Puigt G.,European Center for Research and Advanced Training in Scientific Computation |
Auffray V.,ITK |
Muller J.-D.,Queen Mary, University of London
Journal of Computational Physics
The main approaches of discretising the viscous operator of fluid flow on hybrid meshes are analysed for accuracy, consistence, monotonicity and sensitivity to mesh quality. As none of these approaches is fully satisfactory, a novel method using an approximated finite-element approach is presented and analysed. The methods are compared for the linear heat equation and the Navier-Stokes equations. While the novel approximated finite-element method performs significantly better for the linear heat equation, a stabilised edge-based method performs equally well for the considered test-cases for the Navier-Stokes equations. © 2009 Elsevier Inc. All rights reserved. Source
Itk | Date: 2010-01-15
A method for supervising the treatment of a crop combines a field record, models of plants, diseases or insects and active substances coupled together, and includes, at least once, the following steps:
Garin G.,ITK |
Garin G.,French National Institute for Agricultural Research |
Fournier C.,French National Institute for Agricultural Research |
Andrieu B.,French National Institute for Agricultural Research |
And 4 more authors.
Annals of Botany
Background and Aims Sustainable agriculture requires the identification of new, environmentally responsible strategies of crop protection. Modelling of pathosystems can allow a better understanding of the major interactions inside these dynamic systems and may lead to innovative protection strategies. In particular, functional-structural plant models (FSPMs) have been identified as a means to optimize the use of architecture-related traits. A current limitation lies in the inherent complexity of this type of modelling, and thus the purpose of this paper is to provide a framework to both extend and simplify the modelling of pathosystems using FSPMs. Methods Different entities and interactions occurring in pathosystems were formalized in a conceptual model. A framework based on these concepts was then implemented within the open-source OpenAlea modelling platform, using the platform's general strategy of modelling plant-environment interactions and extending it to handle plant interactions with pathogens. New developments include a generic data structure for representing lesions and dispersal units, and a series of generic protocols to communicate with objects representing the canopy and its microenvironment in the OpenAlea platform. Another development is the addition of a library of elementary models involved in pathosystem modelling. Several plant and physical models are already available in OpenAlea and can be combined in models of pathosystems using this framework approach. Key Results Two contrasting pathosystems are implemented using the framework and illustrate its generic utility. Simulations demonstrate the framework's ability to simulate multiscaled interactions within pathosystems, and also show that models are modular components within the framework and can be extended. This is illustrated by testing the impact of canopy architectural traits on fungal dispersal. Conclusions This study provides a framework for modelling a large number of pathosystems using FSPMs. This structure can accommodate both previously developed models for individual aspects of pathosystems and new ones. Complex models are deconstructed into separate 'knowledge sources' originating from different specialist areas of expertise and these can be shared and reassembled into multidisciplinary models. The framework thus provides a beneficial tool for a potential diverse and dynamic research community. © The Author 2014. Source
Crawled News Article
In 1956, Aldous Huxley and psychiatrist Humphry Osmond were in the throes of a multi-year struggle to come up with a term to describe the mind-altering effects of mescaline when Osmond suggested the neologism “psychedelic,” drawing its roots from the Greek “psyche” (mind) and “delos” (manifest, reveal). Since then, the word has been used to characterize drugs, works of art, and even portraits of Pluto. However, there remains an area—one where the term could be quite literally applied—that has generally been overlooked as being psychedelic work. As a researcher at the University of Southern California's Laboratory of Neuro Imaging (LONI), I spend large parts of my day staring at brain images that are as beautiful as they are data-ful. The images are generated from a wide spectrum of brain mapping data, ranging from diffusion MRI tractography via the Human Connectome Project, to genetic brain maps via the ENIGMA Consortium, and injected neural tracers via the Mouse Connectome Project. Although the data visualization tools used in these studies can vary widely in terms of modality and specific goals, the resultant images are all critical to the understanding and dissemination of results… while being psychedelic as hell. If the term “psychedelic” in its original conception means “mind manifesting,” then the images generated in brain mapping studies are the essence of psychedelia. If we expand the definition to include the “trippy” aesthetic typically referred to as psychedelic, then the algorithmic fractals and rainbow-hued representations of neural connections would not seem out of place in a record shop in the Haight. In a sense, “psychedelic” doesn’t feel adequate as a descriptor; that is, to the extent that the images are literal manifestations of the mind generated using psychedelic colors and motifs, they can be conceived of as being “meta-psychedelic.” Of course, the images used in the studies we work on at LONI are more than just pretty pictures, and any “trippy” qualities are unintended byproducts of trying to show the largest amount of data possible in two dimensions. For this article, however, I purposely kicked the psychedelia up a notch and turned some already beautiful brain pictures into some seriously meta-psychedelic GIFs, with captions explaining the meaning behind all the rainbows tubes and convoluted mesh lattices. What it shows: A slice of the corpus callosum, a thick band of nerve fibers connecting the two hemispheres of the brain. Details: This image was generated using diffusion-weighted MRI, which uses magnetic gradients to measure the non-random movement of water molecules along axons. The movement of water molecules in close proximity to neural pathways is restricted parallel to axons, as opposed to the movement of water molecules in other regions of the brain, which is unrestricted in all directions. The unrestricted flow in other regions is known as isotropic movement, and the restricted flow near neural pathways is known as anisotropic movement. Anisotropic movement of water molecules can be detected through pulsed magnetic gradients using diffusion-weighted MRI and measured to quantify the degree and direction(s) of restricted movement. These data are collected for roughly one million “voxels” (like 3D pixels) per brain, and the diffusion information within each voxel can be modeled with a mathematical function. In the early days of diffusion imaging, the directional information was modeled using a tensor (“diffusion tensor imaging”), which represented all of the diffusion data within each voxel as a single ellipsoid. However, this method proved to be problematic because each voxel is 1-3 cubic millimeters, which is large enough to contain several different populations of nerve fibers all coursing in different directions. Because a solitary ellipsoid can’t represent the directional information of perpendicular fiber bundles (the tensor would be more like a sphere in that case), a new way of modeling diffusion data—the fiber orientation distribution—was developed. Rather than modeling the diffusion data as a single ellipsoid, the fiber orientation distribution (FOD, see next GIF) has multiple arms that enable it to represent multiple fiber directions within each voxel (see the progression from top to bottom in this image). Once the FODs are generated, a probabilistic algorithm is used to more-or-less “connect the dots” between FODs and map out the pathways of axons, a process known as tractography. I use the workflow software LONI Pipeline to run tractography, which provides a graphical user interface for the parallel processing of many brains at the same time. The workflow requires inputs known as “seed regions” and “regions of interest” (the regions of the brain we’re trying to connect), and allows for fine-tuning of variables such as angles of curvature, lengths of tracts, and thresholds to stop tracking. What it shows: These balloon-animal-looking things are individual FODs from voxels in the optic chiasm, which is where half of the nerve fibers from each eye cross to the opposite hemisphere. Details: As explained in the first GIF, the FOD is a way of modeling the strength and direction of anisotropic flow of water within a voxel, i.e. the non-random movement of water molecules along the length of axons. The “arms” of each FOD represent the probability of fiber tracts along a given direction. These particular FODs were generated using a method that can handle a large number of diffusion gradients and an arbitrary number of gradient strengths (b-values). The optic chiasm was chosen for this example because it’s a classic location of “crossing fibers,” i.e. two bundles of axons that cross paths in the same voxel. The old method of modeling fibers using tensor models wouldn’t be able to resolve both directions within the same voxel; it would just come out looking like a blob. With FODs, the directions of fibers are sharp and able to traced with a tractography algorithm with much higher anatomical accuracy. What it shows: the optic chiasm, after tractography has been performed using the FODs generated in the previous GIF. Details: This image was generated in a similar way as the first GIF, i.e. using a probabilistic tractography algorithm to connect FODs and model the most likely paths of brain fibers traveling through the optic chiasm. What it shows: One side of the corticospinal tract (the pathway responsible for voluntary movement) connecting the brain to the spinal cord, superimposed onto the FODs used to track the pathway. Details: The background of this GIF is a screenshot of all of the FODs in one coronal slice of the brain. Using these FODs, probabilistic tractography was run to connect the primary motor cortex to the pyramidal tract in the brainstem. The primary motor cortex was automatically located using a software called FreeSurfer. The pyramidal tract was located manually in the brainstem by yours truly, using a software called ITK-SNAP. What it shows: Different methods of mapping the surface of the hippocampus Details: Surface mapping allows for the detection of changes in brain structures. This GIF demonstrates advances in surface mapping technology that use a distributed network-based model of surfaces. This allows for a highly accurate automated analysis of a large number of subjects. What it shows: Changes in the hippocampus with (a) early mild cognitive impairment, (b) late mild cognitive impairment, and (c) Alzheimer’s disease. Details: This GIF uses the technology demonstrated in the previous GIF to show how the hippocampus changes as Alzheimer’s disease (AD) progresses. The hippocampus is highly implicated in the pathophysiology of AD and is known to decrease in size over the course of the disease. The new mapping technology allows for earlier and more accurate identification of hippocampal changes, which could lead to earlier diagnosis and improved monitoring of treatment efficacy. What it shows: This is a rendering of the surface of the hippocampus that I thought looked particularly trippy, hence the name Trippocampus. What it shows: The pathway from the optic chiasm (see third GIF) to the lateral geniculate nucleus and the primary visual cortex. The top of the GIF (optic chiasm) is anterior, bottom (primary visual cortex) is posterior. Details: This GIF comes from my research on automated methods of mapping the visual pathway using high-resolution diffusion imaging. It was created using the techniques described in the first three GIFs, i.e. fiber orientation distributions and probabilistic tractography. The main challenge in mapping the pathway lies in getting the probabilistic tracking algorithm to navigate the sharp curve of Meyer’s loop (the portion of optic radiation fibers that briefly course anteriorly around the temporal horn before continuing to travel posteriorly to the visual cortex) because the algorithm tends to continue along the path of least resistance, i.e. a straight line. To overcome this, we developed a two-stage tracking method that began tractography first in the optic chiasm, and then used these results to automatically create a second starting location in the lateral geniculate of the thalamus (where Meyer’s loop begins). What it shows: An anterior-to-posterior “flight” between the two sides of the corpus callosum Details: This GIF is the whole-brain version of the first GIF, which was only a small coronal slice of the corpus callosum. It was created using the visualization software TrackVis, which has the ability to automatically take screenshots along a given path. If you’ve ever wondered what it was like to fly between the two hemispheres of your brain (on acid?), now you know.
Guaus A.,French National Institute for Agricultural Research |
Bsaibes A.,ITK |
Cartailler T.,ITK |
Prieur C.,Joseph Fourier University |
And 3 more authors.
Precision Agriculture 2013 - Papers Presented at the 9th European Conference on Precision Agriculture, ECPA 2013
The global sensitivity analysis of a dynamic soil water balance model embedded in a Decision Support System for vineyard water management is achieved via the Sobol variance-based method. The sensitivity analysis is applied sequentially at each simulation step so that the variation of parameter influence over time can be followed. Results allow identification of four soil-related parameters having the highest influence at the vine plot scale, and for various climate scenarios. This provides fundamental information for the operational use of the model, i.e. when few input data are available to the end-user. Source