EP1440413A1 - Dispositif et methode de traitement d'image pour detection de lesions evolutives - Google Patents
Dispositif et methode de traitement d'image pour detection de lesions evolutivesInfo
- Publication number
- EP1440413A1 EP1440413A1 EP02803429A EP02803429A EP1440413A1 EP 1440413 A1 EP1440413 A1 EP 1440413A1 EP 02803429 A EP02803429 A EP 02803429A EP 02803429 A EP02803429 A EP 02803429A EP 1440413 A1 EP1440413 A1 EP 1440413A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- model
- image
- time
- temporal
- sets
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30016—Brain
Definitions
- the invention relates to image processing, especially medical.
- Image processing makes it possible in particular to compare images with one another. According to certain current techniques, the images are compared two by two. Other techniques make it possible to compare the images of a series. Depending on the context, different categories of problems arise to make the images comparable, as we will see.
- medical images can be analyzed temporally and automatically so as to establish a detection of zones with pathological evolution.
- Automatic temporal analysis techniques are particularly used to provide aid in the diagnosis of a patient.
- the performance of the temporal and automatic analysis techniques currently used is limited, in particular in the case of the detection of areas with pathological evolution.
- these techniques are ill-suited for example to quantify a posteriori the effect of drugs administered in the case of therapeutic trials. More generally, it involves detecting, from one image to another, fine changes in areas with pathological changes. It's about being able to do it over time-spaced shots, and also on a phenomenon in progress, that is to say the beginning of which has not been observed.
- the present invention comes to improve the situation.
- the invention relates to an image processing device, comprising:
- each data item comprising a position component and an intensity component
- pre-processing means for modifying the data sets so as to obtain geometrically and intensity-adjusted images
- - comparative processing means suitable for examining sets of temporal sequences of image elements, in order to detect signs of variations therein.
- the processing means comprise:
- the invention relates to an image processing method comprising the following steps: a- receiving a time series of data sets representing comparable digital volume images, each data comprising a position component and an intensity component, b- modifying the data sets so as to obtain geometrically and intensity adjusted images,
- the method comprises the additional steps c- adjusting a parametric model separately on at least some of the sets of temporal sequences of picture elements, which provides pairs of information of the kind: element image, time, d- isolate those of the pairs of information which are representative of a significant variation by statistical analysis.
- FIG. 1a illustrates a computer device comprising an image processing device according to the prior art
- FIG. 1b illustrates a computer device comprising an image processing device according to the invention
- FIG. 2a represents a flowchart of the automatic image analysis method
- FIG. 2b represents a detailed part of the flow diagram of FIG. 2a
- FIG. 2c represents a detailed part of the flow chart of FIG. 2a according to a particular embodiment
- FIG. 3a is a brain image representing an alternation of cubes of a first 3D image and of a second 3D image having a time bias with the first image
- FIG. 3b is a brain image representing an alternation of cubes of the first 3D image of FIG. 3a and of the second corrected 3D image
- FIG. 3c represents the histogram linked to FIG. 3a, this histogram representing the intensities of the first image with respect to the intensities of the second image,
- FIG. 3d represents the histogram linked to FIG. 3b, this histogram representing the intensities of the first image with respect to the intensities of the second corrected image
- FIG. 4a is a brain image representing a first 3D image
- FIG. 4b is a brain image representing a second 3D image having a spatial offset and a time bias with the first image
- FIG. 4c is a brain image representing the second 3D image of FIG. 4b whose spatial offset has been corrected
- FIG. 4d is a brain image representing the second 3D image of FIG. 4c whose temporal bias has been corrected
- FIGS. 5a, 5b, 5c, 5d, 5e, 5f are intensity profiles I evolving over time T from six selected brain points
- FIG. 5g is an example of the parametric model with certain parameters
- FIG. 5i is the average intensity profile model of FIG. 5h superimposed on the normalized discrete measurements
- FIG. 5j is a discretization of the average intensity profile model of FIG. 5i
- FIGS. 6a to 6f represent intensity profiles I on which the rise times are calculated
- FIG. 7 is a schematic representation of the general linear model
- FIG. 8 shows a distribution graph of the sizes of voxel groups according to the particular embodiment.
- image processing in question requires pre-processing.
- image preprocessing techniques are first applied to the images to be analyzed. These techniques, presented below, aim to make the images comparable to one another and to allow the analysis of these images.
- the approximation of images generally requires two preprocessing operations. These operations allow normalization (or calibration) of the images in order to compare them. In medical imaging, these two pre-processing operations apply in the case of a comparison of images from patient acquisitions.
- a first operation is a geometric registration of the images with respect to a reference image.
- this registration can be rigid, that is to say that it does not change the shapes, or else refines, where it will modify the shapes either according to a certain number of degrees of freedom or a more general transformation .
- these are generally three-dimensional images, the volume element of which is also called "voxel”. However, it can also be two-dimensional images.
- the images can present "spatial bias” and “temporal bias”.
- the image may have a “spatial bias”, that is to say that in some of the images there will be areas whose intensity will be greater, that is to say brighter (lighter) areas ) and areas whose intensity will be less important, that is to say areas that are less bright (darker).
- This "spatial bias" can be treated, for example as described in:
- Certain imaging techniques allow a calibration which can be described as absolute, in the sense that it is possible to find the same levels of brightness and contrast in images from examinations spaced several weeks apart, or even several months. This is the case in certain medical imaging techniques for example. Other techniques, which do not allow this absolute calibration, will therefore provide images tainted with a "temporal bias". This is the case, in particular, with nuclear magnetic resonance imaging (MRI) machines.
- MRI nuclear magnetic resonance imaging
- the invention aims in particular to improve the pre-processing of the temporal bias, this sub-characteristic of the invention has an interest as such and is capable of being claimed separately. This improvement is not the main characteristic of the invention, the latter relating more to the temporal, quantitative and automatic analysis of the images.
- the automatic temporal analysis of medical images makes it possible to establish a posteriori a detection of zones with pathological evolution, for example in the case of lesions of multiple sclerosis. This detection is particularly useful for a posteriori monitoring of the effect of certain drugs on the body in the case of therapeutic trials.
- Methods of temporal analysis of medical images have been proposed in the following works: - Patent EP 0 927 405 "Electronic image processing device for the detection of dimensional variations", Jose Calmon, Jean-Philippe Thirion,
- This automatic temporal analysis technique makes it possible to detect areas with pathological evolution over a series of images of a patient.
- this technique is limited to taking into account temporal information for each unit of volume considered (called voxel), for example the temporal intensity profile for each unit of volume.
- a descriptive statistic application of mean, standard deviation
- fuzzy logic are applied to this time information.
- a display screen 1 for example a computer screen, is connected to an operating system 3, itself connected to an image processing device 5.
- This device 5 comprises a data memory 6, a module preprocessing 8 and a comparative processing module 10.
- the data memory 6 is suitable for storing data of image series in 3D, for example voxel data represented by a spatial position of spatial coordinates (x, y , z) for each voxel and an intensity I (x, y, z, t x ) at a given time t for each voxel.
- j is an integer that can vary by 1 year, n being an integer representing the number of different instants at which 3D images of the same patient are acquired.
- These series of 3D images can be series of images of the same patient taken using the MRI imaging technique. As the images are acquired at distant moments in time (a week or more), the patient does not take the same position.
- images 4a and 4b respectively representing a first image acquired at a given instant and a second image acquired at a different instant having a spatial offset with respect to the first image 4a.
- a coordinate (x, y, z) of a voxel of an image does not correspond to the same anatomical point on all the images of the series.
- medical image acquisition techniques can introduce inhomogeneities in the image itself (spatial bias) or between the images acquired at different times (temporal bias) in each image and / or between the images d 'a series. This is illustrated by images 4a and 4c respectively representing a first image acquired at a given instant and a second image, acquired at a different instant, aligned with the first image but having a time bias with respect to the first image 4a.
- the pre-processing module 8 is adapted to perform different pre-treatments:
- FIG. 4c which represents the image 4b aligned in FIG. 4a
- FIG. 4d represents the image 4c corrected with respect to FIG. 4a.
- the comparative processing module 10 is suitable, after image pre-processing, for an automatic temporal analysis of the images, in particular brain images in the case of multiple sclerosis.
- This module of comparative processing 10 is capable of operating according to one of the techniques of the works cited above.
- FIG. 1b is an embodiment of the device according to the invention.
- an image processing device 11 Associated with the display screen 1 and the operating system 3, an image processing device 11 comprises a data memory 16, a pre-processing module 18 and a comparative processing module 20.
- the data memory 16 is able to store, for example in memory, for each 3D image formed of voxels of dimensions 1x1x3 mm, 54 sections of 2D images having as format 256 * 256.
- the pre-treatment module 18 corresponds to the pre-treatment module 8 of the prior art.
- the preprocessing module 18 includes a time bias correction function for MRI voxel intensities based on an algorithm allowing the correction of the joint histogram between two images. This histogram represents the intensities of one image compared to the intensities of the other.
- the algorithm is based on the search for a regression line by the robust orthogonal least squares method, method developed in the publication "CRC Concise Encyclopedia of Mathematics", E. eisstein. CRC Press
- the correction is obtained by applying the transformation thus found to the second image.
- image 3a is formed of alternating cubes of a first 3D image of reference and a second image having a time bias with respect to the first reference image.
- image 3c it represents the joint histogram corresponding to image 3a.
- image 3b presents only a difference between the first reference image and the second corrected image, difference represented by cube B.
- the 3d histogram corresponds to this image 3b representing a comparison between the reference image and the second corrected image.
- the comparative processing module 20 comprises a modeling function 12 and a statistical analysis function 14.
- the modeling function 12 is able to model a curve representing the intensity profile of voxels of evolving zones over time. This function is based on a parametric mathematical model which can represent the shape of the different intensity profiles.
- the statistical analysis function 14 makes a statistical inference. According to this statistical inference, it is possible to determine clusters of points or voxels according to "significantly", in the statistical sense, a model of pathological evolution.
- a second mode preferred embodiment is also described below.
- such a curve typically includes a rising part and a falling part over time.
- a semi-manual method is used to calculate an average asymmetric Gauss type curve model, as shown in FIG. 5g, by fixing the five parameters.
- This average model will be indifferently designated as the average intensity profile model or the parametric average model.
- the discrete temporal sequence of intensity values I (x, y, z, t t ) measured at certain points (x, y, z) of space is approximated at best.
- an operator can select, on the previously corrected 3D images, a set of points corresponding to parts of progressive lesions.
- FIGS. 5a to 5f illustrate six intensity profiles evolving over time from six selected points.
- Profiles corresponding to a "rise time” of less than a week or more than ten weeks are rejected, not corresponding to an intensity profile considered to be pathological.
- Profiles 6 e and 6 f are outliers and are therefore rejected.
- the selected profiles are centered and standardized according to their respective maximum.
- the average model is then calculated according to the normalized profiles and according to a classical least squares estimate.
- the five parameters are calculated according to a least squares estimate using a Powell quadratic convergence method, presented in the following work: [1] - WH Press, SA Teukolsky, .T. Vetterling, andB.P. Flannery. Numerical Recipes. The Art of Computing. Cambridge University Press, 2nd edition, 1997.
- FIG. 5i represents the average model superimposed on the normalized discrete measurements.
- FIG. 5j represents a discretization of the average model obtained.
- the general linear model makes it possible to search for the presence of the average model of intensity profile in any voxel of the image.
- the general linear model is described in the following work:
- each voxel is adjusted according to a general linear model as presented in appendix Ib.
- Figure 7 illustrates this general linear model.
- Each column of a vector of dimension nxx, x being an integer is composed of n components each referring to a discrete instant on the time scale ti.
- Each column of a vector thus forms a profile of discrete values as a function of time.
- the vector Y (dimension nxl) represents the original discrete data of the intensity profile of a given voxel.
- the matrix X (dimension n ⁇ 2) is called the design matrix.
- X has for co-health, as illustrated in figure 7, the discrete values of the average model of intensity profile, represented by the vector XI (dimension n ⁇ l), and a constant value over time, represented by the vector X2 (dimension nxl), to take into account the average intensity of the voxel over time ti.
- the vectors X and Y are known.
- the components ⁇ l and ⁇ 2 of the vector ⁇ (2 ⁇ l) respectively represent the adequacy variable of the average intensity profile model and of the reference level of the intensity profile of a given voxel.
- the vector e represents the vector of the residue terms.
- the vectors ⁇ represent the estimated parameters of the general linear model.
- ⁇ 2 represents the residual variance estimated by the residual mean square.
- the vector c defined in appendix Ic, makes it possible to choose the variable ⁇ 1 of the vector ⁇ .
- the values of the variable ⁇ l are divided by the residual variance.
- Voxels having significantly large values of t on the map of t obtained are selected when their value of t is greater than a selected threshold value Th. Among these selected voxels appear groups of neighboring voxels.
- This theory makes it possible to determine the probability of obtaining groups of voxels of a certain size.
- the groups of voxels having a very low probability of occurrence are considered to be "significant", that is to say, in the case of a search for pathological zones, these groups of voxels are likely to represent a progressive pathological zone . Thanks to this statistical analysis, we can associate a probability to the size of a set of neighboring voxels. Only the sets of neighboring voxels for which this probability of occurrence is sufficiently low are considered significant.
- a very low probability of occurrence may be a probability of 0.01 for example.
- the curves representing the intensity profile of voxels of changing zones over time are modeled by a chosen parametric model.
- the parametric model of appendix I-a presenting an asymmetric Gauss curve with five parameters is chosen.
- a parametric model is adjusted in each voxel on a temporal intensity profile such as those shown in Figures 5a to 5f.
- the parameters of a model are determined for each voxel. It can be difficult to determine all the parameters of a model linked to an intensity profile at the same time.
- These parameters are in fact estimated on a set of profiles normalized in time and in amplitude. Once these parameters have been set, the remaining parameters are evaluated for each voxel, such as for example the maximum amplitude pi, the minimum amplitude parameter p2, and the time parameter at peak p3.
- FIGS 2a and 2b illustrate the automatic image analysis method according to the invention.
- step 110 the 3D images of a patient over time are aligned with a reference image by rigid registration as previously explained.
- a time correction is made on the 3D images, this correction concerns the intensity bias.
- a spatial correction is possibly performed.
- Steps 130 to 150 present general steps particularized by the flow diagram of FIG. 2b.
- a parametric model of the asymmetric Gaussian type comprising 5 parameters is prepared in step 130 to model the intensity profiles of the voxels of the 3D images.
- the parametric model is adjusted to the intensity profiles of the voxels of the 3D images in step 140.
- one (or even several) map making it possible to determine the groups of significantly pathological voxels is developed in step 150
- the pathological zones can be detected by statistical inference in step 160.
- FIG. 2b more particularly illustrates steps 130 to 150 of the method.
- a parametric average model is first developed according to a set of voxel intensity profiles.
- this set of intensity profiles of voxels is chosen by an operator as being a set of profiles representative of voxels presenting progressive lesions.
- the five parameters of the parametric average model are thus fixed.
- the parameters of the general linear model are estimated in each voxel of the brain.
- each intensity profile of a voxel presented in the form of a data vector is represented by a linear combination of the parametric average model, of a constant and of the residues.
- step 152 a value calculation of t is performed for each voxel, that is to say that a card is obtained in t.
- step 154 the map of t makes it possible to determine the groups of voxels having a value greater than a threshold value Th.
- FIG. 2c more particularly illustrates steps 130 to 150 of the method according to a second embodiment.
- a parametric model is chosen, for example that of appendix I-a.
- this model is determined independently for each voxel of a series of patient images.
- the parameters of the model are adjusted for each voxel. This makes it possible to obtain a model with specific parameters for each voxel.
- At least one parameter can be determined according to an average carried out on a sample of voxels in order to reduce the time for calculating the other parameters for each voxel in step 242.
- step 252 a permutation of the images is carried out on the series of images considered for which a particular parametric model has been determined for each voxel.
- the permutation of these images is advantageously random. In this way, each image is randomly assigned to an instant in the series of images. More precisely, an identical permutation is applied to each voxel of each image to preserve the spatial correlation structure of the images.
- a parametric model is then adjusted for each voxel as before.
- a threshold t having an appropriate value is applied to at least one of the parameters determined previously for the voxels of the series of permutated images. In particular, this value t is applied for example to the maximum amplitude of the models obtained. Voxels exceeding this value t are detected as "out of threshold" voxels. Certain voxels outside neighboring threshold form groups of voxels whose size is then determined. The chosen value of t does not influence the validity of the statistics but the sensitivity of the results.
- Steps 252 to 256 are repeated a given number of times, for example N times in step 257.
- This integer is determined so as to correspond to correct computation time requirements and to have a good approximation of the distribution of the probabilities densities.
- N can be equal to 10.
- the significant sizes of the voxel groups outside the threshold are determined by the method described below.
- the null hypothesis is verified when there is no coherent pathological evolutionary process.
- the distribution of voxel group sizes makes it possible to detect an active lesion for the voxel group sizes having a probability of appearing smaller than a critical value under the null hypothesis. So at step
- the significant voxel groups are determined by statistical inference.
- Figure 8 illustrates a size distribution of voxel groups for swapped image series.
- the bars represent the density obtained for each size of group of voxels calculated as the number of voxels per group. Density represents the number of groups of a given size divided by the total number of groups considered.
- ta is 0.2 and ⁇ is 0.01.
- the groups of significant voxels are larger than 10 voxels.
- the improvement of the pre-processing of the temporal bias is a sub-characteristic of the invention which has an interest as such and which can be claimed separately.
- the invention is not limited to the embodiment described above by way of example, it extends to other variants.
- Other statistical studies can be carried out with different distribution tables.
- the method can be applied without permutation of the images.
- the invention also extends to other pathologies than multiple sclerosis.
- ia f ( ⁇ ) P ⁇ . (exp (-g (xp 2 ) 2 ) + p 5
- Ic t c ⁇ . ⁇ / ( ⁇ 2 .c T. (X T .X) " ⁇ 1/2 c)
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
- Image Analysis (AREA)
Abstract
Description
Claims
Applications Claiming Priority (5)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| FR0113192 | 2001-10-12 | ||
| FR0113192A FR2830961B1 (fr) | 2001-10-12 | 2001-10-12 | Dispositif et methode de traitement d'image pour detection de lesions evolutives |
| FR0115780A FR2830962B1 (fr) | 2001-10-12 | 2001-12-06 | Dispositif et methode de traitement d'image pour detection de lesions evolutives |
| FR0115780 | 2001-12-06 | ||
| PCT/FR2002/003396 WO2003044719A1 (fr) | 2001-10-12 | 2002-10-04 | Dispositif et methode de traitement d'image pour detection de lesions evolutives |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| EP1440413A1 true EP1440413A1 (fr) | 2004-07-28 |
Family
ID=26213217
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| EP02803429A Withdrawn EP1440413A1 (fr) | 2001-10-12 | 2002-10-04 | Dispositif et methode de traitement d'image pour detection de lesions evolutives |
Country Status (4)
| Country | Link |
|---|---|
| US (1) | US7822240B2 (fr) |
| EP (1) | EP1440413A1 (fr) |
| FR (1) | FR2830962B1 (fr) |
| WO (1) | WO2003044719A1 (fr) |
Families Citing this family (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4375143B2 (ja) * | 2004-07-06 | 2009-12-02 | 株式会社日立製作所 | 動画像符号化装置 |
| FR2916883B1 (fr) * | 2007-05-29 | 2009-09-04 | Galderma Res & Dev | Procede et dispositif d'acquisition et de traitement d'images pour la detection de lesions evolutives |
| FR2916875B1 (fr) * | 2007-05-29 | 2009-09-04 | Galderma Res & Dev | Procede et dispositif d'acquisition et de traitement de donnees pour la detection de l'evolution dans le temps de lesions evolutives |
| FR2937758B1 (fr) | 2008-10-27 | 2012-09-21 | Techsia | Procede et dispositif de traitement de signaux |
| US8886467B2 (en) | 2008-10-27 | 2014-11-11 | Schlumberger Technology Corporation | Process and apparatus for processing signals |
| JP5775823B2 (ja) * | 2009-01-13 | 2015-09-09 | コーニンクレッカ フィリップス エヌ ヴェ | 画像ベースの臨床試験評価 |
| EP2387708B1 (fr) | 2009-01-16 | 2019-05-01 | New York University | Caractérisation de particules en temps réel automatisée et vélocimétrie tridimensionnelle avec microscopie vidéo holographique |
| US20130315448A1 (en) * | 2012-03-28 | 2013-11-28 | Evan Fletcher | Systems and methods for measuring longitudinal brain change incorporating boundary-based analysis with tensor-based morphometry |
| WO2014093376A1 (fr) * | 2012-12-10 | 2014-06-19 | The Cleveland Clinic Foundation | Fusion d'images à compensation automatisée pour une déformation de cerveau |
| EP3068298B1 (fr) * | 2013-11-15 | 2025-11-05 | New York University | Émission parallèle à auto-étalonnage par empreinte numérique par résonance magnétique dynamique de spin |
| US10983041B2 (en) | 2014-02-12 | 2021-04-20 | New York University | Fast feature identification for holographic tracking and characterization of colloidal particles |
| JP6264949B2 (ja) * | 2014-03-05 | 2018-01-24 | 富士ゼロックス株式会社 | 画像処理装置及びプログラム |
| WO2016077472A1 (fr) | 2014-11-12 | 2016-05-19 | New York University | Empreintes colloïdales pour matériaux mous à l'aide d'une caractérisation holographique totale |
| WO2017048960A1 (fr) | 2015-09-18 | 2017-03-23 | New York University | Détection et caractérisation holographiques de grosses particules d'impureté dans des suspensions épaisses de précision |
| WO2017139279A2 (fr) | 2016-02-08 | 2017-08-17 | New York University | Caractérisation holographique d'agrégats de protéines |
| US10670677B2 (en) | 2016-04-22 | 2020-06-02 | New York University | Multi-slice acceleration for magnetic resonance fingerprinting |
| JP6595729B2 (ja) * | 2016-06-29 | 2019-10-23 | コーニンクレッカ フィリップス エヌ ヴェ | 医療画像における変化検出 |
| US10692279B2 (en) * | 2017-07-31 | 2020-06-23 | Quantum Spatial, Inc. | Systems and methods for facilitating making partial selections of multidimensional information while maintaining a multidimensional structure |
| US12197187B2 (en) * | 2019-06-26 | 2025-01-14 | Hewlett-Packard Development Company, L.P. | Geometrical transformations in additive manufacturing |
| US12555041B2 (en) * | 2019-07-31 | 2026-02-17 | BioSymetrics, Inc. | Methods, systems, and frameworks for debiasing data in drug discovery predictions |
| US11543338B2 (en) | 2019-10-25 | 2023-01-03 | New York University | Holographic characterization of irregular particles |
| US11948302B2 (en) | 2020-03-09 | 2024-04-02 | New York University | Automated holographic video microscopy assay |
| JP7551336B2 (ja) * | 2020-05-21 | 2024-09-17 | キヤノン株式会社 | 情報処理装置、情報処理方法、およびプログラム |
| US11941825B2 (en) * | 2021-10-28 | 2024-03-26 | Canon Medical Systems Corporation | Method and apparatus for registering image volumes |
| US12369861B2 (en) * | 2022-06-03 | 2025-07-29 | BioSymetrics, Inc. | Methods, systems, and frameworks for debiasing data in drug discovery predictions |
| US12572865B2 (en) | 2022-12-27 | 2026-03-10 | International Business Machines Corporation | Automatic and dynamic adaptation of hierarchical reconciliation for time series forecasting |
Family Cites Families (30)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4755792A (en) * | 1985-06-13 | 1988-07-05 | Black & Decker Inc. | Security control system |
| US5068890A (en) * | 1986-10-22 | 1991-11-26 | Nilssen Ole K | Combined signal and electrical power distribution system |
| US5991028A (en) * | 1991-02-22 | 1999-11-23 | Applied Spectral Imaging Ltd. | Spectral bio-imaging methods for cell classification |
| US5546472A (en) * | 1992-08-07 | 1996-08-13 | Arch Development Corp. | Feature guided method and apparatus for obtaining an image of an object |
| DE69332042T2 (de) * | 1992-12-18 | 2003-01-02 | Koninklijke Philips Electronics N.V., Eindhoven | Ortungszurückstellung von relativ elastisch verformten räumlichen Bildern durch übereinstimmende Flächen |
| US5617861A (en) * | 1994-02-16 | 1997-04-08 | Huntington Medical Research Institutes | Magnetic resonance spectral analysis of the brain for diagnosis of clinical conditions |
| US6123919A (en) * | 1994-04-08 | 2000-09-26 | The Trustees Of Princeton University | Magnetic resonance imaging using hyperpolarized noble gases |
| JP3192318B2 (ja) * | 1994-05-20 | 2001-07-23 | 松下電工株式会社 | 無線式情報伝送システム |
| ATE187824T1 (de) * | 1994-10-24 | 2000-01-15 | Fisher Rosemount Systems Inc | Vorrichtung, die einen zugang zu feldgeräten in einem verteilten steuerungssystem gestattet |
| DE19525682A1 (de) * | 1995-07-14 | 1997-01-16 | Basf Ag | Verfahren zur Herstellung von Diarylethanen |
| US5848054A (en) * | 1996-02-07 | 1998-12-08 | Lutron Electronics Co. Inc. | Repeater for transmission system for controlling and determining the status of electrical devices from remote locations |
| US5848121A (en) * | 1996-10-28 | 1998-12-08 | General Electric Company | Method and apparatus for digital subtraction angiography |
| US5859891A (en) * | 1997-03-07 | 1999-01-12 | Hibbard; Lyn | Autosegmentation/autocontouring system and method for use with three-dimensional radiation therapy treatment planning |
| US6243571B1 (en) * | 1998-09-21 | 2001-06-05 | Phonex Corporation | Method and system for distribution of wireless signals for increased wireless coverage using power lines |
| US6778817B1 (en) * | 1998-12-01 | 2004-08-17 | Phonex Corporation | Method and system for combining wireless phone jack and RF wireless communications |
| US6397288B1 (en) * | 1999-09-03 | 2002-05-28 | Xid Wireless Technology Inc | Rf PC interface |
| US6653932B1 (en) * | 1999-09-29 | 2003-11-25 | Skyworks Solutions, Inc. | System and method for achieving wireless communications coverage in a local area |
| CA2634667A1 (fr) * | 1999-11-01 | 2001-05-10 | Arthrovision, Inc. | Evaluation de la progression d'une maladie au moyen de l'imagerie par resonance magnetique |
| US6492897B1 (en) * | 2000-08-04 | 2002-12-10 | Richard A. Mowery, Jr. | System for coupling wireless signals to and from a power transmission line communication system |
| US6822946B1 (en) * | 2000-08-24 | 2004-11-23 | Motorola, Inc | Wireless bridge for a broadband network |
| FR2831002B1 (fr) * | 2001-10-16 | 2004-01-02 | Victor Kossikouma Agbegnenou | Systeme de communication polyvalent sans fil |
| US7346332B2 (en) * | 2002-01-25 | 2008-03-18 | Ksc Industries Incorporated | Wired, wireless, infrared, and powerline audio entertainment systems |
| US7113763B2 (en) * | 2002-06-03 | 2006-09-26 | Nokia Corporation | Bluetooth access point and remote bluetooth modules for powerline based networking |
| US20040198236A1 (en) * | 2002-08-30 | 2004-10-07 | Scott Paine | Wireless telephone jack |
| US20050076149A1 (en) * | 2002-12-04 | 2005-04-07 | Macphy Technologies, Inc. | Method and apparatus for providing broadband wireless access services using the low voltage power line |
| US6972688B2 (en) * | 2003-01-28 | 2005-12-06 | Gateway Inc. | Power supply with modular integrated networking |
| US7065350B2 (en) * | 2003-03-28 | 2006-06-20 | Sony Corporation | Apparatus and method for communicating a wireless data signal in a building |
| TW200501624A (en) * | 2003-05-14 | 2005-01-01 | Digitaldeck Inc | Infrared signal distribution and management system and method |
| JP2005073141A (ja) * | 2003-08-27 | 2005-03-17 | Toshiba Corp | データ伝送装置および電源制御方法 |
| US7092693B2 (en) * | 2003-08-29 | 2006-08-15 | Sony Corporation | Ultra-wide band wireless / power-line communication system for delivering audio/video content |
-
2001
- 2001-12-06 FR FR0115780A patent/FR2830962B1/fr not_active Expired - Fee Related
-
2002
- 2002-10-04 WO PCT/FR2002/003396 patent/WO2003044719A1/fr not_active Ceased
- 2002-10-04 EP EP02803429A patent/EP1440413A1/fr not_active Withdrawn
- 2002-10-04 US US10/492,015 patent/US7822240B2/en not_active Expired - Fee Related
Non-Patent Citations (1)
| Title |
|---|
| See references of WO03044719A1 * |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2003044719A1 (fr) | 2003-05-30 |
| US20050141757A1 (en) | 2005-06-30 |
| US7822240B2 (en) | 2010-10-26 |
| FR2830962B1 (fr) | 2004-01-30 |
| FR2830962A1 (fr) | 2003-04-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2003044719A1 (fr) | Dispositif et methode de traitement d'image pour detection de lesions evolutives | |
| Roy et al. | Magnetic resonance image example-based contrast synthesis | |
| Manjón et al. | Diffusion weighted image denoising using overcomplete local PCA | |
| Ullmann et al. | Automatic labeling of vertebral levels using a robust template‐based approach | |
| Ahmad et al. | Edge sharpness assessment by parametric modeling: application to magnetic resonance imaging | |
| Lemieux et al. | Automatic segmentation of the brain and intracranial cerebrospinal fluid in T1‐weighted volume MRI scans of the head, and its application to serial cerebral and intracranial volumetry | |
| Gago et al. | An end-to-end framework for intima media measurement and atherosclerotic plaque detection in the carotid artery | |
| Carone et al. | ICA-based denoising for ASL perfusion imaging | |
| CN111537931A (zh) | 快速磁共振多参数成像方法和装置 | |
| US11887298B2 (en) | Fluorescence lifetime imaging using deep learning | |
| FR2884013A1 (fr) | Systeme et procede d'amelioration d'images volumetriques | |
| Goldfryd et al. | Deep semi-supervised bias field correction of Mr images | |
| Rachmadi et al. | Limited one-time sampling irregularity map (LOTS-IM) for automatic unsupervised assessment of white matter hyperintensities and multiple sclerosis lesions in structural brain magnetic resonance images | |
| EP4314854B1 (fr) | Procédé d'atténuation du bruit d'images résultant d'acquisitions multiples par résonance magnétique | |
| Pinon et al. | One-Class SVM on siamese neural network latent space for Unsupervised Anomaly Detection on brain MRI White Matter Hyperintensities | |
| Olson et al. | Optimization of q-space sampling for mean apparent propagator MRI metrics using a genetic algorithm | |
| Prastawa et al. | Automatic segmentation of neonatal brain MRI | |
| Darıcı et al. | A comparative study on denoising from facial images using convolutional autoencoder | |
| FR2943156A1 (fr) | Procede d'elaboration d'un dispositif de prediction, utilisation, support de stockage d'information et appareil correspondants. | |
| US20220349972A1 (en) | Systems and methods for integrated magnetic resonance imaging and magnetic resonance fingerprinting radiomics analysis | |
| Poojary et al. | A novel approach for bone age assessment using deep learning | |
| Wang et al. | Rapid and automatic detection of brain tumors in MR images | |
| US8224048B2 (en) | Computer implemented method for correction of magnetic resonance images | |
| Avants et al. | Sparse unbiased analysis of anatomical variance in longitudinal imaging | |
| Hou et al. | A fast and automatic method to correct intensity inhomogeneity in MR brain images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
| 17P | Request for examination filed |
Effective date: 20040410 |
|
| AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR IE IT LI LU MC NL PT SE SK TR |
|
| AX | Request for extension of the european patent |
Extension state: AL LT LV MK RO SI |
|
| RIN1 | Information on inventor provided before grant (corrected) |
Inventor name: STOECKEL, JONATHAN Inventor name: AYACHE, NICHOLAS Inventor name: MALANDAIN, GREGOIRE Inventor name: REY, DAVID |
|
| 17Q | First examination report despatched |
Effective date: 20100726 |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: EXAMINATION IS IN PROGRESS |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE APPLICATION IS DEEMED TO BE WITHDRAWN |
|
| 18D | Application deemed to be withdrawn |
Effective date: 20180501 |