Automatic three-dimensional segmentation of short axis cardiac magnetic resonance images with delayed staining

FIELD: physics.

SUBSTANCE: system (200) has breaking unit (210) for breaking an image into a plurality of image areas, each image area displaying an area of the anatomical structure, and an adaptation unit (220) for adaptation of the referenced image to the evaluation function-based image, wherein the evaluation function is a function of parameters of the reference image and index numbers of the image and relative positions thereof on the image, and based on a criterion which must be satisfied by the index numbers of the evaluation function, wherein the evaluation function is determined based a plurality of image areas. Determination of the evaluation function based on a plurality of image areas enables to calculate optimum values of input into the value of the evaluation function within each image area.

EFFECT: high reliability of automatic detection of a cardiac muscle in short axis layer images.

11 cl, 9 dwg

 

The technical FIELD TO WHICH the INVENTION RELATES.

The present invention relates to the segmentation of images and, in particular, to the definition of the contours of the anatomical structures in the image calculated by the layer image data, and in addition to defining the contours of the anatomical structures in the image data.

The LEVEL of TECHNOLOGY

Assessment of the viability of the heart tissue is of great importance when planning surgical intervention and therapy after a heart attack. In particular, the proportion of viable myocardium is a major factor in determining whether to bring the patient the benefit of the operation of revascularization. In addition to estimating the thickness and thickening of the wall of the left ventricle with high resolution to visualize normal, ischemic and non-viable areas using methods of contrast enhancement and, in particular, magnetic resonance imaging (MRI delayed contrast enhancement (LEMR). For localization and quantification of non-viable tissue, the first stage is that define the boundaries of the endo - and epicardial contours (in other words, segment the myocardium) in each layer (usually 10-12 layers) LEMR-data (data obtained magnetic resonance study with delayed contrast enhancement) volume in cross section along the short axis, and is monotonous and length of the part, when done manually. However, automatic segmentation of the myocardium is a complex task and is rarely implemented in current commercial products. Designing an automatic way of determining the boundaries of endo - and epicardial contours is difficult mainly because of the heterogeneity of myocardial tissue, resulting from accumulation of contrast agent in ischemic and non-viable areas.

Figure 1 shows the image calculated for layer LEMR-SA data-image (image on the short axis). The first picture 11 represents an image without contours of the myocardium. The second shot 12 represents the image with the contours of the myocardium, drawn manually by experts. The boundary between a pool of 110 blood and abnormal tissues within 120 attack is localized with difficulty. The scar region 120 appears to be white and healthy part 130 is dark, and the surrounding organs vary from grey to dark tones. In addition, the boundaries of the white areas often appear very fuzzy, particularly if they are located close to the pool of blood, which especially complicates precise localization of the endocardium. Therefore, the problem lies in the selection of the structure type of the myocardium, which may contain both dark and bright areas of a structured environment.

The algorithm for automatic segmentation for segmental and SA images, obtained by the method of LEMR, described in the article by E. Dikici, T. O'donnell, R. Setser and R.D. White, "Quantification of delayed enhancement MR images",Proc. of the 7th International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI'04), LNCS Series, Vol. 3216, Springer, pp. 250-257, 2004.

The algorithm for the automatic recognition of the boundaries of the myocardium using stochastic schemes with active contours are also described in the work of C. Pluempitiwiriyawej et al: “Active contour with automatic initialization for myocardial perfusion analysis” 2005 27thAnnual International Conference of the IEEE Engineering in Medicine and Biology Society (IEEE Cat. N0. 05ch37611 C) IEEE Piscataway, NJ, USA, 2006, page 4.

In the publication WO 2006/033066 A1 describes a system for determining the boundaries of anatomical structures in the image calculated in the layer LEMR data SA images and symbols necrotic areas in the myocardium of the heart on the basis of combining functional imaging with SA-images obtained by the method of LEMR.

DISCLOSURE of INVENTIONS

Create an alternative solution to the problem of determining the boundaries of the contours of the myocardium in the layer LEMR-image data volume on the short axis in the presence of scar tissue may be useful, including for determining the boundaries of other structures in the images obtained using different methods of image capture.

Thus, in accordance with an aspect of the invention features a system for determining the boundaries of anatomical structures in the image, the computed p is the layer image data, the system contains

block partitioning to split the image into multiple areas, each image area represents the region of the anatomical structure; and

- block adaptation for adapting the reference image to the image based on the evaluation function, and the evaluation function is a function of the parameters of the reference image and the amounts of images and their relative positions in the image, and on the basis of criteria, which must satisfy the calculated value of the evaluation function,

moreover, the evaluation function is defined on the basis of many areas of the image.

In other words, the evaluation function depends on the partitioning of the image into several sections. For example, the evaluation function can contain a member that is dependent on the average gray level or variance of the gray levels of the pixels contained in the image area. Definition of evaluation function based on multiple areas of the image allows us to calculate the optimal value of the contribution to the value of the evaluation function within each area of the image. For example, in one embodiment, the magnitude of the contribution can be determined by the type of tissue of the myocardium, recognized in the area of the site. The evaluation function can contain members, depending on the type of the recognized tissue. If detected the fabric of the first type, to contribute to the value of the evaluation function is composed of members of the corresponding tissue of the first type. If the identified fabric of the second type, to contribute to the value of the evaluation function is composed of members of the corresponding tissue of the second type. This approach has been particularly useful to highlight the contours of the myocardium in the layer LEMR-image data volume on the short axis in the presence of scar tissue. The invention allows optimization of the processing sections of the image containing scar tissue in the myocardium. The invention can also be applied to the segmentation of other normal or pathological anatomical structures obtained by various methods of obtaining the image.

In one embodiment, system block adaptation is additionally made with the possibility of detection of abnormal tissue in an anatomical structure of an image area from a variety of areas of the image, and the evaluation function contains a member that contributes to the evaluation function, when the above-mentioned image area is detected mentioned abnormal tissue. Abnormal tissue anatomical structures can be recognized by the average intensity or variance of the pixel values of the image area, combined, for example, with the reference image. The evaluation function can contain a member that corresponds to ANO the social fabric, located in the image area. The said member may be composed with the opportunity to contribute to the value of the evaluation function when the section of the partitioning of the image is detected abnormal tissue.

In one embodiment, the system adaptation of the reference image based on the search parameter values of the reference image, when the evaluation function satisfies the condition criterion using the absorbing of the search algorithm. This choice provides the best compromise between computational complexity and sensitivity to initial conditions. For example, absorbing search algorithms are more robust than algorithms gradient descent and less complex than the algorithms of dynamic programming. The use of absorbing the search algorithm provides a useful ability to use non-differentiable merit function.

In one embodiment, the system of the reference image is a closed tape, limited external and internal closed contour. The reference image in the form of a closed ribbon is useful for modeling of epicardial and endocardial contours of the myocardium.

In one embodiment, system partitioning is a pie break. Sectoral split with the center in the center of the left as the fishing rod is suitable for determining the boundaries of the myocardium and other anatomical structures, which can be defined by closed contours.

In one embodiment, the system further comprises a unit combining to combine model surfaces with lots of reference images, each reference image adapts to the image calculated by the layer image data, and at least one reference image adapted by block 220 of the adaptation system 200. The advanced system is designed to define the boundaries of anatomical structures in the image data using such useful features that at least one reference image is optimally adapt the unit to adapt the system.

In one embodiment, system block combination contains

block affine transformation affine combine model surfaces with multiple reference images, thereby creating affine-combined surface model;

- block of local deformation for local paffing combining affine-combined surface model with multiple reference images, thereby creating locally-combined surface model and

the unit specifications for the adaptation of locally combined surface model to the image data.

The application of the affine combination with subsequent reaffinned the combination increases the em reliability of the method, used by block matching. Optionally, an affine transformation can be reduced to a transformation. The specification of model surfaces it is important to perform final adjustments to the model surface in the vicinity of abnormal tissue.

In accordance with an additional aspect of the invention proposes a method of determining the boundaries of anatomical structures in the image calculated by the layer image data, the method contains the following steps:

- phase partitioning to split the image into multiple areas, each image area represents the region of the anatomical structure; and

stage adaptation for adapting the reference image to the image based on the evaluation function, and the evaluation function is a function of the parameters of the reference image and the amounts of images and their relative positions in the image, and on the basis of criteria, which must satisfy the calculated value of the evaluation function,

moreover, the evaluation function is defined on the basis of many areas of the image.

In accordance with an additional aspect of the invention features a computer program product to be loaded in a computer system, computer software product contains commands to define the boundaries of anatomica is some structure in the image, calculated on the layer image data, and a computer system includes a processor unit and memory, and computer software product after loading the supplies mentioned CPU unit's functional ability to perform the following tasks:

- split the image into multiple areas, each image area represents the region of the anatomical structure; and

- adaptation of the reference image to the image based on the evaluation function, and the evaluation function is a function of the parameters of the reference image and the amounts of images and their relative positions in the image, and on the basis of criteria, which must satisfy the calculated value of the evaluation function,

moreover, the evaluation function is defined on the basis of many areas of the image.

In accordance with an additional aspect of the invention, the system in accordance with the invention is contained in the device receiving the image.

In accordance with an additional aspect of the invention, the system in accordance with the invention is contained in the workstation.

Specialists in the art should be obvious that at least two of the above embodiments, performances and/or aspects of the invention can be combined in any way that p is establece useful.

On the basis of the present description, a person skilled in the technical field can be created modifications and variations of the device receiving the image, the workstation, of the method and/or computer program product, which correspond to the described modifications and variations of the system. Specialist in the art will understand that the method is applicable to data from multidimensional images, for example, 2-dimensional (2-D)3-dimensional (3-D) or 4-dimensional (4-D) images obtained by different data collection methods, for example, but without limitation, standard radiography, computed tomography (CT), magnetic resonance imaging (MRI), ultrasound (US), positron emission tomography (PET), single photon emission computed tomography (SPECT) and medical radiology (NM).

BRIEF DESCRIPTION of DRAWINGS

These and other aspects of the invention are clearly set forth and explained in the example below of embodiments and embodiments and with reference to the accompanying drawings, in which

Figure 1 - image is calculated by layer LEMR data SA images, without contours of the myocardium and with the contours of the myocardium, drawn manually by experts;

Figure 2 - sketch the block diagram of an exemplary variant of the implementation of the system;

Figure 3 - reference image in the form of a closed ribbon;

Figure 4 - an example is izbieniya image into four quadrants;

Figure 5 - results of automatic segmentation contours of the myocardium system in accordance with one embodiment;

6 - the results of the automatic segmentation contours of the myocardium in accordance with another embodiment;

7 is a block diagram of the sequence of operations in an exemplary performance of the method;

Fig diagram of an exemplary variant of the implementation of the device receiving the image and

Fig.9 is a diagram of an exemplary case for the workstation.

Identical digital positions are used to denote similar parts in all the figures.

The IMPLEMENTATION of the INVENTION

Figure 2 schematically shows a block diagram of an exemplary variant of the implementation of the system 200 for determining the boundaries of anatomical structures in the image calculated by the layer image data, the system 200 contains

block 210 split to split the image into multiple areas, each image area represents the region of the anatomical structure; and

block 220 adaptation for adapting the reference image to the image based on the evaluation function, and the evaluation function is a function of the parameters of the reference image and the amounts of images and their relative positions in the image, and on the basis of the criterion that udovletvorat the calculated value of the evaluation function,

moreover, the evaluation function is defined on the basis of many areas of the image.

A sample implementation of the system 200 further comprises the following blocks:

block 230 combining to combine model surfaces with lots of reference images, each reference image adapts to the image calculated by the layer image data, and at least one reference image adapted by block 220 of the adaptation system 200;

block 260 to control the sequence of actions performed in the system 200;

- the user interface 265 for interaction with a user of the system 200 and

block 270 memory for data storage.

In one embodiment, system 200 block 230 combination contains

block 232 affine transformation affine combine model surfaces with multiple reference images, creating, thus, affine combined surface model;

block 234 local deformation for local paffing combining affine combined surface model with multiple reference images, creating, thus, locally combined surface model and

block 236 refinement to adapt locally combined surface model to the image data.

In one embodiment, the system is s 200 has three input connector 281, 282 and 283 for the input data. The first input connector 281 is arranged to receive data from data storage media, for example, but without limitation, hard disk, magnetic tape, flash memory or an optical disc. The second input connector 282 is configured to receive data from a user input device, for example, but without limitation, a mouse or touch screen. The third input connector 283 is configured to receive data from a user input device such as keyboard. Input connectors 281, 282 and 283 are connected to the input unit 280 controls.

In one embodiment, system 200 includes two output connector 291 and 292 for outgoing data. The first output connector 291 is configured to output data in the storage medium data, for example, but without limitation, hard disk, magnetic tape, flash memory, or optical disk. The second output connector 292 is configured to output data to a display device. Output connectors 291 and 292 receive data via the output unit 290 controls.

Specialist in the art will understand that there are many ways to connect input devices to the input connectors 281, 282 and 283 and output devices output soy is initely 291 and 292 of the system 200. The above methods include, but without limitation, wired and wireless communications, digital network, for example, but without limitation, a local area network (LAN) and wide area network (WAN), the Internet, digital telephone and an analog telephone network.

In one embodiment, system 200, the system 200 includes a block memory 270. The system 200 is configured to receive input data from external devices via any of the input connectors 281, 282 and 283 and store the received input data in block 270 memory. Loading the input data in block 270 memory allows quick access units of the system 200 to the appropriate sections of the data. The input data can include, for example, image data and the reference image. Optionally, the input data may optionally contain a surface model. Unit 270 of the memory can be performed using devices such as, but without limitation, random access memory chips (RAM)chip constant memory (ROM) and/or the hard disk drive and hard disk. Unit 270 of the memory may be further configured to store the output. The output can contain, for example, the reference image, adapted to the layer image data, in accordance with the present invention. Optionally, the output can optionally contain a combined fashion is ü surface. Block memory 270 may also be configured to receive and/or deliver data from/to block/blocks of the system 200 containing block 210 partitioning block 220 adaptation block 230 the alignment unit 232 affine transformation, the block 234 of the local deformation, block 236 clarification unit 260 controls and user interface 265, bus 275 memory. Block 270 memory is additionally configured to provide output data to external devices via any of the output connectors 291 and 292. Storing data blocks of the system 200 in block 270 memory can effectively improve the performance of units of the system 200, and the transmission rate of the output data of the blocks of the system 200 to the external device.

In an alternative embodiment, the system 200 may not contain any unit 270 of the memory or bus 275 memory. The input data used by the system 200, can be supplied at least one external device, such as an external memory or a processor, connected(th) with blocks of system 200. Similarly, the output generated by the system 200, may be submitted in at least one external device, such as external memory or a processor, connected(th) with blocks of system 200. The blocks of the system 200 may be configured to receive data each each internal connection or via the data bus.

In one of the variations is the implementation of the system 200, the system 200 includes block 260 to control the sequence of actions performed in the system 200. The control unit may be configured to receive and supply management data from/to block/blocks of the system 200. For example, the block 220 adaptation can be performed so that after the adaptation of the reference image to the image calculated by the first layer image data, to apply the control data of the reference image is adapted to the image in block 260 control, and unit 260 controls can be executed with a possibility of control data "to adapt the reference image to the image calculated by the second layer image data in block 220 adaptation. The calculation of the image by the first and second layers may be performed by block 220 adaptation. In an alternative embodiment, the control function may be implemented in another block of the system 200, such as in block 220 of adaptation.

In one embodiment, system 200, the system 200 includes a user interface 265 for interaction with a user of the system 200. The user interface 265 may be configured to receive user input data to select the reference image for adapting to the image from a set of reference images available in the system 200, or data user input for splitting the image. The user interface may further provide a means by which to represent the adapted reference image for observation by the user. Optionally, the user interface may receive user input for selecting the mode of operation of the system, for example to select multiple members of the evaluation function. Specialist in the art will understand that the user interface 265 system 200 can usefully implement more functions.

Embodiments of the present invention is described below with reference to the boundary definition epicardial and endocardial contours and surfaces in LEMR data SA image. Specialist in the art will understand that the system is also suitable for the segmentation of other structures, including, but without limitation, liver, pancreas and blood vessels, the images obtained by scanning methods CT (computed tomography) or MR (magnetic resonance).

In one embodiment of the present invention the reference image suitable for modeling of the myocardium, is closed by the tape, as shown in figure 3. The reference image is implemented interpolation splines using as fewer nodes to reduce the number of model parameters. In one embodiment, the parameters of the splines set using 8 nodes. The use of a larger number of nodes to assign the parameters of the reference image is ment improves the accuracy of the calculations the cost of increasing computation time. The use of splines provides optimally smooth solutions. In addition, the specific choice of the basic interpolation functions and nodal points, as opposed to approximating functions and control points, increases the stability of the algorithm and the viability of the optimization strategy. The use of splines to define the boundaries of the contours in the images described in the article by P. Brigger, J. Hoeg, and M. Unser, "B-Spline Snakes: A Flexible Tool for Parametric Contour Detection",IEEE Transactions on Image ProcessingVol. 9, No. 9, pp. 1484-1496, 2000. In particular, the infarction model in the form of a belt structure 30 formed by the virtual Central line 31 containing many nodes (schematically depicted by dots) and described by the parameters width (schematically depicted by arrows). As the Central line, and width are continuous interpolating spline functions with a discrete set of parameters (xi,yi,wi), wherexi,yidenote the coordinates of the i-th node in the coordinate system of the image, andwirefers to the width of tape around the i-th node of the Central line. The advantages of the above representations are a natural link between the epicardium 301 and endocardium 302 and the flexibility of discrete choice data. Timing paths you can take any desired accuracy using fast spline-filter is written in the article M. Unser, A. Aldroubi and M. Eden, "B-spline Signal Processing; Part 1 - Theory",IEEE Transactions on Signal ProcessingVol. 41, No. 2, pp. 821-832, 1993.

In one embodiment of the invention, the block 220 adaptation designed to determine the myocardium, the evaluation axis of the left ventricle and the initialization reference image in the image. Optionally, the system 200 may include a separate block for each task, for example, the recognition unit, the evaluation unit axis and the initialization block of the reference image. The system 200 is designed for reliable recognition of the myocardium in the image layers on the short axis to set the starting position of the geometric reference image. The above task is performed in three stages: first, the myocardium, which appears as a dark ring, is recognized in each layer; secondly, according to the above-mentioned recognition evaluate the left ventricle and, finally, appreciate the robust initialization of the model on the basis of the recognition of the myocardium, and evaluation. For recognition of the myocardium on the image layers on the short axis in one embodiment, use image conversion Hava. Conversion Hava described, for example, in the original P. V. C. Hough, "Method and means for recognizing complex patterns, U.S. patent 3069654, 1962. Conversion Hava is intended to recognize the ring of geometric shapes. Because the myocardium p is establece as a dark ring, the myocardium can be recognized as the maximum response and the convolution of the image with radially symmetric kernel, simulating a dark ring. Radial profile of the mentioned kernel is given by the Laplacian of Gaussian, offset by the radius of the ring, and its width is directly related to the expected thickness of the myocardium. Convolution is calculated in the radial coordinate in the Fourier transform of the field that corresponds to the multiplication of the transform image with the analytical Hankel transform of the kernel for each angular frequency. The Hankel transform is described, for example, Bracewell, R., The Hankel Transform",The Fourier Transform and Its Applications, 3rd ed.New York: McGraw-Hill, pp. 244-250, 1999. The above operation is one-dimensional and, therefore, very fast. When performing the inverse transform and the same procedure is repeated for different radii of the rings the maximum response convolution finally determine the optimal center and radius for the initialization of the deformed reference image. Evaluation axis of the left ventricle reliably determined on the basis of the center of optimal dark rings, recognized in the image of each layer. This operation is performed using the method of least median of squares containing the step of discarding outlier, not to account for possible false data detection infarction. The method is described, n is the sample, in the work of P. J. Rousseeuw, A. M. Leroy, Robust regression and outlier detection, John Wiley & Sons, Inc., New York, NY, 1987.

And finally, the original geometric position of the reference image for the image of each layer is determined as follows. If the identified optimal heart attack, received the recognition of the rings, is near the point of intersection of the axis of the left ventricle and the current layer, and if the radius of the myocardium corresponds to the pre-specified radius of the reference image, the reference image is initialized with the detected optimum center and radius. In another case, the original center of the reference image is the point of intersection of the axis with the current layer and the radius is calculated from the detected optimal ring, using the parabolic model radii.

In one embodiment of the present invention the evaluation function is a power function of the reference image in the form of a closed ribbon. The energy function contains the members of the internal energy and the members of the external energy. Members of the internal energy reaches a minimum when the geometry of the reference image is not deformed and is identical to the typical geometry of the reference image, based on a priori knowledge. Members of the external energy reaches a minimum when the reference image is deformed so that some the e part of the reference image block some characteristic features, detected in the image. The condition of the criterion is that the energy function has reached a minimum.

As an integral part of members of the internal and external energy in addition to the above members or in combination with them additional possible members of the energy functions include, but are not limited to, members belonging to the following parameters:

-Roundness: Central line 31 should not be too much to deviate from the circle. A possible choice is the use of measurable criteria of dispersion of curvature relative to the Central line of its medium, as a circle has constant curvature.

-Regularity: Changes in width along the center line 31 should be limited. This condition provides adequate coupling between circuits. A possible choice is also measurable criterion dispersion width, variable relative to its mean value.

-The homogeneity of the pool of blood: Excluding papillary muscles, a pool of blood should be homogeneous. A possible alternative is the variability of the internal region after excluding the lowest part of the histogram to account for the possible presence of papillary muscles.

-The homogeneity of myocardium: Area of infarction should be evenly distributed and dark, if the fabric is normally the th, otherwise, it should be light.

-The contrast of the endocardium: A pool of blood should locally to demonstrate a higher intensity than the myocardium, which can be measured by the sum of 1-dimensional filters contrast (defined, for example, by the first derivative of the Gauss filter) along the normals to the contour of the endocardium.

-Contrast epicardium: Epicardium should be presented with positive, negative, contrast, or contrast in the form of a comb, depending on the surrounding organs. Mentioned the contrast can be measured by the sum of 1-dimensional filters of contrast in the form of a comb (defined, for example, the second derivative of the Gauss filter) along the normals to the contour of the epicardium.

-The contrast of myocardium: The average intensity of the myocardium must be less than the average intensity of a pool of blood.

In one embodiment, the evaluation function is a weighted sum of the members of the evaluation function. The weights are user-defined algorithm parameters.

Alternatively or additionally, the evaluation function can be a force field acting on the reference image, and the condition of the criterion may be that the force field was, essentially, zero.

The above-mentioned evaluation function and condition criteria are based on the entire image and fit with the image is of different types, for example, magnetic resonance images of the heart obtained in the mode of filming. However, in the case LEMR data SA-images given conditions sufficient to obtain a reliable automatic detection of infarction and necessary characteristic features defined on the basis of areas of the image. Mentioned characteristic features are designed to search for a reliable solution of the problem of segmentation of images, when the myocardium is present scars (bright areas).

In one embodiment of the present invention, the image is divided into four quadrants Q1, Q2, Q3, and Q4. Figure 4 shows the mentioned approximate partitioning of the image. The location of the four quadrants determined by the provisions of the center 41 of the left ventricle and the center 42 of the right ventricle. In one embodiment of the present invention the determination of the center of the right ventricle based on the best correlation between the circular reference image with the same intensity as the intensity of the pool of blood in the left ventricle and images in the annular region outside the myocardium. Quadrants break infarction and initialized the reference image into four parts. Member of the evaluation function can be defined on the basis of the split. Members defining contrast, adapt to the quadrants on the basis of anatomicheskoi information. For example, it is assumed that Q1 infarction should be darker than the surrounding body (right ventricle), whereas in Q3 surrounding the body is light and also dark, but often visible thin layer of fat. In Q2 and Q4 suitable filters contrast in the form of a comb, as outside of the myocardium alternating dark and light fabrics. Thus, instead of computing the defining criteria of members throughout the myocardium each member is calculated separately in each quadrant. This approach allows the algorithm to detect abnormal tissue if necessary and, consequently, to control the adaptation process. It is advisable to note that although the splitting of the myocardium specifies different areas of the myocardium, the said sections are connected to adjacent areas, set partitioning, and interact with them.

Alternatively, in block 210 split system 200 can be used another way to split the image, randomized or based on image analysis, for example, using pattern recognition or segmentation of the object. It is advisable to specify that the proposed partitioning of the image does not have to lead to the breaking of anatomical structures, such as the heart, acknowledged substructures, such as the ventricles and the Atria. The purpose of partitioning is to identify additional members of the evaluation function, PR is that each member has its own area on the site structure, to provide more detail and, therefore, higher precision.

In one embodiment of the present invention in each quadrant are performing multiple tests for the detection of possible scarring or ischemic zones. For example, the abnormal tissue recognize, if

- averaged intensity of the myocardium within the quadrant is high compared with the average intensity of a pool of blood;

- the cumulative difference of the gray levels of the myocardium relative to the expected value of abnormal tissue (e.g., 255) is lower than the accumulated difference between the intensities relative to the expected value of healthy myocardium (for example, 0);

- mentioned test is confirmed if the local gradient inside quadrant of the myocardium is low, which means that the region is homogeneous.

In one embodiment of the present invention if in quadrant recognized the scar, the members of the evaluation function in quadrant adapted as follows:

as the scar seems bright and normal myocardium is dark, the scar appears lighter than the surrounding organs, therefore, the expected contrast along the boundary changes (for example, from 0 to 255);

for the same reason, the uniformity within the myocardium is defined as the accumulated difference relative to (or in the form of disperse and relative to) the expected value of abnormal tissue (e.g., 255), instead of the expected values healthy myocardium (e.g., 0).

The above test can also be a member of the evaluation function, for example, in the form of the coefficient in the above-mentioned member. If the test value is 1 (true), then the member will contribute to the evaluation function. If the test value is 0 (false), then the member will not contribute to the evaluation function.

In one embodiment of the present invention the evaluation functionFmay be expressed by the following energy function:

F(p,I)=Fs(C,w)+Fc(Ci,Co,I)+Fr(M,B,I),

where

-pmeans the parameter vector containing the parameters of the reference image,pi=(xi,yi,wi), andwmean vector width, containing the values of the width of thewidescribing the coordinates of the nodes of the Central line;

-C,CiandComean respectively the Central line, the inner and outer contours of the reference image in the form of a closed ribbon;

-Imeans image, i.e. a map that assigns an intensity areas of the image; and

-MandBmean respectively the area of infarction and the pool of blood, defined by the placement of the reference image in the form of a closed ribbon in the picture the AI.

The first member of theFs(C,w) is a member form the reference image and is a member of the internal energy. For example,

Fs(C,w)=λ001|κ(s)-κ|2ds+λ101|w'(s)|2ds,

where κ(s) denotes the curvature of the Central line, andκmeans the average curvature. Specialist in the art will understand thatsis a parameter of the spline representation of the model of the reference image in the form of a closed ribbon. As the curvature and the width of the Central line must be small changes.

The second member of theFc(Ci,Co,Iis describing the circuit member, which is designed to pull the wall of the epicardium and endocer is in the preferred location of the image gradients. The said member can be expressed in the form

Fc(Ci,Co,I)=λ01Iin(s)2ds-λ01|Ion(s)|3ds,

whereIin=I(Ci(s)).n(s)andIon=I(Co(s)).n(s),Imeans the gradient of the image andn(s) means directed outward normal to the center line. For the implementation of the said member use gradient filters that reflect a priori knowledge about the relative intensity of normal and anomalous parts of the myocardium, as described in the paragraph on the detection of abnormal tissue.

The third member of theFr(M,B,I) is a member of describing the area. The gray levels of the pool of blood should be distributed uniformly. In addition, normal tissue of the myocardium appear dark and abnormal tissue are represented light, which leads to a strong global contrast with a pool of blood. Therefore, the term, the region has the form:

Fr(M,B,I)=λ4|M|M|I(x,y)-m|dxdy+λ5|B|B|I(x,y)-b|dxdy+λ6 Cglob,

where the areaBit has a medium intensity ofband the area of|B|and the intensity expected for the fieldMattack equal tom.Cglobmean global contrast between the two areas.

Below is a description of the detection of abnormal tissue. To this end, the image calculated for layer LEMR-SA data image is pre-processed before the deformation of the reference image: a combination of the intensity distributions is measured using the algorithm the expectation-maximization that allows you to stretch the range to saturate the darkest and lightest parts of the image. The mentioned areas are respectively to a healthy and abnormal tissues, which, as expected from the above, must be submitted in the form of a homogeneous areas the minimum intensity ofhand the maximum intensity ofa in the new range. Then for each site of the myocardiumMi=MQirecognized potential scarring or ischemic zone, if the following three conditions:

- averaged intensity of the myocardium within the quadrant is high compared with the average intensity of blood:

Mi|I(x,y)|dxdy>b;

the variance of the intensity relative to thealower than relative to theh:

Mi|I(x,y)-a|dxdy<Mmi> i|I(x,y)-h|dxdy;

the sum of the values of the local gradients within the areaMiattack is small.

If the plot ofMirecognized the scar, the defining criteria members adapted respectively to scar. Because the scar is lighter than the surrounding organs, gradient filters that specify the expected contrast along the border, are the reverse. For the same reason, the expected value ofmwithin the myocardium, which is used inFras wellainstead ofh.

In one embodiment, the evaluation function is non-differentiable function of the parameters (xi,yiwithe reference image, whereiis the index of the node on the Central line, in particular, because of the nonlinear exceptions histogram inner region to account for papillary muscles. Therefore, Ichnya the gradient descent methods are not suitable for minimization. In addition, any scheme of gradient descent has a natural limitation due to a heightened sensitivity to local minima. The optimization scheme is based on absorbing algorithm based only on the direct calculation criteria. This choice ensures a good compromise between computational complexity and sensitivity to initial conditions. Applied optimization scheme is more robust than gradient descent scheme and less complex than dynamic programming, which is the traditional solution for global optimization of spline models. Strategy optimization on the basis of absorbing the algorithm can be described by the following iterative algorithm:

Repetition

(A) Around each node one by one,

To search for a value (xi, yi, wiparameter of the i-th node in predetermined ranges,

To find the optimal move in a position that gives the lowest energy

(B) Repeatedly take samples and to move the node along the Central line (move)to enhance invariance to rotation and parameterization,

until you find a stable condition.

Figure 5 shows the results of the automatic segmentation contours of the myocardium in one embodiment, the system 200. Shown Ave is dimensional images contain different types of abnormal tissue: large bright transmural scars, small subendocardialnah scars, diffuse or fuzzy light areas. Bright areas are properly contained within the segmented myocardium. The results are satisfactory, but some minor inaccuracies remain along the border on hard segmentierung images. The mentioned inaccuracies generally correspond to the anomalous thickness of the myocardium, which is too thin, if the scar is removed from the muscles, or too large, if the segmentation result contained the surrounding structure. To reduce the above-mentioned inaccuracies and to obtain a 3-dimensional surface model of the myocardium requires additional processing LEMR data SA images. In order to solve the above-mentioned problem, in one embodiment, the system contains a block 230 combining for combining 3-dimensional surface model with lots of reference images, each reference image adapts to the image calculated by the layer image data, in block 220 the adaptation system 200. Suitable surface model can be obtained from the definition of epicardial borders and endocardial surfaces in the movie sequences magnetic resonance (MR) images.

Research method LEMR usually performed in approximately 20 minutes after receiving MR-cinoposledovanie, which demonstrates DV is a provision of the myocardium during the cardiac cycle. During the mentioned 20 minutes the patient does not move on the table. Therefore, all anatomical information that can be extracted from cinoposledovanie, is a useful a priori knowledge for segmentation of the volume obtained by the method of LEMR. In particular, the segmentation of the myocardium get throughout cinoposledovanie images, where each image corresponds to the phase of the cardiac cycle, through:

- segmentation of the myocardium in the image, the corresponding end-diastolic phase of the cardiac cycle, using, for example, the same approach as in the case of SA-images obtained by the method of LEMR. For segmentation mentioned end-diastolic MR images does not require any partitioning of the image;

- application of the segmentation result in end-diastolic phase to cinoposledovanie in order to segment each image in the sequence, as described, for example, Hautvast, G.; Lobregt, S.; Breeuwer, M. & Gerritsen, F. Automatic contour propagation in cine cardiac magnetic resonance images,IEEE Transactions on Medical Imaging, 2006, 25, 1472-1482. At the end of each phase will receive two 3-dimensional surface models, for example a polygonal mesh describing the endocardium and the epicardium. Mentioned surface models contain important information about the shape and thickness.

To use the aforementioned information system 200 can you alnite with automatic phase selection, which corresponds to the received LEMR-SA data-images, for example, by reading the attributes of images in the DICOM standard (standard transmission and storage of medical images), and the use of appropriate 3-dimensional surface models as a priori information. However, even if the patient does not move between research in movie mode and method LEMR, the patient can relax or breathe more or less deeply. Therefore, the model surface is separated from the phase cinema mode, you cannot directly be applied to the SA image obtained by the method of LEMR, and therefore, the phase alignment is required. In an alternative embodiment, the system 200 in accordance with the present invention can be applied to a suitable surface model obtained in any other way.

In one embodiment, system 200 is made with the possibility of additional combine model surfaces with lots of reference images, each reference image adapts to the image calculated by the layer image data in block 220 of the adaptation system 200. The specified action is performed in two stages:

-Strict or affine transformationattracting surface models adapted to the reference images. The transformation that leads to optimal SOG is osowaniu between surface models and service reference images, calculate and apply to surface models.

-Local refinementmodels of surfaces, depending on the position of the reference image and the gray levels of the image. More trust epicardial contour of the reference image, which is usually more accurate than endocardial the contour of the reference image. Information about the thickness of the provided surface models, used for the introduction of restrictive due to the local deformation. Near contours searches white pixels that were recovered bright areas, which could be lost during the adaptation of the reference image. Of the above operations, it is important to apply consistently, as an affine transformation leads to a reliable result, while local refinement adjusts the location of the surface models under the assumption that they are, in General, are optimal. Consequently, the use of only local refinement does not lead to optimal results.

In one embodiment of the present invention, the deformation is performed using the so-called affine-local strategy: first, find an affine transformation, which gives the best match between the net and service circuits, without changing the geometry of the grid; then set the characteristic strengthF3DAve is kladiva at each vertex of the mesh. The power ofF3Dconsider the original shape of the mesh (Fint), the distance to the corresponding 2-dimensional contour (Fcont) and thickness (Fth) attack:

F3D=Fint+Fcont+Fth.

And, finally, the mesh locally precise by replacing the force of attraction circuit power (FI) test the intensity of the image to ensure that the content of scars, i.e. bright areas inside the final contours of the myocardium, which leads to the following force:

F3DRefine=Fint+FI+Fth.

The result of segmentation of the myocardium contains model epicardial and endocardial surfaces. In addition, the intersection of epicardial and endocardial surfaces with each data layer SA of the image obtained by LEMR, determines clarified the contours of the epicardium and endocardium in each image calculated by the layer.

The effectiveness of the method was quantitatively evaluated by the database 27 volumes containing abnormal tissue types, including, but without limitation, in large transmural scars, small subendocardialnah scars, scattered indistinct light zone, through a comparison of the contours produced by the automatic method in accordance with the present invention, with contours drawn manually former is artami. The average deviation between the contours obtained manually and automatically, is about 1.5 pixels. Figure 6 shows the results of the automatic segmentation contours of the myocardium in one embodiment, the system 200. The first column contains the results of the adaptation of the reference image to the image computed from the data layers of images obtained from different patients. Arrows indicate areas in which adaptation can be enhanced further. In the second row of images shows the contours obtained from the combined model surfaces for surfaces of the myocardium. The third column shows the results of manual segmentation by experts. The quality of the segmentation is high, as the contours safely surround both normal and anomalous parts of the myocardium. The results allow us to reliably estimate the percentage of nonviable tissue, for example, by determining the boundaries of the scar tissue in each image and compute the area mentioned scar tissue.

Specialist in the art it will be obvious that the system 200 can be a useful tool to assist the physician in many aspects of its work.

Specialists in the art will further understand that other possible embodiments of the system 200. Between p the by, you can override the system blocks and redistribute their functions. Although the described embodiments of apply for medical images, it is also possible other use of the system, not related to medical applications.

The blocks of the system 200 can be implemented using a processor. Features of the mentioned blocks are usually performed under software product management system software. During execution of the software product system software is usually loaded into memory, such as RAM (memory), and executes it. The program may be loaded from a background memory device, such as a ROM (permanent memory), hard disk, or magnetic and/or optical storage device, or may be downloaded via a network such as the Internet. Optionally, a specialized chip can provide the aforementioned functions.

7 shows a block diagram of the sequence of operations in an exemplary performance of method 700 of determining the boundaries of anatomical structures in the image calculated by the layer image data. The method 700 starts at step 705 host to initialize the reference image in the image. In the execution of the method 700 step 705 of placing includes a step 7052 evaluation axis to estimate the axis of the left ventricle, the stage 7054 recognition for having raspoznavaniya infarction and stage 7056 initialization to determine the initial position of the reference image in the image. After step 705 placement method 700 proceeds to step 710 split to split the image into multiple areas, such as quadrants, with each image area represents the region of the anatomical structure. After step 710 partitioning method 700 proceeds to step 720 adaptation for adapting the reference image to the image on the basis of the evaluation function, while the evaluation function is a function of the parameters of the reference image and the amounts of images and their relative positions in the image, and on the basis of criteria, which must satisfy the calculated value of the evaluation function, and the evaluation function is defined on the basis of many areas of the image. After step 730 adaptation method 700 proceeds to step 730 combining to combine model surfaces with lots of reference images, each reference image is adapted to the image calculated by the layer image data, at step 720 adaptation of the method 700. In the execution of the method 700, the step of combining includes a step 732 affine transformation affine combine model surfaces with many reference images and create, thus, affine-combined surface model, then step 234 local deformation for local paffing combining affine-with the absorbed surface model with multiple reference images and create thereby locally-combined surface model, and then step 736 refinement to adapt locally-combined surface model to the image data. After step 730 combining the method 700 ends.

Specialist in this field of technology can change the order of certain steps or to perform some steps concurrently using threading models, multi-processor systems or multiple procedures, without deviating from the principles of the present invention. Optionally, at least two stages of the method in accordance with the present invention can be combined in a single step. Optionally, the step of the method in accordance with the present invention can be divided into many stages.

On Fig schematically presents a sample implementation of the device 800 image acquisition, using the system 200, while the said device 800 receiving image contains block 810 obtain CT images associated internal connection with the system 200, the input connector 801 and the output connector 802. The scheme effectively improves the ability of the device 800 image acquisition due to the provision of the aforementioned device 800 of obtaining images of the useful features of the system 200.

Figure 9 schematically presents a sample implementation of the workstation 900. The workstation contains the system bus 901. About essor 910, memory 920, adapter 930 disk I/o and user interface (UI) 940 have a working connection to the system bus 901. Disk storage device 931 has a working connection to the adapter 930 disk I/o. Keyboard 941, mouse 942 and display 943 have a working relationship with UI 940. System 200 in accordance with the present invention, executed in the form of a computer program, stored in the disk storage device 931. Workstation 900 is configured to load the program and input data in the memory 920 and program execution in the processor 910. The user can enter information into the workstation 900 using keyboard 941 and/or mouse 942. The workstation is configured to display information to a display device 943 and/or drive 931. Specialist in the art it will be obvious that there are many other options for implementing the workstation 900 known in the art, and is presented here an implementation option is intended to illustrate the invention and is not subject to interpretation in the sense of limiting the invention, the concrete option implementation.

It should be noted that the above embodiments of illustrate and not limit the present invention, and specialists in the art will be able razrabotat the alternative implementation, not beyond the scope of the claims appended claims. In the claims, none of the items in brackets cannot be interpreted in the sense of the limitations of the claim. The wording "comprising" does not exclude the presence of elements or steps not listed in the claim or in the description. The indefinite article, denoting the singular before the element does not exclude the presence of many of the above elements. The invention can be implemented using hardware, contains several distinct elements, and by using a computer with a stored program. In the claims, listing several blocks of the system, a number of the aforementioned blocks may be implemented by one and the same item of hardware or software. The application of the definition of the first, second, third, etc. does not indicate any order. Mentioned adjectives should be interpreted as assignment.

1. The system (200) for determining the boundaries of anatomical structures in the image (11; 12), calculated on the layer image data, characterized in that the system (200) includes
block (210) split to split image (11; 12) on many of the plots (Q1; Q2; Q3; Q4) images, where each image represents the region of the anatomical structure; the
block (220) adaptation to adapt the reference image (30) to the image (11; 12) on the basis of the merit function and the evaluation function is a function of the parameters of the reference image and the amounts of images and their relative positions in the image, and on the basis of criteria, which must satisfy the calculated value of the evaluation function,
moreover, the evaluation function is defined on the basis of many areas of the image.

2. The system (200) according to claim 1, in which the block (220) adaptation is additionally configured to detect abnormal tissue in an anatomical structure of an image area from a variety of areas of the image, and the evaluation function contains a member that contributes to the evaluation function, when the above-mentioned image area is detected mentioned abnormal tissue.

3. The system (200) according to claim 1, in which the adaptation of the reference image based on the search parameter values of the reference image, when the evaluation function satisfies the condition criterion using the absorbing of the search algorithm.

4. The system (200) according to claim 1, in which the reference image is a closed ribbon (30), limited external (301) and internal (302) closed contours.

5. The system (200) according to claim 1, in which the split is a sector split.

6. The system (200) according to claim 1, ei is niteline containing block (230) combining to combine model surfaces with lots of reference images, each reference image is adapted to the image calculated by the layer image data, and at least one reference image adapts unit (220) adaptation of the system (200).

7. The system (200) according to claim 6, in which the block (230) combination contains
block (232) affine transformation affine combine model surfaces with multiple reference images, creating, thus, affine-combined surface model;
block (234) local deformation for local paffing combining affine-combined surface model with multiple reference images, creating, thus, a locally-combined surface model; and
block (236) specifications for the adaptation of locally combined surface model to the image data.

8. The method (700) of the definition of the boundaries of anatomical structures in the image (11; 12), calculated on the layer image data, characterized in that the method (700) includes the following stages:
- step (710) split to split image (11; 12) on many of the plots (Q1; Q2; Q3; Q4) images, where each image represents the region of the anatomical structure; and
stage (720) adaptation to adapt the reference image (30) to the image (11; 12) on the basis of the merit function and the evaluation function is a function of the parameters of the reference and what the considerations applying and numerical values of the image and their relative positions in the image, and on the basis of criteria, which must satisfy the calculated value of the evaluation function,
moreover, the evaluation function is defined on the basis of many areas of the image.

9. The device (800) obtain an image containing the system (200) according to claim 1.

10. Workstation (900)containing the system (200) according to claim 1.

11. The computer-readable medium containing stored thereon a computer program product, which when implemented by a computer system prompts the computer system to perform a method of determining the boundaries of anatomical structures in the image (11; 12), calculated on the layer image data, and a computer system includes a processor unit and a memory, wherein the computer program product, after boot, provides mentioned the CPU unit's functional ability to perform the following tasks:
- split image (11; 12) on many of the plots (Q1; Q2; Q3; Q4) images, where each image represents the region of the anatomical structure; and
- adaptation of the reference image (30) to the image (11; 12) on the basis of the merit function and the evaluation function is a function of the parameters of the reference image and the amounts of images and their relative positions in the image, and on the basis of criteria, which must satisfy vicis the military value of the evaluation function,
moreover, the evaluation function is defined on the basis of many areas of the image.



 

Same patents:

FIELD: information technology.

SUBSTANCE: method can be used for remote control of production process infrastructure in an apparatus (1) for extracting, processing and/or producing materials. This process is controlled by a control system performed with one or more computers, and with one or more computer display devices (10). At least one computer has a graphic three dimensional model of the process infrastructure, where the position of the process infrastructure in the three-dimensional model is selectable. The selected position is identified from an observation point in that model. Information on observation points from that model is used to arrange one or more cameras in the real apparatus. Images of the real apparatus are combined with the three-dimensional model using various techniques.

EFFECT: improved control of a technological process.

19 cl, 11 dwg

FIELD: medicine.

SUBSTANCE: invention relates to means for diagnosing neurodegenerative diseases. Device contains module of obtaining images which receives visual data about patient's brain state, and image analyser, made with possibility of determining quantitative index, which shows degree of development of neurodegenerative disease of patient's brain on the basis of visual data with application of probability mask for determination of studied areas on the image, specified by visual data. Method of clinical assessment includes stages of obtaining visual data and their analysis for determination of quantitative index, which makes it possible to assess degree of development of neurodegenerative diseases of patient's brain with application of probability mask. Software carrier contains computer programme, settings of data processing device for its performance of at least one of method stages.

EFFECT: invention facilitates early diagnostics and control of neurodegenerative diseases, for instance, Alzheimer's disease.

25 cl, 8 dwg

FIELD: information technologies.

SUBSTANCE: modules from a module 23 for calculation of a quantitative index of blur extent to a module 27 for calculation of a quantitative index of colour intensity withdraw the quantitative value of the specified characteristic from the input image, and the quantitative index is calculated using a separately taken characteristic, which characterises assessment of the input image on the basis of this characteristic. Thus, the module 24 for calculation of the quantitative index of brightness withdraws characteristics and the brightness value as a quantitative value from the input image, and calculates the quantitative index of brightness, which characterises assessment based on distribution of brightness value in the area occupied by the object, in the input image. The module 28 for calculation of the total quantitative index calculates the total quantitative index on the basis of each quantitative index of separately taken characteristics, and this total index characterises assessment of image fixation condition for the input image.

EFFECT: more accurate and effective assessment of the input image.

7 cl, 28 dwg

FIELD: physics.

SUBSTANCE: method includes an image forming step (710); a step (720) of placing a thickness gauge (21) in image data space; a scaling step (730) for selecting the scale of the thickness gauge (21) using a scaling factor in said data space; a step (740) of moving the thickness gauge (21) in said data space; and a thickness gauge forming step (750) for imaging thereof; wherein the object is measured based on the scaling factor.

EFFECT: high accuracy of determining points on the surface of the structure of a measured object on a medical image.

10 cl, 9 dwg

FIELD: physics.

SUBSTANCE: when performing repetitive scans on a patient using a magnetic resonance imaging machine or the like, patients often tend to move as they relax during a lengthy scanning session, causing movement in the volume or portion of the patient being scanned. A prospective motion correction component accounts for patient movement by calculating transformation data characterising patient movement in multiple planes, as well as rotational movement, and a host evaluates the change in position relative to a most recent scanning geometry of the patient or dynamic volume. In this manner, correction or adjustment to the scanning geometry employed by an associated scanner is made only for the difference between the current geometry and the most recent geometry in order to exclude redundant adjustment that can result in oscillatory over- and under-compensation during adjustments.

EFFECT: cutting scanning time and improving quality of scanning with real time motion correction.

23 cl, 10 dwg

FIELD: information technologies.

SUBSTANCE: video cameras are arranged in pairs, at the same time they are synchronised with each other and calibrated in respect to each other by three coordinates in general objects, forming a stereo module from two video cameras, stereo modules are arranged at the specified fixed distance from each other in the amount of two and more, at the same time each stereo module realises independent 3D reconstruction of its visible part of the person's face, the reconstructed parts of the person's face are combined into a common 3D reconstruction of the person's face, at the same time continuous or periodical calibration of stereo modules is carried out between each other by video images of cameras without suspension of 3D reconstruction of faces, using the 3D reconstruction of the person's face built with the help of all stereo modules, comparison is carried out between the face of the identified person and the basic face image, using comparison results, the personal identification is carried out.

EFFECT: improved probability of personal identification.

4 cl, 2 dwg

FIELD: medicine.

SUBSTANCE: group of inventions relates to medicine, visualization of vessels and their connection with pathological changing. Data of 3-dimensional image, reflecting spatially changing degree of connection of vessels between areas of data in 3-dimensional image and pathological changing, are created. Data can be represented by means of displaying maximal intensity projection (MIP), where image brightness represents degree of vessel participation in blood supply of pathological changing. Corresponding computer-readable carriers are used in method realisation. Described methods of visualisation can be useful in visualisation of connectedness with structures which are not pathological changes, and in visualisation of connectedness which is not connectedness of vessels.

EFFECT: inventions provide representation of degree of connectedness of vessels by means of relative brightness of vessel with increased reliability with respect to image interference.

12 cl, 4 dwg

FIELD: medicine.

SUBSTANCE: invention relates to medical equipment. System of visualisation in vivo contains capsule with animal, which includes one or more holders or bed, cylindrical cover, sensors of physiological parameters, identifier and docking interface, providing control, heating and anesthesia for one or more animals. System contains one first means of visualisation for collection in vivo data of animals in the field of visualisation of the first visualisation means, which includes docking port and reading device. System contains reconstructing processor, which reconstructs presented image from in vivo data for visualisation, installation unit or preparation unit, which includes docking port. System contains research work station, which includes electronic interface for interaction with the first visualisation means, user interface, which provides user with possibility of projecting new research and modification of existing research. System contains display devices, which displays vital parameters of animals during visualisation and preparation, and system of animal control and anesthesia for control of vital signals from animals during influence with sedative medication and for provision of regulated introduction of anesthesia into object via several docking interfaces.

EFFECT: application of the invention will make it possible to assess several biological parameters simultaneously within one visualisation procedure.

13 cl, 7 dwg

FIELD: information technology.

SUBSTANCE: method, apparatus and system for model-based playfield registration. An input video image is processed. The processing of the video image includes extracting key points relating to the video image. Further, whether enough key points relating to the video image were extracted is determined, and a direct estimation of the video image is performed if enough key points have been extracted and then, a tomographic matrix of the final video image based on the direct estimation is generated.

EFFECT: high stability of search and acceleration thereof, smaller search space.

12 cl, 11 dwg

FIELD: information technology.

SUBSTANCE: invention relates to an image forming apparatus in which a developing apparatus is installed, having a plurality of developer transporting elements, and a latent image formed on an image bearing element is developed by the developing apparatus. The image forming apparatus includes an image forming unit which serves as the support of the image bearing element, which enables said element to rotate, a developing unit which includes a first developer transporting element which is meant to develop the latent electrostatic image formed on the image bearing element, and a second developer transporting element which is meant to develop the latent electrostatic image formed on the image bearing element; a first gap adjusting element which is meant to adjust the gap between the image bearing element and the first developer transporting element; a second gap adjusting element which is meant to adjust the gap between the image bearing element and the second developer transporting element; a propelling element which is meant to move the developing unit towards the image forming unit so that the first gap adjusting element and the second gap adjusting element are pressed to a support region formed on the image bearing element; and a positioning element which is meant for positioning the developing unit relative the image forming unit.

EFFECT: reduced image deterioration due to the SD gap arising from the relative alignment error of paired hollow rollers.

7 cl, 20 dwg

FIELD: information technology.

SUBSTANCE: during training, a random number N with length of n bits is obtained; n artificial neurons are created; neuron inputs are randomly connected to memory cells with data on inclination of dermal ridges of fragments of examples of the classified fingerprint picture; further, the neurons are trained to output a random number N when presenting examples of the classified picture; the random number N is the hashed and the reference hash value is stored, as well as the connection of the neural network and its training parameters; and during authentication, there is repeated breaking up of the scanned picture into fragments, calculation of the average inclination on the fragment and the calculated value is transmitted to inputs of the neural network; the hash result is calculated at the output of the neural network and then compared with the reference value; in case of a match, personal authentication is continued, otherwise the authentication procedure is stopped once the time allocated for an authentication session runs out.

EFFECT: safeguarding the finger print picture and its biometric parameters which are used during biometric authentication, and providing conditions wherein it is impossible to find the owner of biometric parameters from biometric authentication data, while also cutting the average time of loading the authentication server.

2 cl, 3 dwg

FIELD: information technology.

SUBSTANCE: fragments of a digital image similar top human faces are detected in all regions of the digital image; the indicator of the fraction of pixels of skin colours inside the region of a fragment is determined on said fragments; a face is detected if a given threshold value is exceeded; if the threshold value of the fraction of skin colour inside the region of the found face is exceeded, the initial colour palette of human skin is adjusted; outlines of the object of the colour skin are determined; an indicator of geometric similarity of the found outline with typical body outlines is calculated; fragments similar to typical naked body parts are searched; the image is analysed for pornographic visual content.

EFFECT: high accuracy of analysing content of a digital image.

16 cl, 2 dwg

FIELD: information technologies.

SUBSTANCE: method is proposed to identify contours of moving objects, including detection of moving pixels according to a principle of inter-frame subtraction, detection of direction of movement of detected pixels with account of direction of movement of adjacent pixels, generation of moving objects by combination of adjacent pixels with one direction and availability of pixels in an eight-connected vicinity, spatial differentiation of detected objects by a Sobel operator, skeletonisation of contour lines on a gradient image by the method of suppression of points of non-maximum brightness, threshold processing of a skeleton gradient image on the basis of maximum and minimum of enthropy.

EFFECT: higher accuracy in detection of areas of moving objects and higher speed for identification of image contours.

1 dwg

FIELD: physics; computer engineering.

SUBSTANCE: invention relates to computer engineering and can be used for detecting image contours and subsequent input of the result into a computer. The device has a bus controller, RAM for reading (recording) intermediate image processing results, image filter unit, a contour detecting unit and a microprocessor.

EFFECT: more accurate detection of image contours.

1 dwg

FIELD: physics.

SUBSTANCE: invention relates to location of part of a document captured in an image. A method is proposed for determining part of a document, which corresponds to the captured image, when the user makes a flourish in the document. Location of some of the images will be determined, for example by analysing pattern on the capture image, or from results of pixel-by-pixel comparison of the image with the document. Location of other images will be determined through segmentation of the sequence of images in a group of corresponding forms of the flourish. Information relating to localised images in the segment can then be used for determining position of non-localised images in the segment. The search area of a non-localised image in the document is established based on the position of the previous localised image and maximum speed or maximum acceleration of the pen. Turning and scale of the non-localised image are evaluated, as well as turning and scale of the localised image, and the non-localised image is deformed using this turning and scale. Pixel-by-pixel comparison between the deformed non-localised image and the search area in the document is done. If deformation of the non-localised image is successfully matched, parametres of image transformation are further specified.

EFFECT: increased efficiency of locating part of a document.

10 cl, 25 dwg

FIELD: physics, image processing.

SUBSTANCE: invention is related to digital processing of image in process of scanning and copying, and especially to the field of colour and black-and-white text segmentation, when text is automatically extracted from scanned document. According to the present invention, method of text segmentation by colour criterion consists in performance of the following operations: initial image is broken down into non-overlapping units of pixels; new image Z is generated, in which each pixel represents corresponding unit of initial image; in process of scanning serial units are selected from initial image; classification is carried out for current unit by criterion "monochromatic/colour" in space of opposite colours; initial colour unit RGB is broken down into monochromatic units R, G and B; detector of Laplacian-Gaussian edges is applied with specified threshold T to monochromatic unit; number of edges is calculated for each pixel in Z; classification of "text/non-text unit" is performed by comparison of edge number with set threshold C; classified channels are combined, using logical operator OR.

EFFECT: provision of universal approach to preliminary processing of initial document, providing for faster printing of copy and saving of multifunctional device resources.

4 cl, 3 dwg

The invention relates to information technology in medicine

The invention relates to automatic control and computer engineering and can be used to analyze the geometric parameters of the images of objects in computer vision systems

FIELD: physics, image processing.

SUBSTANCE: invention is related to digital processing of image in process of scanning and copying, and especially to the field of colour and black-and-white text segmentation, when text is automatically extracted from scanned document. According to the present invention, method of text segmentation by colour criterion consists in performance of the following operations: initial image is broken down into non-overlapping units of pixels; new image Z is generated, in which each pixel represents corresponding unit of initial image; in process of scanning serial units are selected from initial image; classification is carried out for current unit by criterion "monochromatic/colour" in space of opposite colours; initial colour unit RGB is broken down into monochromatic units R, G and B; detector of Laplacian-Gaussian edges is applied with specified threshold T to monochromatic unit; number of edges is calculated for each pixel in Z; classification of "text/non-text unit" is performed by comparison of edge number with set threshold C; classified channels are combined, using logical operator OR.

EFFECT: provision of universal approach to preliminary processing of initial document, providing for faster printing of copy and saving of multifunctional device resources.

4 cl, 3 dwg

FIELD: physics.

SUBSTANCE: invention relates to location of part of a document captured in an image. A method is proposed for determining part of a document, which corresponds to the captured image, when the user makes a flourish in the document. Location of some of the images will be determined, for example by analysing pattern on the capture image, or from results of pixel-by-pixel comparison of the image with the document. Location of other images will be determined through segmentation of the sequence of images in a group of corresponding forms of the flourish. Information relating to localised images in the segment can then be used for determining position of non-localised images in the segment. The search area of a non-localised image in the document is established based on the position of the previous localised image and maximum speed or maximum acceleration of the pen. Turning and scale of the non-localised image are evaluated, as well as turning and scale of the localised image, and the non-localised image is deformed using this turning and scale. Pixel-by-pixel comparison between the deformed non-localised image and the search area in the document is done. If deformation of the non-localised image is successfully matched, parametres of image transformation are further specified.

EFFECT: increased efficiency of locating part of a document.

10 cl, 25 dwg

Up!