This application claims the benefit of Korean Patent Application No. 10-2010-0068674, filed on Jul. 15, 2010, in the Korean Intellectual Property Office, the entire disclosure of which is incorporated herein by reference for all purposes.
1. Field
The following disclosure relates to a method and apparatus for processing an image, and a medical image system employing the apparatus.
2. Description of the Related Art
In radiography that uses dual energy, calibration of an image processing apparatus is generally performed using a wedge phantom having a stepped shape or a triangular shape, and by estimating a functional relation between a dual energy logarithm signal corresponding to a negative logarithm of a signal output from a radiography detector, and a thickness of an object to be imaged. The estimated functional relation obtained from the performing of the calibration may allow a dual energy radiation image to be converted into a material ratio and a material thickness.
Provided are methods and apparatuses for processing an image, wherein an input parameter obtained from a dual energy radiation image is processed by applying a calibration model.
Provided are medical image systems employing the apparatuses.
Additional aspects will be set forth in part in the description which follows and, in part, will be apparent from the description, or may be learned by practice of the presented embodiments.
According to an aspect of the present invention, a method of processing an image is provided. The method includes generating a calibration model by learning an intensity-target data set obtained from a parameter of a test subject, and estimating a target. The estimating includes applying the calibration model to a parameter of an input subject and calibrating the parameter of the input subject.
The parameter of the test subject may be intensity obtained from a dual energy radiation image of a phantom for calibration, and the parameter of the input subject is intensity obtained from a dual energy radiation image of the input subject.
The parameter of the test subject may include additional information about the phantom for calibration, and the parameter of the input subject comprises additional information about the input subject.
The additional information may be information about total thicknesses of the phantom for calibration and the input subject, and different calibration models may be selected according to the total thickness of the input subject.
The generating may include learning the intensity-target data set by using support vector regression (SVR).
The phantom for calibration may be formed by overlapping at least two ramp wedge phantoms formed of at least two materials.
The method may include when the parameter of the input subject exists outside an intensity range used while generating the calibration model, projecting the parameter of the input subject within the intensity range used while generating the calibration model, and calibrating the parameter of the input subject with an intensity closest to the parameter of the input subject.
One of the two materials may be polycarbonate and another of the two materials may be polyethylene.
According to another aspect of the present invention, an apparatus for processing an image is provided. The apparatus includes a calibration model generating unit configured to generate a calibration model by learning an intensity-target data set obtained from a parameter of a test subject, and a target estimating unit configured to estimate a target. The estimating includes applying the calibration model to a parameter of an input subject and calibrating the parameter of the input subject.
The calibration model generating unit may include an intensity-target mapping unit configured to obtain the intensity-target data set by mapping intensity and the target obtained from the parameter of the test subject, and a learning unit configured to learn the intensity-target data set by applying a regression analysis to the intensity-target data set.
The parameter of the test subject may be intensity obtained from a dual energy radiation image of a phantom for calibration, and the parameter of the input subject is intensity obtained from a dual energy radiation image of the input subject.
The parameter of the test subject may include additional information about the phantom for calibration, and the parameter of the input subject comprises additional information about the input subject.
The additional information may include information about total thicknesses of the phantom for calibration and input subject, and different calibration models may be selected according to the total thickness of the input subject.
The apparatus may include when the parameter of the input subject exists outside an intensity range used while generating the calibration model, the target estimating unit projects the parameter of the input subject within the intensity range used while generating the calibration model, and calibrating the parameter of the input subject with an intensity closest to the parameter of the input subject.
The parameters of the test subject and the input subject may be respectively intensities obtained from a radiation image.
The radiation image may include at least a first energy image and a second energy image.
The apparatus may be disposed in a remote place.
The phantom for calibration may be formed by overlapping at least two ramp wedge phantoms formed of at least two materials, and one of the two materials is polycarbonate and another of the two materials is polyethylene.
According to another aspect of the present invention, a method of processing an image is provided. The method includes generating an input parameter including intensities of first and second energy images, applying a calibration model to the generated input parameter, calibrating an error of the input parameter by applying the calibration model, and estimating a target based on the calibrated input parameter and the error.
Other features and aspects may be apparent from the following detailed description, the drawings, and the claims.
Throughout the drawings and the detailed description, unless otherwise described, the same drawing reference numerals will be understood to refer to the same elements, features, and structures. The relative size and depiction of these elements may be exaggerated for clarity, illustration, and convenience.
The following detailed description is provided to assist the reader in gaining a comprehensive understanding of the methods, apparatuses, and/or systems described herein. Accordingly, various changes, modifications, and equivalents of the systems, apparatuses and/or methods described herein will be suggested to those of ordinary skill in the art. Also, descriptions of well-known functions and constructions may be omitted for increased clarity and conciseness.
Referring to
Meanwhile, a regression analysis may be applied while learning the intensity-target data set. More specifically, a support vector regression (SVR) may be applied, as the regression analysis, but any other method may be used to learn the intensity-target data set.
In this example, the phantom for calibration used as the test subject may be formed by overlapping at least two ramp wedge phantoms formed of at least first and second materials. When the dual energy radiation image is used for mammography, the first and second materials may be 1) polycarbonate corresponding to a glandular tissue and 2) polyethylene corresponding to an adipose tissue, respectively. When a total thickness of the phantom for calibration is uniform, the first and second materials may be combined in all possible ratios of thickness.
For example, when the total thicknesses of the test subject are 3 cm, 4 cm, and 5 cm, respectively, an actual thickness of a predetermined material corresponding to an intensity combination of the first energy image (high) and the second energy image (low) may be shown in Tables 1 through 3 below according to each total thickness.
In other words, referring to Table 1, when the total thickness of the phantom for calibration of a compression paddle is 3 cm, the actual thickness of the first material, for example, the adipose tissue, forming the phantom for calibration is 1 mm.
The intensity-target data set may be formed in response to each total thickness of the test subject as described above, and the intensity-target data set according to each total thickness may be learned by performing a regression analysis on the intensity-target data set, thereby generating the calibration model according to each total thickness.
The input parameter generating unit 130 may generate the parameter of the input subject from the dual energy radiation image obtained with respect to the input subject. Here, the parameter of the input subject may be intensity obtained from the dual energy radiation image of the input subject. For example, the intensity may be obtained from each of at least a first energy image constituting a low energy image, and a second energy image constituting a high energy image. In this example, the parameter of the input subject may further include additional information related to a total thickness of the input subject, in addition to the intensity obtained from the dual energy radiation image. By further including the additional information related to the total thickness as the parameter of the input subject, different calibration models may be selected while considering different total thicknesses of the input subject to calibrate the parameter of the input subject. Accordingly, the addition information may allow calibration accuracy to be increased.
The target estimating unit 150 may estimate a target corresponding to each combination of intensities by calibrating the parameter of the input subject by applying the calibration model to the parameter of the input subject. By including not only the intensities of the first and second energy images but also the total thickness of the input subject as the parameter of the input subject, the calibration model corresponding to the total thickness of the input subject may be selected. More specifically, the intensities of the first and second energy images may be received as the parameter of the input subject. If the combination of the intensities of the first and second energy images is outside a range of the intensity-target data set to which the calibration model belongs, a corresponding target may be estimated by calibrating the combination inside the range, by referring to a default calibration model or a selected calibration model and the intensity-target data set used to generate each calibration model. In other words, if the parameter of the input subject has an undefined value, the undefined value may be mapped to a value closest to a functional relation of 1) a thickness of the first material, 2) a thickness of the second material, 3) the intensity of the first energy image, and 4) the intensity of the second energy image, which are defined in the intensity-target data set used to generate the calibration model. Thus, by including the total thickness of the phantom for calibration as the additional information in addition to the dual energy radiation image, 1) different calibration models may be generated in response to the total thickness of the phantom for calibration and 2) different calibration models may be selected in response to the total thickness of the input subject. Accordingly, an estimation error may be reduced in regard to the thickness of the target estimated from the dual energy radiation image of the input subject. Also, the estimated thickness of the target may be a thickness of a predetermined material forming the input subject.
Accordingly, the estimated thickness may be similar to a pre-known thickness 1) even when a material of the input subject and a material of the phantom for calibration are different, 2) even when a dual energy radiation image of the input subject is damaged by noise or the like, or 3) even when a difference between the thickness of the phantom for calibration and a thickness of a material to be obtained from the dual energy radiation image of the input subject is remarkably large.
For example, the target estimating unit 150 estimates the target, for example thickness information, through the calibration using the selected calibration model in response to each combination of the intensities of the first and second energy images. The thickness information is shown in Table 4 below.
In other words, referring to Table 4, when the input subject has a predetermined total thickness, an actual thickness of 1) the first material, for example, adipose tissue, which the input subject is formed of, is estimated to be 15 mm if the intensity of the first energy image is 20 and 2) the intensity of the second energy image is 35.
The display unit 170 may display the thickness information of the target, for example, the first material, estimated by the target estimating unit 150 through the calibration according to each combination in 3-dimensions (3D).
The storage unit 190 may store target information that is estimated by the target estimating unit 150 through the calibration according to each combination.
The estimated target information may be transmitted to a transmitter that transmitted the parameter of the input subject or to the remote medical image system through the communicator, by wire or wirelessly.
Referring to
The intensity-target mapping unit 230 may map each combination of the intensities of the first and second images in the dual energy radiation image of the test subject to the actual thickness of the predetermined material forming the test subject in each combination to the target. In the example embodiment, mapping results may be stored as a lookup table.
The learning unit 250 may learn the intensity-target data set provided by the intensity-target mapping unit 230 by applying the regression analysis to the intensity-target data set. Thus, the learning unit may generate the calibration model for the intensity-target data set.
Referring to
The learning unit 350 may learn the intensity-target data set provided by the intensity-target mapping unit 330 by applying the regression analysis to the intensity-target data set. Thus, the calibration model for the intensity-target data set may be generated according to the total thickness of the test subject.
Referring to
Meanwhile, as shown in
Accordingly, the phantom for calibration, which is able to deal with various total thicknesses is required, and thus, for example, the phantom 410 of
Also, in SVR, the regression analysis may be performed on a data distribution by using a kernel assuming nonlinearity, and thus a fitting error may be remarkably reduced even when an intensity-target relation of a subject to be imaged may be very nonlinear.
Also, since the total thicknesses of the phantom for calibration and input subject are used as parameters for generating the calibration model, a fitting error may be remarkably reduced even for the input subject having a different total thickness from the phantom for calibration.
Referring to
In operation 830, the calibration model may be applied to the input parameter generated in operation 810. Thus, when the total thickness of the input subject is included in the input parameter, a calibration model corresponding to the total thickness of the input subject may be selected.
In operation 850, an error of the input parameter may be calibrated by applying the calibration model in operation 830, and a target corresponding to the calibrated input parameter, for example, a thickness of a predetermined material forming the input subject, may be estimated.
As described above, according to the one or more of the above example embodiments, even when error data outside a normal range is provided as an input parameter while processing a dual energy radiation image, the input parameter may be calibrated by projecting the input parameter onto a location nearest to the error data within the normal range. Accordingly, a desired target, for example, a thickness of a predetermined material forming an input subject, may be accurately obtained from the dual energy radiation image of the input subject.
Program instructions to perform a method described herein, or one or more operations thereof, may be recorded, stored, or fixed in one or more computer-readable storage media. The program instructions may be implemented by a computer. For example, the computer may cause a processor to execute the program instructions. The media may include, alone or in combination with the program instructions, data files, data structures, and the like. Examples of computer-readable media include magnetic media, such as hard disks, floppy disks, and magnetic tape; optical media such as CD ROM disks and DVDs; magneto-optical media, such as optical disks; and hardware devices that are specially configured to store and perform program instructions, such as read-only memory (ROM), random access memory (RAM), flash memory, and the like. Examples of program instructions include machine code, such as produced by a compiler, and files containing higher level code that may be executed by the computer using an interpreter. The program instructions, that is, software, may be distributed over network coupled computer systems so that the software is stored and executed in a distributed fashion. For example, the software and data may be stored by one or more computer readable recording mediums. Also, functional programs, codes, and code segments for accomplishing the example embodiments disclosed herein can be easily construed by programmers skilled in the art to which the embodiments pertain based on and using the flow diagrams and block diagrams of the figures and their corresponding descriptions as provided herein. Also, the described unit to perform an operation or a method may be hardware, software, or some combination of hardware and software. For example, the unit may be a software package running on a computer or the computer on which that software is running.
A number of examples have been described above. Nevertheless, it will be understood that various modifications may be made. For example, suitable results may be achieved if the described techniques are performed in a different order and/or if components in a described system, architecture, device, or circuit are combined in a different manner and/or replaced or supplemented by other components or their equivalents. Accordingly, other implementations are within the scope of the following claims.
Number | Date | Country | Kind |
---|---|---|---|
10-2010-0068674 | Jul 2010 | KR | national |