Image analysis – Image transformation or preprocessing
Reexamination Certificate
1997-11-04
2001-09-11
Bella, Matthew C. (Department: 2621)
Image analysis
Image transformation or preprocessing
C382S132000, C382S154000, C382S190000, C324S309000, C600S425000
Reexamination Certificate
active
06289135
ABSTRACT:
FIELD OF THE INVENTION
The invention concerns the processing of sequential three-dimensional digital images of the same region, with a view to detecting movements in a shape of interest common to the images.
The invention applies more particularly, but not exclusively, to so-called medical images and especially to sequential digital images of the regions of the heart as, for example, the left ventricle.
In certain fields, it may be useful to analyse the spatial-temporal evolution of a region. This is, in particular, the case in the medical field, and yet more particularly in cardiology, because the tracking of the cardiac muscle (the heart) with respect to time may allow a physician to detect anomalies, and thus possibly prevent accidents.
PRIOR ART
In the field of medical imaging, a three-dimensional (3D) image is constituted by a set of image data obtained by tomoscintigraphy, or by nuclear magnetic resonance (MRI), or more generally any type of apparatus capable of acquiring images based on variations in intensity. Each elementary part (or voxel) of a region represented by a 3D image is defined by an image data item, comprising (Cartesian) position information and at least one intensity (measured value).
The expert has developed various techniques for trying to track the motions of the left ventricle of the heart. This is, in particular, the case with the technique termed “models of a deformable contour type” (see for example, A. Amini and J. Duncan, “Bending and Stretching Models for LV Wall Motion Analysyis from Curves and Surfaces” in “Image and Vision Computing”, Vol. 10 pp. 418-430, August 1992), or the technique termed “Spring-mass Meshes” (see, for example, C. Nastar, “Vibration Modes for non-rigid Motion Analysis in 3D Images”, in European Conference in Computer Vision, May 1994), or yet again the techniques termed “free-deformed superquadrics” and “volumetric superquadrics” (see, for example, respectively E. Bardinet, L. D. Cohen and N. Ayache; “Tracking and Motion Analysis of the Left Ventricle with deformable Superquadrics” in Medical Image Analysis, 1(2), 1996; and J. Park, D. Metaxas and L. Axel, “Analysis of Left Ventricular Motion based on volumetric deformable Models and MRI-SPAMM”, in Medical Image Analysis, 1(1): 53-71, March 1996.
In these techniques, the tracking relies on conservation constraints based on constraints of proximity, differential surface properties or data regarding displacement or speed contained in certain MRI type images, such as “tagged images” or “phase contrast images”.
The drawback presented by these techniques is related to the fact that they are based on correlations between successive pairs of images and that they therefore do not offer a regular, continuous and periodic transformation with respect to time. Moreover, the majority of these techniques do not provide any intuitive parameters (or chosen data) representing the motion of the analyzed shape of interest. It is, therefore, necessary to use far from trivial, and therefore lengthy and expensive, calculations to allow their analyses to be utilized.
OBJECT OF THE INVENTION
The object of the present invention is therefore to improve the situation in the field of sequential digital image processing.
SUMMARY OF THE INVENTION
For this purpose, it proposes an electronic device for the processing of sets of image data representing respectively three-dimensional digital images of the same region, taken at successive instants, and comprising:
means operative to define a three-dimensional shape of interest, as for example an active organ with cyclic motion, on the basis of at least one of said sets of image data, as well as to extract from each of the sets of image data, image data of the characteristic points of the defined shape of interest;
processing means to establish, on the basis of the image data representing the characteristic points of the shape of interest of each image, a four-dimensional (4D) transformation in space and time (or a spatial-temporal transformation), and preferably planispheric, making it possible to pass continuously from the characteristic points extracted from a starting image of the sequence, for example the first image, to the characteristic points of another image of the sequence; and
display means operative to display selected data derived from parameters of the transformation, and preferably representing the spatial-temporal evolution of the image data of the characteristic points of the shape of interest.
The starting image is understood to mean any one of the images of the sequence or a stored reference image representative of the shape of interest at a known instant, which defines a starting space and which serves as a temporal reference.
In other words, for a 3D image sequence, a 4D transformation is determined, making it possible to pass from the characteristic points of one image (representing its contour for example) to the characteristic points of another image, and possibly to revert to the starting characteristic points (temporal continuity and optionally periodicity) and on the basis whereof it is easy to draw intuitive motion data.
In a particularly advantageous way, the four-dimensional (4D) transformation is a planispheric transformation. The processing means are then capable of defining in the three-dimensional Cartesian space, a so-called “planispheric” three-dimensional reference system formed by three vectors that are orthogonal to one another and respectively parallel to the characteristic directions of the shape of interest in the Cartesian space. Such a reference system is equivalent to a combination of spherical and cylindrical reference systems. It is particularly advantageous for processing shapes of interest, such as that of the left ventricle of the heart (or myocardium), since it makes it possible to represent it substantially in the form of a disk, thus considerably simplifying the analysis of the tracking of its dynamic evolution.
By way of example, in the case of a left ventricle, the three easily recognisable characteristic directions are called apico-basal, septo-lateral and infero-anterior.
To allow the planispheric reference system to be used, the processing means is capable, on the one hand, of applying to the image data in the initial Cartesian space, a first transfer function making it possible to transform their Cartesian spatial coordinates into planispheric spatial coordinates and, on the other hand, of applying to these image data in the planispheric space, a second transfer function making it possible to transform their spatial planispheric coordinates into Cartesian spatial coordinates.
In the case where the planispheric reference system is used, the 4D transformation may thus be represented by the composition of the first coordinate transfer function, of a transfer function of the characteristic points making it possible to pass from the three-dimensional planispheric coordinates of the characteristic points of one image to the three-dimensional planispheric coordinates of the characteristic points of another image of the sequence, and of the second transfer coordinate function.
According to yet another characteristic of the invention, the selected data derived from the transformation, represent three canonical motions orthogonal to one another namely: a radial motion relative to the origin of the planispheric reference system, a rotation relative to one of the vectors of this planispheric reference system (the apico-basal axis in the case of a left ventricle), and an elevation relative to this same vector of the planispheric reference system.
Preferably, the processing means are arranged to apply, to the image data representing image data of the characteristic points in the planispheric reference system, an operator of the “least squares” type, so as to determine the transfer function of the characteristic points.
In a first application to conventional digital 3D images (each image data item comprising a 3D position information plus one intensity), the processing means are arranged to determine the transf
Ayache Nicholas
Declerck Jèrôme
Feldmar Jacques
Bella Matthew C.
Desire Gregory
Foley & Lardner
Inria Institut National de Recherche en Informatique et en Antom
LandOfFree
Electronic image processing device for the detection of motions does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Electronic image processing device for the detection of motions, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Electronic image processing device for the detection of motions will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2523114