cambio, como se demuestra en esta tesis, la paralelización de las distintas subtareas y su implementación sobre una GPU arrojan los resultados deseados de ejecución con tasas de refresco interactivas. Asimismo, se propone una técnica para la evaluación rápida de funciones de complejidad arbitraria especialmente indicada para su uso en una GPU. En primer lugar se estudia la aplicación de técnicas de síntesis de imágenes virtuales a partir de únicamente dos cámaras lejanas y no paralelas—en contraste con la configuración habitual en TV 3D de cámaras cercanas y paralelas—con información de color y profundidad. Empleando filtros de mediana modificados para la elaboración de un mapa de profundidad virtual y proyecciones inversas, se comprueba que estas técnicas son adecuadas para una libre elección del punto de vista. Además, se demuestra que la codificación de la información de profundidad con respecto a un sistema de referencia global es sumamente perjudicial y debería ser evitada. Por otro lado se propone un sistema de detección de objetos móviles basado en técnicas de estimación de densidad con funciones locales. Este tipo de técnicas es muy adecuada para el modelado de escenas complejas con fondos multimodales, pero ha recibido poco uso debido a su gran complejidad computacional. El sistema propuesto, implementado en tiempo real sobre una GPU, incluye propuestas para la estimación dinámica de los anchos de banda de las funciones locales, actualización selectiva del modelo de fondo, actualización de la posición de las muestras de referencia del modelo de primer plano empleando un filtro de partículas multirregión y selección automática de regiones de interés para reducir el coste computacional. Los resultados, evaluados sobre diversas bases de datos y comparados con otros algoritmos del estado del arte, demuestran la gran versatilidad y calidad de la propuesta. Finalmente se propone un método para la aproximación de funciones arbitrarias empleando funciones continuas lineales a tramos, especialmente indicada para su implementación en una GPU mediante el uso de las unidades de filtraje de texturas, normalmente no utilizadas para cómputo numérico. La propuesta incluye un riguroso análisis matemático del error cometido en la aproximación en función del número de muestras empleadas, así como un método para la obtención de una partición cuasióptima del dominio de la función para minimizar el error. ABSTRACT The evolution of smartphones, all equipped with digital cameras, is driving a growing demand for ever more complex applications that need to rely on real-time computer vision algorithms. However, video signals are only increasing in size, whereas the performance of single-core processors has somewhat stagnated in the past few years. Consequently, new computer vision algorithms will need to be parallel to run on multiple processors and be computationally scalable. One of the most promising classes of processors nowadays can be found in graphics processing units (GPU). These are devices offering a high parallelism degree, excellent numerical performance and increasing versatility, which makes them interesting to run scientific computations. In this thesis, we explore two computer vision applications with a high computational complexity that precludes them from running in real time on traditional uniprocessors. However, we show that by parallelizing subtasks and implementing them on a GPU, both applications attain their goals of running at interactive frame rates. In addition, we propose a technique for fast evaluation of arbitrarily complex functions, specially designed for GPU implementation. First, we explore the application of depth-image–based rendering techniques to the unusual configuration of two convergent, wide baseline cameras, in contrast to the usual configuration used in 3D TV, which are narrow baseline, parallel cameras. By using a backward mapping approach with a depth inpainting scheme based on median filters, we show that these techniques are adequate for free viewpoint video applications. In addition, we show that referring depth information to a global reference system is ill-advised and should be avoided. Then, we propose a background subtraction system based on kernel density estimation techniques. These techniques are very adequate for modelling complex scenes featuring multimodal backgrounds, but have not been so popular due to their huge computational and memory complexity. The proposed system, implemented in real time on a GPU, features novel proposals for dynamic kernel bandwidth estimation for the background model, selective update of the background model, update of the position of reference samples of the foreground model using a multi-region particle filter, and automatic selection of regions of interest to reduce computational cost. The results, evaluated on several databases and compared to other state-of-the-art algorithms, demonstrate the high quality and versatility of our proposal. Finally, we propose a general method for the approximation of arbitrarily complex functions using continuous piecewise linear functions, specially formulated for GPU implementation by leveraging their texture filtering units, normally unused for numerical computation. Our proposal features a rigorous mathematical analysis of the approximation error in function of the number of samples, as well as a method to obtain a suboptimal partition of the domain of the function to minimize approximation error.

Odor space and olfactory processing: Collective algorithms and neural implementation

Several basic olfactory tasks must be solved by highly olfactory animals, including background suppression, multiple object separation, mixture separation, and source identification. The large number N of classes of olfactory receptor cells—hundreds or thousands—permits the use of computational strategies and algorithms that would not be effective in a stimulus space of low dimension. A model of the patterns of olfactory receptor responses, based on the broad distribution of olfactory thresholds, is constructed. Representing one odor from the viewpoint of another then allows a common description of the most important basic problems and shows how to solve them when N is large. One possible biological implementation of these algorithms uses action potential timing and adaptation as the “hardware” features that are responsible for effective neural computation.

Nuclear Domains Enriched in RNA 3′-processing Factors Associate with Coiled Bodies and Histone Genes in a Cell Cycle–dependent Manner

Nuclear domains, called cleavage bodies, are enriched in the RNA 3′-processing factors CstF 64 kDa and and CPSF 100 kDa. Cleavage bodies have been found either overlapping with or adjacent to coiled bodies. To determine whether the spatial relationship between cleavage bodies and coiled bodies was influenced by the cell cycle, we performed cell synchronization studies. We found that in G1 phase cleavage bodies and coiled bodies were predominantly coincident, whereas in S phase they were mostly adjacent to each other. In G2 cleavage bodies were often less defined or absent, suggesting that they disassemble at this point in the cell cycle. A small number of genetic loci have been reported to be juxtaposed to coiled bodies, including the genes for U1 and U2 small nuclear RNA as well as the two major histone gene clusters. Here we show that cleavage bodies do not overlap with small nuclear RNA genes but do colocalize with the histone genes next to coiled bodies. These findings demonstrate that the association of cleavage bodies and coiled bodies is both dynamic and tightly regulated and suggest that the interaction between these nuclear neighbors is related to the cell cycle–dependent expression of histone genes.

Subdivisions of auditory cortex and processing streams in primates

The auditory system of monkeys includes a large number of interconnected subcortical nuclei and cortical areas. At subcortical levels, the structural components of the auditory system of monkeys resemble those of nonprimates, but the organization at cortical levels is different. In monkeys, the ventral nucleus of the medial geniculate complex projects in parallel to a core of three primary-like auditory areas, AI, R, and RT, constituting the first stage of cortical processing. These areas interconnect and project to the homotopic and other locations in the opposite cerebral hemisphere and to a surrounding array of eight proposed belt areas as a second stage of cortical processing. The belt areas in turn project in overlapping patterns to a lateral parabelt region with at least rostral and caudal subdivisions as a third stage of cortical processing. The divisions of the parabelt distribute to adjoining auditory and multimodal regions of the temporal lobe and to four functionally distinct regions of the frontal lobe. Histochemically, chimpanzees and humans have an auditory core that closely resembles that of monkeys. The challenge for future researchers is to understand how this complex system in monkeys analyzes and utilizes auditory information.

Mechanisms and streams for processing of “what” and “where” in auditory cortex

The functional specialization and hierarchical organization of multiple areas in rhesus monkey auditory cortex were examined with various types of complex sounds. Neurons in the lateral belt areas of the superior temporal gyrus were tuned to the best center frequency and bandwidth of band-passed noise bursts. They were also selective for the rate and direction of linear frequency modulated sweeps. Many neurons showed a preference for a limited number of species-specific vocalizations (“monkey calls”). These response selectivities can be explained by nonlinear spectral and temporal integration mechanisms. In a separate series of experiments, monkey calls were presented at different spatial locations, and the tuning of lateral belt neurons to monkey calls and spatial location was determined. Of the three belt areas the anterolateral area shows the highest degree of specificity for monkey calls, whereas neurons in the caudolateral area display the greatest spatial selectivity. We conclude that the cortical auditory system of primates is divided into at least two processing streams, a spatial stream that originates in the caudal part of the superior temporal gyrus and projects to the parietal cortex, and a pattern or object stream originating in the more anterior portions of the lateral belt. A similar division of labor can be seen in human auditory cortex by using functional neuroimaging.

Amino-terminal protein processing in Saccharomyces cerevisiae is an essential function that requires two distinct methionine aminopeptidases.

We previously characterized a methionine aminopeptidase (EC 3.4.11.18; Met-AP1; also called peptidase M) in Saccharomyces cerevisiae, which differs from its prokaryotic homologues in that it (i) contains an N-terminal zinc-finger domain and (ii) does not produce lethality when disrupted, although it does slow growth dramatically; it is encoded by a gene called MAP1. Here we describe a second methionine aminopeptidase (Met-AP2) in S. cerevisiae, encoded by MAP2, which was cloned as a suppressor of the slow-growth phenotype of the map1 null strain. The DNA sequence of MAP2 encodes a protein of 421 amino acids that shows 22% identity with the sequence of yeast Met-AP1. Surprisingly, comparison with sequences in the GenBank data base showed that the product of MAP2 has even greater homology (55% identity) with rat p67, which was characterized as an initiation factor 2-associated protein but not yet shown to have Met-AP activity. Transformants of map1 null cells expressing MAP2 in a high-copy-number plasmid contained 3- to 12-fold increases in Met-AP activity on different peptide substrates. The epitope-tagged suppressor gene product was purified by immunoaffinity chromatography and shown to contain Met-AP activity. To evaluate the physiological significance of Met-AP2, the MAP2 gene was deleted from wild-type and map1 null yeast strains. The map2 null strain, like the map1 null strain, is viable but with a slower growth rate. The map1, map2 double-null strains are nonviable. Thus, removal of N-terminal methionine is an essential function in yeast, as in prokaryotes, but yeast require two methionine aminopeptidases to provide the essential function which can only be partially provided by Met-AP1 or Met-AP2 alone.

Processing and cryopreservation of placental/umbilical cord blood for unrelated bone marrow reconstitution.

Clinical evidence of hematopoietic restoration with placental/umbilical cord blood (PCB) grafts indicates that PCB can be a useful source of hematopoietic stem cells for routine bone marrow reconstitution. In the unrelated setting, human leukocyte antigen (HLA)-matched donors must be obtained for candidate patients and, hence, large panels of frozen HLA-typed PCB units must be established. The large volume of unprocessed units, consisting mostly of red blood cells, plasma, and cryopreservation medium, poses a serious difficulty in this effort because storage space in liquid nitrogen is limited and costly. We report here that almost all the hematopoietic colony-forming cells present in PCB units can be recovered in a uniform volume of 20 ml by using rouleaux formation induced by hydroxyethyl starch and centrifugation to reduce the bulk of erythrocytes and plasma and, thus, concentrate leukocytes. This method multiples the number of units that can be stored in the same freezer space as much as 10-fold depending on the format of the storage system. We have also investigated the proportion of functional stem/progenitor cells initially present that are actually available to the recipient when thawed cryopreserved PCB units are infused. Progenitor cell viability is measurably decreased when thawed cells, still suspended in hypertonic cryopreservative solutions, are rapidly mixed with large volumes of isotonic solutions or plasma. The osmotic damage inflicted by the severe solute concentration gradient, however, can be averted by a simple 2-fold dilution after thawing, providing almost total recovery of viable hematopoietic progenitor cells.