Search

Additional file 3 Fig. S3. Methodology for manual phenotyping of maize ears. (A) Sample ear from the biological diversity panel. (B) Sample ear from the environmental diversity panel grown under water deficit conditions showing partial pollination and/or ovary and grain abortion. Red areas: example of grains used for manual phenotyping of basal (BGC), median (MGC) and apical (AGC) number of grains per cohort, particularly difficult to characterize on "scattered" ears. To standardize the measurement, the method consisted in counting the number of rows (= lines of grains along the ear perpendicular to the cohorts) with at least one grain per row for each third of the ear. Green areas: example of grains used for manual phenotyping of the average number of cohorts per ear (CN). The measurement was repeated 4 times around the ear. White zone: fertile zone of the ear (FZ), where the surface occupied by grains visually represents more than 50% of the visible surface of the ear. Grey zones: basal (BA) and apical (AA) abortion zones, where the surface occupied by grains visually represents less than 50% of the visible surface of the ear. The lengths of the FZ, BA, and AA zones were measured manually along the main axis of the ear.

Additional file 4 Fig. S4. (A) Illustration for 3 ears of the processing used to extract ear length from ear masks. The black line represents the centerline of pixels along the main axis of the ear (vertical axis, starting from the bottom to the top of the ear), the measured ear length is the number of pixels in this line. (B) Illustration of the processing procedure to calculate the number of grains per cohort. Grain objects were reduced to a one-pixel wide vertical line along the principal axis. (C) Illustration of the processing procedure to calculate the number of grain cohorts. Grain objects were reduced to a one-pixel wide horizontal line along the ear axis perpendicular to the principal axis.

Additional file 8 Fig. S8. Comparison of Earbox (y) and reference (x) data. (A) Number of grains per cohort in the basal third. (B) Number of grains per cohort in the apical third. (C) Length of the basal aborted zone in centimeters. (D) Length of the apical aborted zone in centimeters. Green dots: ears from biological diversity panel; red dots: ears from environmental diversity panel. Empty red dots: scattered ears of the environmental diversity panel (Fig.3B). Grey line: bisector line. Black line: linear regression of data. R²: correlation coefficient between x and y values, RMSE: root mean square error, n: number of observations in each graph.

Additional file 1 Fig. S1. Choice of ear rotation angle based on Earbox specifications. (A) Heatmap of the theoretical cumulative ear percentage seen in all 6 images for roller rotation angles between 0° and 360° and diameters between 2cm and 6cm. Values are calculated assuming 120° of the ear circumferences is captured in each image. The dotted line represents the selected roller rotation angle, for which no critical diameter is encountered, capturing 100% of the information from each ear. (B) Match between theoritical and measured rotation angle for a set of ears chosen to represent diameter diversity in maize. Dots: measured and theoritical rotation angles. Colors: ear diameters. Dotted line: y = x. Solid line: linear regression. R²: correlation coefficient between x and y values, RMSE: root mean square error.

Additional file 6 Fig. S6. Output of the estimation of the number of cohorts along the perpendicular ear axis. For each position along the ear diameter (x), the number of cohorts (y) is calculated for each ear shown in Fig. 5 with the methodology presented in Supplementary Fig. S3. Black line, ear with white grains; red line, ear with vine grains; golden line, aborted ear with yellow grains. Dotted lines represent the maximum number of cohorts for each curve. The maximum number of cohorts is used as the output of the routine workflow and its average over the 6 ear sides is used for the correlation in Fig. 6. Each curve represents data from a single image/ear side.

Additional file 5 Fig. S5. Illustration of the image correction applied to project the distances and positions of the reference points onto a hypothetical circular section of the ear. Orange circle, boundaries of the ear. The reference image measurement (dref) is corrected using the horizontal distances measured between its extreme reference points (black dots) and the center of the ear (dmax and dmin). The final corrected measured is an estimate of the length of the arc resulting from the projection of dref onto a hypothetical perfect circle of radius (Rear).

Additional file 2 Fig. S2. Steps of the deep learning iteration workflow. (A) Steps of the deep learning (DL) workflow and (B,C,D) example images used for 3 typical ears representing the problems encountered and the manual correction performed. (B) Typical ear with good empirical grain masks used directly for training the first iteration of deep learning (B2) and requiring minimum manual corrections (B3 to B4) to produce the routine segmentation method (B5). (C, D) Typical ears with false detections and erroneous grain segmentations from the empirical segmentation (C2, D2) and requiring major manual corrections (step 3 to 4) used for training the second deep learning iteration to produce the routine segmentation method (C5, D5). Orange boxes, image acquisition. White boxes, image, or data processing. Green boxes processed images or data. Green areas, manual grain mask corrections.

Additional file 9 Fig. 9. Examples of grain dimensions as a function of cohort and position along the ear across 9 contrasting ears. Each point represents the average dimension of all grains classified in the same cohort (Fig. 5P, 5Q, 5R) across the 6 sides (images) of the ear. Red dots, average grain height in centimeters. Green dots, average grain width in centimeters. Error bars, standard deviation. Errors for position from ear base are shown but are smaller than the dots.

Additional file 7 Fig. S7. Illustration of the cohort classification algorithm. Grains are classified in cohorts (each identified by a color) by scanning the ear from bottom to top starting with the lowest grain (bold black dots on the barycenter). The classification is done sequentially, one cohort at a time. For each cohort, grains whose barycenter lie within a specific range of pixels along the main axis of the ear are classified into a common cohort, and removed for the rest of the classification process. Black dots: grain barycenter’s. Colors: grains classified in the same cohort.

In this paper, we describe a new wavelet-based data processing that performs simultaneously compression and fast multidimensional interpolation of hemispherical angular features. This algorithm is applied for display aspect simulation to allow predicting the rendering of any content on a given display under any illuminations conditions. Such rendering is based on a complex set of data composed of emitting display properties as a function of angle as well as its reflective behavior (BRDF). This record was migrated from the OpenDepot repository service in June, 2017 before shutting down., {"references":["In this paper, we describe a new wavelet-based data processing that performs simultaneously compression and fast multidimensional interpolation of hemispherical angular features. This algorithm is applied for display aspect simulation to allow predicting the rendering of any content on a given display under any illuminations conditions. Such rendering is based on a complex set of data composed of emitting display properties as a function of angle as well as its reflective behavior (BRDF)."]}

Proceedings of Acoustics Week in Canada 1998, October 28-30, 1998, London, Ontario, Canada

The continual need for more accurate and effective techniques in radiation therapy makes it necessary to devise new control means combining high spatial resolution as well as high dose accuracy. Intensity modulated radio therapy (IMRT) allows highly conformed fields with high spatial gradient and therefore requires a precise monitoring of all the multileaf positions. In response to this need, the authors have developed a new 2D tissue equivalent dosimeter with high spatial resolution. A plastic scintillator sheet is sandwiched between two polystyrene blocks and the emitted light is captured by a high resolution camera. A newly developed procedure described herein allows efficient discrimination of the scintillation from the parasitic Čerenkov radiation. This processing is applied on the cumulated image from a sequence of images taken during an irradiation field at a rate of 10 images/s. It provides a high resolution mapping of the cumulated dose in quasireal time. The dosimeter is tissue equivalent (ICRU-44) and works both for electrons and photons without complex parameter adjustment since phantom and detector materials are identical. Instrument calibration is simple and independent of the irradiation conditions (energy, fluence, quality, ...). In this article, the authors present the principle of the dosimeter and its calibration procedure. They compare the results obtained for photons and electron beams with ionization chamber measurements in polystyrene. Technical specifications such as accuracy and repeatability are precisely evaluated and discussed. Finally, they present different IMRT field measurements and compare DOSIMAP measurements to TPS simulations and dosimetric film profiles. The results confirm the excellent spatial resolution of the instrument and its capacity to inspect the leaf positions for each segment of a given field.

New radiation therapy techniques such as IMRT present significant efficiency due to their highly conformal dose distributions. A consequence of the complexity of their dose distributions (high gradients, small irradiation fields, low dose distribution, ...) is the requirement for better precision quality assurance than in classical radiotherapy in order to compare the conformation of the delivered dose with the planned dose distribution and to guarantee the quality of the treatment. Currently this control is mostly performed by matrices of ionization chambers, diode detectors, dosimetric films, portal imaging, or dosimetric gels. Another approach is scintillation dosimetry, which has been developed in the last 15 years mainly through scintillating fiber devices. Despite having many advantages over other methods it is still at an experimental level for routine dosimetry because the ?erenkov radiation produced under irradiation represents an important stem effect. A new 2D water equivalent scintillating dosimeter, the DosiMap, and two different Cerenkov discrimination methods were developed with the collaboration of the Laboratoire de Physique Corpusculaire of Caen, the Comprehensive Cancer Center François Baclesse, and the ELDIM Co., in the frame of the MAESTRO European project. The DosiMap consists of a plastic scintillating sheet placed inside a transparent polystyrene phantom. The light distribution produced under irradiation is recorded by a CCD camera. Our first Cerenkov discrimination technique is subtractive. It uses a chessboard pattern placed in front of the scintillator, which provides a background signal containing only Cerenkov light. Our second discrimination technique is colorimetric. It performs a spectral analysis of the light signal, which allows the unfolding of the Cerenkov radiation and the scintillation. Tests were carried out with our DosiMap prototype and the performances of the two discrimination methods were assessed. The comparison of the dose measurements performed with the DosiMap and with dosimetric films for three different irradiation configurations showed discrepancies smaller than 3.5% for a 2 mm spatial resolution. Two innovative discrimination solutions were demonstrated to separate the scintillation from the Cerenkov radiation. It was also shown that the Dosi- Map, which is water equivalent, fast, and user friendly, is a very promising tool for radiotherapy quality assurance.

National audience

National audience

International audience; no abstract

Le rôle des facteurs génétiques de l'hôte dans le contrôle de l'infection plaustre a été clairement établi dans les modèles animaux. Chez l'homme, des résultats de plus en plus nombreux sont en faveur d'un contrôle génétique du paludisme. Dans une population de la ville de Bobo-Dioulasso (Burkina Faso), nous avons mis en évidence, par une analyse de ségrégation réalisée sur le phénotype charge parasitaire sanguine, l'existence d'un génie majeur et sont donc en faveur d'un contrôle polygénique de la charge parasitaire. L'influence de cet effet génétique est plus importante chez les enfants que chez les adultes. Une analyse de liaison a alors été réalisée chez les enfants et a mis en évidence une liaison entre le phénotype et la région q31-33 du chromosome 5. Dans cette région est localisé un gène que nous avons dénommé Pfil 1 (#Plasmodium falciparum$ infection levels 1) qui explique presque 50% de la variance de la charge parasitaire sanguine et joue un rôle fondamental dans le contrôle de l'infection. Cette région 5q31-33 contient de nombreux gènes codant pour des cytokines régulant les lymphocytes T. L'identification des gènes contrôlant l'infection palustre ouvre de nouvelles voies pour le développement de stratégies préventives et thérapeutiques. Il devrait être possible dans un proche avenir d'identifier les individus à risque, individus qui pourront bénéficier en priorité des mesures préventives et thérapeutiques. Enfin, une meilleure connaissance de l'effet de ces gènes contrôlant les réponses immunitaires protectrices est d'un intérêt majeur pour le développement de vaccins. (Résumé d'auteur)

International audience; no abstract