KR102194009B1 - POCS-based external information restoration method applicable to field modulation CT acquisition system - Google Patents

Abstract

In the present invention, a reconstructed image containing external information is obtained by reconstruction of a non-cut sinogram (SNT) using a POCS-based reconstruction algorithm, and the obtained reconstructed image is again forward projection (forward projection). To generate a virtual sinogram (SOUT) including external information, and use the virtual sinogram (SOUT), the uncut sinogram (SNT), and the cut sinogram (ST) The present invention relates to a POCS-based external information restoration method applicable to a field modulation CT acquisition system that can obtain a high-quality image including external information by obtaining a reconstructed image using a nogram.
The POCS-based external information restoration method in the field-modulated CT acquisition system of the present invention obtains projection data by irradiating X-rays on the subject at an angle of 360 degrees centered on the subject, adjusting the field of view, and cutting projection. A projection data acquisition step of alternately acquiring projection data of an image and projection data of an uncut projection image; The sinogram is obtained from the projection data received from the projection data acquisition unit, and the uncut sinogram (SNT) and the cut sinogram (ST) are extracted from the obtained sinogram. Nogram acquisition step; Applying and forwarding the POCS algorithm, which acquires a reconstructed image by applying the POCS algorithm to the uncut sinogram (ST), and obtains a virtual sinogram (SOUT) including external information by forward projecting the obtained reconstructed image Projection data recovery step by projection; A CT image by performing filtered back projection (FBP) reconstruction on the projection data obtained by summing the uncut sinogram (SNT), the cut sinogram (ST) and the virtual sinogram (SOUT). It characterized in that it comprises a; obtaining, FBP reconstructed CT image acquisition step.

Description

POCS-based external information restoration method applicable to field modulation CT acquisition system}

The present invention relates to a POCS (projection onto convex sets, convex set projection technique)-based reconstruction method that improves the reconstruction accuracy of external information in a field modulation CT acquisition system, and in more detail, POCS-based reconstruction Reconstructing the uncut sinogram (SNT) using an algorithm to obtain a reconstructed image containing external information, and then re-projecting the obtained reconstructed image forward (forward projection) By creating a sinogram (SOUT), a complete sinogram using the existing sinograms, such as uncut sinogram (SNT) and cut sinogram (ST), and the virtual sinogram (SOUT). A POCS-based external information restoration method applicable to a field modulation CT acquisition system that can complete a gram and obtains a reconstructed image using the completed sinogram to obtain an image of good quality including external information will be.

Existing ROI (Region of Interest) CT imaging is an imaging technique that can reduce dose while reducing unnecessary exposure to patients. However, since the existing ROI (region of interest) CT cannot reconstruct external information, a technique called a field modulation CT acquisition method, which is a new imaging technique, has been proposed and studied.

For example, in the Proc SPIE Int Soc Opt Eng. on March 22, 2016, a paper by J. Webster Stayman et al., "Fluence-field side X using a multi-aperture device. -ray CT (Fluence-Field Modulated X-ray CT using Multiple Aperture Devices)".

(https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4838035/)

As another example, "Dynamic fluence field modulation in computed tomography using a multi-aperture device" published by Gang GJ et al. on March 21, 2019 at the Society of Medical Physics (Phys Med Biol.) using multiple aperture devices)".

(https://www.ncbi.nlm.nih.gov/pubmed/30939459)

Therefore, since the field modulation CT acquisition method and its system are known, a separate description thereof will be omitted below.

When the field-modulated CT acquisition method is used, external information can be restored to some extent, but there is a limit to complete restoration, so the accuracy and quality of the external information are degraded. Here, field modulation is the same concept as beam modulation.

To solve this, the present invention proposes to completely restore insufficient external information by using a POCS-based restoration algorithm.

That is, the sinogram obtained through the field modulation CT acquisition system is composed of a nontruncated sinogram (SNT) and a truncated sinogram (ST), and the present invention is a POCS-based Using a reconstruction algorithm, reconstruction is performed on the uncut sinogram (SNT) to obtain a reconstructed image containing external information, and the obtained reconstructed image is again forward projection to provide a virtual space containing external information. By creating a nogram (SOUT), a complete sinogram can be completed using the existing sinograms SNT, ST, and SOUT, and when a reconstructed image is obtained using the completed sinogram, external information We propose a POCS-based external information restoration method applicable to a field-modulated CT acquisition system that can acquire images of good quality including.

In general, CT X-ray scanners have a cylindrical shape, and an X-ray detector is provided on one side of the cylindrical shape, and an X-ray tube is provided on the other side (symmetrical part) of the X-ray detector, and is inserted into the cylindrical shape. The resulting subject is imaged. The transmitted rays emitted from the X-ray tube and transmitted through the subject are detected by the X-ray detector. In the X-ray detector, the amount of transmitted rays passing through the subject is added to the accumulated value.

To image the internal object by irradiating only the transmission line while leaving the subject in the cylindrical shape as it is, take a picture by rotating the subject from 0 to 180 degrees, or by rotating the beam while leaving the subject as it is, and passing through the subject. By calculating the amount of one beam, a multi-faceted CT image is created.

At this time, the amount of transmission is detected by the sensor of the X-ray detector, which is rotated from 0° to 180°, and the amount of transmission is detected by the sensor of the X-ray detector. The 360 degree representation of an object is obtained by mirroring data up to 180 degrees.

Here, a visualization of a graph obtained by taking an X-ray picture at various angles is called a sinogram.

The sinogram is a visualization of the cumulative transmission value that has passed through a specific part of the subject, on the horizontal axis, the angle from 0 to 180 degrees, and the height of the corresponding graph based on the central axis for each angle using contrast. That is, the sinogram is the amount of absorbed X-rays displayed on the XY plane. That is, the closer to the white, the more absorption occurs, and when the absorption occurs considerably more, it can be seen that the object blocks the transmitted beam.

In general, a sinogram is a result of physically performing a mathematical radon transformation, and to obtain an original image, it is obtained by performing an inverse radon transformation.

The POCS-based algorithm used in the present invention has the advantage of obtaining a more accurate and high-quality image by approaching an iterative algorithm, and also, it can restore more accurate external information than the existing ROI CT or interpolation method. Therefore, it can be usefully used in a field modulation CT acquisition method.

As a prior art, Korean Patent Laid-Open Publication No. 10-2011-0040164 relates to a sinogram restoration apparatus and method, which is the first image corresponding to a sinogram by reconstructing a sinogram. A metal material identification unit that creates an (initial image) and identifies a metal object in the initial image, and a metal area that creates a metal projection area that is an area corresponding to the metal material by forward projection of the metal material Reconstruction that generates a reconstructed sinogram by detecting a peripheral area related to the pixel value inside the metal projection area among the peripheral areas outside the secondary and metal projection area, and acquiring the pixel value inside the metal projection area from the detected surrounding area And an image generator configured to reconstruct the reconstructed sinogram to generate a reconstructed image corresponding to the reconstructed sinogram, and to combine the metallic material with the reconstructed image.

In Korean Patent Publication No. 10-2011-0040164, the accuracy and quality of external information is not considered because it does not take into account a cut sinogram, an uncut sinogram, and a virtual sinogram (SOUT) including external information. This falls.

The problem to be solved by the present invention is to obtain a reconstructed image including external information by performing reconstruction on a non-cut sinogram (SNT) using a POCS-based reconstruction algorithm, and then forward the obtained reconstructed image again. Projection) to generate a virtual sinogram (SOUT) including external information, and the conventional sinograms such as uncut sinogram (SNT) and cut sinogram (ST), and the virtual sinogram A complete sinogram can be completed using SOUT, and when a reconstructed image is obtained using the completed sinogram, a field modulation CT acquisition system that can acquire a good quality image including external information It is to provide an applicable POCS-based external information restoration method.

In order to solve the above problems, the POCS-based external information restoration method in the field modulated CT acquisition system of the present invention is, in the projection data acquisition unit, by irradiating X-rays to the subject at a 360 degree angle centered on the subject. Projection data is acquired, but by adjusting the field of view, the projected data of the cut-projected image and the projected data of the uncut-projected image are alternately acquired, and the number of acquired projected data is the'total number of projected data' A projection data acquisition step in which the number of projection data of is'the number of cut-off projection data and the number of projection data of the non-cut-off projection image is'the number of non-cut projection data; The calculation processing unit includes: a sinogram acquisition step of obtaining a sinogram of the total number of projection data from projection data of the total number of projection data acquired in the projection data acquisition step; The operation processing unit includes a non-cut sinogram extraction step of extracting a non-cut sinogram (SNT) of the number of non-cut projected data from the sinogram obtained in the sinogram acquisition step; The operation processing unit may include an image reconstruction step using a POCS algorithm for obtaining a reconstructed image by applying a POCS algorithm to the uncut sinogram (SNT) obtained in the uncut sinogram extraction step; The operation processing unit performs forward projection of the reconstructed image obtained in the image reconstruction step by the POCS algorithm using a virtual CT projector, and restores the projection data of the total number of projection data, which is a virtual sinogram (SOUT) including external information. , Projection data recovery step by forward projection; It characterized in that it comprises a.

The POCS-based external information restoration method includes: a cutting sinogram extraction step of extracting a cut sinogram (ST) of the number of cut projection data from the sinogram obtained in the sinogram acquisition step; A first summing step of summing the uncut sinogram (ST) extracted in the non-cut sinogram extraction step and the cut sinogram (ST) extracted in the cut sinogram extraction step; The operation processing unit, on the summation result of the uncut sinogram (SNT) and the cut sinogram (ST) output in the first summing step, the virtual sinogram output in the projection data recovery step by forward projection ( SOUT) to obtain projection data of the total number of projection data; Filtered back projection (FBP) on the summation result of the uncut sinogram (SNT), cut sinogram (ST), and virtual sinogram (SOUT) output in the second summation step ) Performing reconstruction, obtaining a reconstructed CT image, obtaining an FBP reconstructed CT image; further includes.

The cut-out projection image is an image photographed using only the region of interest as an irradiation field, and the uncut projection image is an image photographed using the entire view provided by the projection data acquisition unit as an irradiation field.

In the filtered reverse projection (FBP) reconstruction, the summation result output in the second summation step is corrected with a correction filter, and back-projection is performed using a virtual CT projector to obtain a reconstructed CT image.

The total number of projected data, cut projected data, and uncut projected data can be values set by user settings before the projected data acquisition step, and the total projected data number is 360, the cut projected data number is 72, and the ratio The number of cut projection data may be 288.

In addition, the present invention is characterized by a recording medium storing a computer program source for a POCS-based external information restoration method in a field modulated CT acquisition system.

In addition, the field-modulated CT acquisition system of the present invention obtains projection data by irradiating X-rays to the subject at an angle of 360 degrees centered on the subject, and by adjusting the field of view, the projection data of the cut projection image and, A projection data acquisition unit that alternately acquires projection data of an uncut projection image; Obtain a sinogram from the projection data received from the projection data acquisition unit, extract an uncut sinogram (SNT) and a cut sinogram (ST) from the obtained sinogram, and extract the uncut sinogram (ST) To obtain a reconstructed image by applying the POCS algorithm to the system, a virtual sinogram (SOUT) including external information is obtained by forward projecting the obtained reconstructed image, and the uncut sinogram (SNT) and the truncated sinogram And an operation processing unit configured to obtain a CT image by performing filtered back projection (FBP) reconstruction on projection data obtained by summing (ST) and the virtual sinogram SOUT. .

The projection data acquisition unit acquires projection data by irradiating X-rays on the subject at a 360-degree angle centered on the subject, and alternately acquiring the projected data of the cut-off projection image and the projection data of the non-cut-off projection image , The number of acquired projected data is'total projected data', the number of projected data of the cut-out projected image is'cut-off projected data', and the number of projected data of uncut-off projected images is'non-cut projected data'. do.

The calculation processing unit includes: a sinogram acquisition unit for obtaining a sinogram of the total number of projection data from projection data of the total number of projection data acquired by the projection data acquisition unit; A non-cut sinogram extraction unit for extracting a non-cut sinogram (SNT) of the number of non-cut projection data from the sinogram obtained by the sinogram acquisition unit; A POCS algorithm processing unit for obtaining a reconstructed image by applying a POCS algorithm to the non-cutting sinogram (SNT) obtained by the non-cutting sinogram extraction unit; Projecting the reconstructed image obtained from the POCS algorithm processing unit in the forward direction using a virtual CT projector to obtain projection data of the total number of projection data, which is a virtual sinogram (SOUT) including external information. A projection data recovery unit by forward projection; A cut sinogram extraction unit for extracting a cut sinogram (ST) of the number of cut projection data from the sinogram obtained by the sinogram acquisition unit; A first adder for summing the uncut sinogram (ST) extracted from the uncut sinogram extraction unit and the cut sinogram (ST) extracted from the cut sinogram extraction unit; To the sum result of the uncut sinogram (SNT) and the cut sinogram (ST) output from the first adder, a virtual sinogram (SOUT) output from the projection data recovery unit by forward projection is added, A second adder for obtaining projection data of the total number of projection data; Filtered inverse projection (FBP) reconstruction is performed on the sum result of the uncut sinogram (SNT), the cut sinogram (ST), and the virtual sinogram (SOUT) output from the second adder, It further includes; an FBP reconstructed CT image acquisition unit that obtains a reconstructed CT image.

The POCS-based external information restoration method applicable to the field-modulated CT acquisition system of the present invention uses a POCS-based reconstruction algorithm to reconstruct a non-cut sinogram (SNT) to obtain a reconstructed image containing external information. , By forward projection of the obtained reconstructed image again to create a virtual sinogram (SOUT) containing external information, a complete sinogram using the existing sinograms SNT, ST, and SOUT Can be completed, and if a reconstructed image is obtained using the completed sinogram, a good quality image including external information can be obtained.

ROI CT image reconstruction has been developed to reduce the radiation dose and scattered radiation, and this method is relatively inexpensive and results in lower radiation doses than those encountered with conventional CT scans. However, the reconstructed image obtained by ROI CT scan does not contain external information. This can be limited to treatment plans or diagnoses that require external information. In the present invention, the quality of the reconstructed image is improved by introducing a field modulation (beam modulation) acquisition based on the POCS algorithm and a new cropped image restoration method.

That is, in the past, the accuracy and quality of external information were degraded by approaching based on an analytical algorithm, but the present invention has the advantage of obtaining more accurate and high quality images by approaching an iterative algorithm with a POCS-based algorithm. There is this.

In addition, the POCS-based reconstruction algorithm applied in the present invention is an efficient algorithm capable of restoring more accurate external information than the existing ROI CT or interpolation method, and is an algorithm that can be usefully used in a field modulation acquisition method.

1 is a block diagram showing an operation processing unit for applying a POCS-based external information restoration method applicable to the field modulation CT acquisition system of the present invention.
2 is a schematic diagram of a general field modulation acquisition method.
3 shows a 3D voxel abdominal DICOM phantom and sinogram obtained by simulation.
4 is an explanatory diagram illustrating a projection data acquisition unit for applying a field modulation method in a field modulation CT acquisition system.
5 is a flowchart of an image restoration method using a POCS algorithm in the present invention.
6 shows the three ROIs of the abdominal phantom.
7 is an example of a sinogram obtained from the abdominal DICOM phantom according to the present invention.
8 is. An example of reconstructed abdominal CT image is shown.
9 is an enlarged view of three ROIs in reconstructed abdominal CT images obtained by each method.
10 shows SSIM values obtained by four ROI-based methods.

Hereinafter, a POCS-based external information restoration method applicable to the field modulation CT acquisition system of the present invention will be described in detail with reference to the accompanying drawings.

The present invention proposes an alternative CT acquisition method called field modulation, that is, field modulation. The field modulation acquisition method obtains both uncut (SNT) and truncated projection (ST), so that external information of the ROI can be determined. Therefore, unlike conventional ROI imaging, external information can be restored using field modulation collection. However, the ratios of the internal and external projection information obtained by field modulation do not match. Thus, the reconstructed image carries truncation artifacts, which limits the ability to recover accurate external information. In order to solve this problem, the present invention proposes an interpolation-based restoration method and a POCS (projection to convex set, convex set projection method or convex projection) algorithm-based method.

The field modulation CT acquisition system 10 of the present invention includes a field modulation projection data acquisition unit 50 and an operation processing unit 100, and the operation processing unit may be a computer including MATLAB.

The projection data acquisition unit 50 is a means provided in a conventional field modulation CT acquisition system, and is a means for acquiring projection data by modulating an X-ray irradiation field in order to acquire a CT image using a field modulation method.

The projection data acquisition unit 50 modulates the X-ray irradiation field, so that the circumference (i.e., the subject) of the subject or the subject (hereinafter, collectively referred to as'subject' for convenience of description) should be centered on the subject. When the projection data (projection image) is acquired at an angle of 360 degrees.

At this time, when a projection image is acquired around the subject, a truncated projection image and a non-truncated projection image may be alternately acquired according to a user's setting.

Here, the cut-out projection image may be referred to as an image photographed by reducing the irradiation field only with the region of interest, and the non-cut projection image may be referred to as an image photographed with the entire view as the irradiation field.

In addition, the operation processing unit 100 uses the projection data obtained by the projection data acquisition unit 50 to perform calculation processing to obtain an image of good quality including external information restored by a POCS-based external information restoration method. It is a means to do.

1 is a block diagram showing an operation processing unit for applying a POCS-based external information restoration method applicable to the field modulation CT acquisition system of the present invention.

The operation processing unit 100 of the present invention includes a sinogram acquisition unit 120, a non-cut sinogram extraction unit 130, a POCS algorithm processing unit 140, a projection data recovery unit 150 by forward projection, and a cutting A sinogram extraction unit 160, a first adder 165, a second adder 170, and a filtered back projection (FBP) reconstructed CT image acquisition unit 180 are included.

The sinogram acquisition unit 120 is a projection data (projection image) obtained from the projection data acquisition unit 50 around a subject or a subject (hereinafter, collectively referred to as'subject' for convenience of explanation) at an angle of 360 degrees. From these, a sinogram of the total number of projection data is obtained. Here, the'total number of projection data' refers to the total number of (total) projection data obtained by the projection data acquisition unit 50 for convenience of explanation. In addition, the'sinogram of the total number of projection data' refers to a sinogram obtained by accumulating as much as the total number of projection data acquired by the projection data acquisition unit 50.

In general, projection data is a 2D image, and one section (1D) is extracted from the projected data, and the section (1D) for each angle is extracted and accumulated as a sinogram. In other words, the stack of 360 lines of projection data is marked as 360 sinograms.

For example, when the projection data acquisition unit 50 obtains one projection data at an angle of 1 degree around the subject or the circumference of the subject, 360 pieces obtained by projecting X-rays at an angle of 360 degrees around the subject From the projection data, 360 sinograms are obtained.

In addition, the projection image (data) acquired by the projection data acquisition unit 50 includes a cut-out projection image and a non-cut projection image. The number of cut-out projection images and the number of non-cut projection images may vary depending on the user's setting when shooting.

For example, according to a user's setting at the time of acquisition of projection data, out of the 360 projection data, 72 cut-out projection images may be used and 288 non-cut projection images.

The non-cut sinogram extraction unit 130 extracts a non-cut sinogram (SNT) of the number of non-cut projection images from the sinogram obtained by the sinogram acquisition unit 120. Here, the'number of uncut projected images (number of uncut projected data)' refers to the number of non-cut projected images (data) obtained by the projection data acquisition unit 50. In addition, the'non-cutting sinogram of the number of uncut projected images (SNT)' is a sinogram formed by stacking lines of projection data corresponding to the number of non-cutting projected images.

For example, in the sinogram obtained by the sinogram acquisition unit 120, the non-cut sinogram extraction unit 130 is a non-cut sinogram (SNT) of 72, which is the number of non-cut projection images, that is, 72 ratio. Cut sinogram (SNT) can be extracted. Here, the number of uncut projection images and the number of cut projection images are values set by the user.

The POCS algorithm processing unit 140 obtains a reconstructed image by applying a projection onto convex sets (POCS) algorithm to the non-cut sinogram (SNT) obtained from the non-cut sinogram extraction unit 130 do.

In general, the POCS (projection onto convex sets) algorithm is given a closed convex set Ci(i=1, 2,..., m) having a certain shape in a certain norme space, It refers to a series of algorithms that calculate one point in Ci of a common subset [numerical formula] by using the distance projection to each Ci. That is, the POCS algorithm is a method of forming a plurality of convex sets from the features of the original image, and repeatedly obtaining the original image belonging to the common part of each convex set by using nonlinear projection action elements.

The POCS (reconstruction) algorithm suppresses fringe artifacts due to the limitation of projection views by applying the total variation (TV) minimization to the algebraic reconstruction technique (ART) algorithm. In the traditional CT imaging problem, the sampling procedure can be considered as a discrete linear transform, and the imaging model links the two vectors through the following linear model.

g _o = Af

However, A is a system matrix of size M × N, that is, a system matrix representing the size of a unit image in the original image, and g ₀ is a measurement vector of model data, that is, data (vector) of the original image, and image f is a vector having M pixels and N voxels, that is, represents a detected image. The model data g ₀ is different from the measurement data g, and the measurement data g can be considered in the optimization formula as follows.

The inverse transformation (inversion, inversion, and inversion) of the equation of a linear model can be formulated as an optimization problem as follows, and the optimization problem can be solved using an optimization-based algorithm.

The reconstructed image, that is, the inverse transform (f ^* ) of the image f can be expressed as follows.

f ^* = argmin∥f∥ _TV D(f)≤ε

Here, D(f) is D(f) =|Af-g|.

D(f) =|Af-g| represents the Euclidean data divergence between the measured data g and the imaging model Af, and ε is a preselected positive parameter to accommodate the discrepancy between the measured data g and the imaging model Af .

The POCS algorithm was published in "Image reconstruction in circular cone-beam computed tomography by constrained, total-variation minimization", published in 2008 in Vol. 53, No. 17 of the Journal of Medical and Biophysics. In addition, it is also known in "Image Reconstruction and Restoration by the Method of Convex Convex Projections" published in January 1994 in MEDICAL IMAGING TECHNOLOGY Vol.12 No.l of Japan. Therefore, a more detailed description of the POCS algorithm is omitted here.

The projection data recovery unit 150 by forward projection forward-projects the reconstructed image obtained by the POCS algorithm processing unit 140 to recover projection data, which is a virtual sinogram SOUT including external information. The number of projection data obtained here is equal to the'total number of projection data'. Also, here, it is also possible to use a virtual CT projector for projection.

For example, if 360 sinograms obtained by the sinogram acquisition unit 120 are acquired, the projection data recovery unit 150 by forward projection forward projection of the reconstructed image obtained from the POCS algorithm processing unit, and external information. It restores 360 projection data, which is the included virtual sinogram SOUT.

The cut sinogram extraction unit 160 extracts a cut sinogram ST of the number of cut projection images from the sinogram obtained by the sinogram acquisition unit 120.

Here, the'number of cut-out projection images (number of cut-out projection data)' refers to the number of cut-out projection images (data) obtained by the projection data acquisition unit 50. In addition, the'cutting sinogram of the number of cut-out projection images (SNT)' is a sinogram formed by stacking lines of projection data corresponding to the number of cut-out projection images.

For example, the cut sinogram extraction unit 160 is a cut sinogram ST of 288 (a value set by the user), which is a preset number of cut projection images from the sinogram obtained by the sinogram acquisition unit. 288 cut sinograms (ST) can be extracted.

The first adder 165 includes an uncut sinogram (ST) extracted from the uncut sinogram extraction unit 130 and a cut sinogram (ST) extracted from the cut sinogram extraction unit 160 Add up

The second adder 170 converts the sum result of the uncut sinogram (SNT) and the cut sinogram (ST) output from the first adder 165 into the projection data recovery unit 150 by forward projection. Add the virtual sinogram (SOUT) output from (for example, 360 projection data). The number of projection data of this summation result is equal to the'total number of projection data'.

The FBP reconstructed CT image acquisition unit 180 includes a non-cut sinogram (SNT), a cut sinogram (ST), and a virtual sinogram (SOUT) output from the second adder 170. From the summation result (eg, 360 projection data), reconstruction is performed with filtered back projection (FBP) to obtain a filtered back projection (FBP) reconstructed CT image.

In general, filtered reverse projection (FBP) refers to performing a correction through filtering and performing reverse projection on a broken portion of an image. Here, the filtered reverse projection (FBP) is the sum of the uncut sinogram (SNT), the cut sinogram (ST), and the virtual sinogram (SOUT), and the correction filter (e.g., Ramp) Filter, Ram-Lak, etc.) and back projection.

Filtered inverse projection (FBP) is a well-known technique, for example, "Image Reconstruction: Commentary for Clinicians 2nd-Sequential Approximation," published in August 2014 in Japanese Tomography Research Society Magazine Vol. Image reconstruction method", a more detailed description is omitted here.

2 is a schematic diagram for explaining a method of obtaining field modulation according to the present invention.

In the field modulation CT acquisition system, an X-ray from an X-ray source 11 is irradiated to a subject (e.g., a phantom in the phantom stage of Fig. 2) 20 at various angles to obtain an image. , A truncated projection image 21 and a non-truncated projection image 22 are alternately acquired. That is, as shown in FIG. 2, the cut projection image 21 and the non-cut projection image 22 are alternately obtained, and external information may be obtained differently from the existing ROI CT.

<Field Modulation Acquisition>

The field modulation CT acquisition system of the present invention improves the accuracy of reconstruction of external information by using the POCS algorithm. In order to simulate this, computer tomography is simulated by including engineering software (eg, MATLAB R2017b program) that provides a numerical analysis and programming environment to the operation processing unit 100. In addition, for the simulation, a 3D voxel phantom of 512 × 512 × 512 cubic size uses an abdominal DICOM (Digital Imaging and Communications in Medicine) image (image according to the DICOM format (standard)) obtained from a clinical CT system. And used a virtual CT projector to create a sinogram.

In this simulation, 360 projection data obtained by the second adder 170 (that is, the uncut sinogram (SNT) and the cut sinogram (ST) and the virtual From the result of summation of the sinogram (SOUT)), a CT image was obtained through reverse projection (FBP) reconstruction filtered by the FBP reconstructed CT image acquisition unit 180. ROI CT images were reconstructed using 360 cropped projection data. The CT image obtained through field (beam) modulation acquisition was reconstructed using 288 truncated projection data and 72 maximum size (non-cut) projection data.

3 shows a 3D voxel abdominal DICOM phantom and sinogram obtained by simulation.

Figure 3 (a) is an abdominal DICOM phantom, Figure 3 (b) is a sinogram obtained through a conventional scan, Figure 3 (c) is a sinogram obtained through field (beam) modulation acquisition, 3D is a sinogram obtained through field (beam) modulation acquisition after removing the cut part.

Since most of the 288 cut-out projection data is cut off, from the sinogram obtained by field modulation (Fig. 3(c)), the cut-out part using values of adjacent pixels is removed by extrapolation. (Fig. 3(d)).

4 is an explanatory diagram illustrating a projection data acquisition unit 50 for applying a field modulation method in the field modulation CT acquisition system 10.

4A is an explanatory diagram for explaining acquisition of non-truncated projection data (images). An X-ray is projected from the X-ray tube 11a whose irradiation field is adjusted according to the non-cut projection to obtain a non-cut projected image 22 including the ROI.

4B is an explanatory diagram illustrating acquisition of truncated projection data (image), and from the X-ray tube 11b whose irradiation field is adjusted according to the truncated projection, the X-ray is only the region of interest (ROI). By projecting, an uncut projection image 22 including only the ROI is obtained.

FIG. 4C illustrates acquisition of a truncated projection image and a non-truncated projection image alternately.

By rotating the X-ray tube and adjusting the irradiation field for each angle, that is, the X-ray tube 11a whose irradiation field is adjusted according to the non-cut projection, and the X-ray tube 11b whose irradiation field is adjusted according to the cut projection are alternately positioned, X-rays are projected, and cut projection data and non-cut projection data are acquired alternately.

<How to restore the cut image>

The truncated image restoration procedure is necessary to completely remove truncation artifacts from the CT image obtained by field modulation acquisition. The cut projection data was corrected using the two proposed methods.

The first method is the spline interpolation method. Projection data obtained through field modulation acquisition contains information outside the sampled ROI. We populated 288 truncated projection data via spline interpolation method using information outside of 72 sampled ROIs.

In general, a connection polynomial that applies a low-order polynomial to a subset of data points is called a spline function, and the spline function provides an excellent approximation to the behavior of a function that changes rapidly locally. A line drawn using a mathematical function is a curvature expression, which is not called a line but is called a spline. Lagrange interpolation method and Newton's interpolation method have problems such as vibrating when the number of points is large, and as a method to solve these problems, a method that can be combined by creating a function that is dividedly viewed as a function that can be flexibly connected. There is this. This segmented polynomial is called a spline function, and the problem of obtaining an interpolation function using a spline function is called spline interpolation.

The spline becomes a form in which distortion energy is minimized by fixing a point (node) to pass through. This energy is proportional to the value obtained by integrating the square of the curvature of the spline with respect to the length of the arc, and between nodes is expressed as a third-order polynomial, and in the polynomial of the next section, the first and second derivatives are connected in succession at the nodes. From this, the cubic spline (cubic spline) function is used among the spline functions.

This is, given the node x _i (i=1, 2, ..., n) and its function value f(x _i ), each interval [x _i , x _i+1 ](i=1, 2, ... , n-1) is interpolated with a third-order polynomial. At the end points connecting each section, the value of the function and the first and second derivatives are made the same. And the coefficient of the polynomial is determined from the condition that satisfies the end condition at both ends x ₁ and x _n .

Since the spline interpolation method is known in the Naver Encyclopedia, a more detailed description will be omitted.

The second method uses the POCS algorithm.

5 is a flowchart of an image restoration method using the POCS algorithm in the operation processing unit 100 of the present invention. That is, FIG. 5 shows a schematic diagram of a POCS algorithm and a method of reconstructing an entire image.

In the image acquisition step, the projection data acquisition unit 50 of the field-modulated CT image acquisition system projects (photographs) an X-ray around the subject (that is, at an angle of 360 degrees around the subject) to obtain projection data, At this time, the acquired number of projection data (for example, 350) is acquired as'the total number of projection data' (S110).

Projected images (data) acquired by the projection data acquisition unit 50 include cut and non-cut projection images, and the number of cut and non-cut projection images varies depending on the user's setting when shooting. I can.

In the sinogram acquisition step, the total number of projection data sinograms (that is, 360 sinograms) from the projection data of the total number of projection data (eg 360) received from the projection data acquisition unit 50 in the image acquisition step Obtain (S120).

In the uncut sinogram extraction step, the uncut sinogram (SNT) (for example, 72 uncut sinogram) of the number of uncut projection images is extracted from the sinogram obtained in the sinogram acquisition step (S130). ).

In the image reconstruction step by the POCS algorithm, a reconstructed image is obtained by applying the POCS algorithm to the uncut sinogram (SNT) obtained in the uncut sinogram extraction step (S140).

This is the projection data recovery step by forward projection. The reconstructed image obtained in the image reconstruction step by the POCS algorithm is forward projected, and the total number of projection data (e.g. 360), which is a virtual sinogram (SOUT) containing external information. The projection data of is restored (S150).

In the truncated sinogram extraction step, a truncated sinogram (ST) (eg, 288 truncated sinogram) of the number of truncated projection images is extracted from the sinogram obtained in the sinogram acquisition step (S160).

In the first summing step, the uncut sinogram (ST) extracted in the uncut sinogram extraction step and the cut sinogram (ST) extracted in the cut sinogram extraction step are added (S165).

In the second summing step, the summing result of the uncut sinogram (SNT) and the cut sinogram (ST) output in the first summing step, and the virtual cyno output in the projection data recovery step by forward projection The projection data of the total number of projection data, which is a gram (SOUT), is added (S170).

Filtered Inverse Projection (FBP) Reconstructed CT image acquisition step, uncut sinogram (SNT), cut sinogram (ST), and virtual sinogram (SOUT) output in the second summing step Filtered inverse projection (FBP) reconstruction is performed from the projection data (that is, projection data of the total number of projection data) of the summation result of, to obtain a filtered inverse projection (FBP) reconstructed CT image (S180).

As shown in FIG. 5, the reconstructed image obtained from the projection data consisting of 72 samples using the POCS algorithm is somewhat blurry and the image quality is slightly degraded, but sufficient information to fill the ROI external information of 288 cropped projection data can be provided. . 360 projection data were recovered using reconstructed images obtained with the POCS algorithm using forward projection. The recovered ROI external information of 288 truncated projection data is merged with 288 truncated projection data to fill the external information. The final reconstructed image is obtained using 288 filled projection data and 72 unprojected projection data. The final completed sinogram, 72 uncut projection data obtained by field modulation (SNT), 288 cut projection data obtained by field modulation (ST), and 360 reconstructed non-cutting by forward projection (SOUT) 5 shows that the external information (SOUT) of the sinogram is combined.

<Data Analysis>

6 shows three ROIs (regions of interest) of the abdominal phantom, that is, ROI 1, ROI 2, and ROI 3.

The image quality of the three ROIs (see FIG. 6) was evaluated using the Structural Similarity Index Measure (SSIM) value and the peak signal-to-noise ratio (PSNR). During field modulation acquisition, the inner ROI was set to ROI 1, and the outer ROI was set to ROI 2 and ROI 3. The index SSIM (Structed similarity image matching) value used to evaluate the degree of image restoration is calculated as follows.

Here, α, β, and γ are most commonly set to a set value of 1. ℓ, c, and s are luminance, contrast, and structural elements, respectively, and are defined as follows.

Where u _x and u _y are the local means of the reference image and the comparison image, respectively. σ _x and σ _y are the standard deviations of the reference image and the comparison image, respectively. σ _xy is the cross-covariance of the two images. C1, C2, and C3 are positive integers and are used to stabilize the equation when the denominator is small.

7 is an example of a sinogram obtained from the abdominal DICOM phantom according to the present invention.

7A is a reference sinogram. FIG. 7B shows a sinogram generated by field modulation acquisition using simple extrapolation in order to remove the truncated portion of the sinogram. 7C shows the reconstructed sinogram obtained by the spline interpolation method. 7D shows the reconstructed sinogram obtained by the method proposed in the present invention based on the POCS algorithm. The sinogram restored in (d) of FIG. 7 is the closest to the reference sinogram.

8 is. An example of reconstructed abdominal CT image is shown.

FIG. 8A is a reference CT image and shows a reference CT image of an abdominal DICOM phantom. Figure 8(b) shows the reconstructed image obtained using the ROI scan. In this reconstructed image, you can hardly see the outline of the external information. FIG. 8C shows a sinogram generated by field modulation using simple extrapolation. Fig. 8(d) shows a reconstructed image obtained using field modulation using an interpolation method, and Fig. 8(e) shows a reconstructed image obtained using field modulation acquisition using the method proposed in the present invention. That is, FIGS. 8C to 8E show reconstructed images obtained using field modulation acquisition. 8C and 8D show stick artifacts, even though the outline of the external information is visible. This stick artifact was removed using the method of the present invention based on the POCS algorithm, as shown in Fig. 8(e).

9 is an enlarged view of three ROIs in reconstructed abdominal CT images obtained by each method.

FIG. 9(a) shows the reference images of ROI 1, ROI 2, and ROI 3, and FIG. 9(b) shows the reconstructed images obtained by scanning ROI 1, ROI 2, and ROI 3, and FIG. (c) shows the images obtained by field (beam) modulation using simple extrapolation for ROI 1, ROI 2, and ROI 3, and FIG. 9(d) shows the interpolation method for ROI 1, ROI 2, and ROI 3 An image obtained using field (beam) modulation is shown, and (d) of FIG. 9 shows images obtained through field (beam) modulation of the present invention for ROI 1, ROI 2, and ROI 3.

During the fill (beam) modulation acquisition, the inner ROI was ROI 1, and the outer ROI was ROI 2 and ROI 3. In (b) of FIG. 9, the ROI scan of ROI 1 indicates that the cutting artifacts were not completely removed and slightly brightened. There is little external information visible in the ROI scans of ROI 2 and ROI 3. However, among the information obtained using field (beam) modulation, the interpolation method of Fig. 9(c) and the proposed method of the present invention of Fig. 9(d) have been restored to a reasonable level.

10 shows four ROI-based methods (i.e., a method of obtaining a reconstructed image by scanning an ROI, a method of obtaining a reconstructed image by field (beam) modulation using simple extrapolation for an ROI, and an interpolation method for an ROI. It shows the SSIM value obtained by the method of obtaining an image reconstructed by field (beam) modulation, and a method of obtaining an image reconstructed by field (beam) modulation using the POCS algorithm for ROI of the present invention.

At ROI 1, all ROI-based methods yielded high SSIM values close to 1. Reconstructed images obtained from ROI scans have slightly lower SSIM values due to cutting artifacts. In particular, reconstructed images obtained by field modulation acquisition have higher SSIM values for external information than those obtained by ROI scan. The highest SSIM value was obtained by this method based on the POCS algorithm.

In the present invention, the sinogram indicates that the projection data obtained by beam modulation contains sufficient external information for the ROI. Using this external information, an image similar to the original can be reconstructed by the POCS-based external information restoration method of the present invention. The reconstructed image of the abdominal phantom showed that the image restoration method was the most effective among the interpolation method and the POCS-based method. Truncation artifacts were included in the ROI scan because the area outside the ROI (region of interest) was cut off. Since the projection data obtained by ROI scan is only ROI information, the external information of the reconstructed image is incomplete and there is no information that can be used to fill in the external information.

On the other hand, the projection data obtained by field modulation acquisition by simple extrapolation includes information outside the ROI, which can be reconstructed from the reconstructed image using field modulation acquisition. However, since there is more ROI internal information than external information, an ROI boundary exists in the reconstructed image as shown in FIG. 8 and the image quality is slightly degraded. In order to solve this problem, the present invention proposes to fill in external information using a spline interpolation method and a restoration method based on the POCS algorithm. This method was evaluated by calculating the SSIM value to determine the most effective method. All ROI-based methods had high SSIM values close to 1 in ROI 1. However, the reconstructed image obtained using the ROI scan had a slightly lower SSIM value, so it was rather high due to the pixel value trimming artifact around ROI 1.

How well the external information is restored is determined by checking ROIs 2 and 3 in the reconstructed image. Overall, the SSIM values of the field modulation scheme for ROIs 2 and 3 were high due to the sinogram containing external information. The interpolation method and the POCS-based method fill the sample space between external information outside the ROI, so the SSIM value is increased. However, since the interpolation method used only external information of 72 projection data to fill the value, stick (striped) artifacts occur in the reconstruction process as shown in Figs. 6 and 7. While this could improve the SSIM value to some extent, it was not sufficient to use the image for diagnostic or radiotherapy planning.

If the POCS-based method shown in FIG. 5 is used, a value between external information can be predicted using only external information of 72 projection data. The POCS-based method yielded the highest SSIM value because 72 external information is used to predict and fill 360 external information, which is closer to the actual value.

The sparse sampling collection of scattering data can miss important anatomical information and limit practical clinical use for medical diagnosis. However, since the projection data inside the ROI obtained by acquiring field modulation of the present invention is not sampled information, a perfect anatomical image can be provided and restored to an appropriate level outside the ROI. The field modulation acquisition used in the present invention is expected to reduce the exposure by about 53% compared to the conventional CT image. This is proportional to the size of the field of view set for image acquisition.

The present invention proposes an ROI-based field modulation acquisition method capable of reconstructing external information while reducing the amount of exposure. In addition, the present invention is based on the POCS algorithm and can be used to obtain reconstructed images of the highest quality. In addition, the possibility of using field modulation acquisition using the POCS algorithm of the present invention in CT images was verified.

As described above, although the present invention has been described by the limited embodiments and drawings, the present invention is not limited to the above embodiments, which is various modifications and variations from these descriptions to those of ordinary skill in the field to which the present invention belongs. Transformation is possible. Accordingly, the spirit of the present invention should be grasped only by the claims set forth below, and all equivalent or equivalent modifications thereof will be said to belong to the scope of the present invention.

10: field modulated CT acquisition system 11: X ray source
20: subject 21: cut projection image
22: uncut projection image 50: projection data acquisition unit
100: operation processing unit 120: sinogram acquisition unit
130: uncut sinogram extraction unit 140: POCS algorithm processing unit
150: projection data recovery unit by forward projection 160: cut sinogram extraction unit
165: first adder 170: second adder
180: FBP reconstructed CT image acquisition unit

Claims (15)

The projection data acquisition unit acquires projection data by irradiating X-rays to the subject at an angle of 360 degrees centered on the subject, and by adjusting the field of view, the projected data of the cut and projected image and the projected data of the non-cut projected image Is acquired alternately, and the number of acquired projection data is'total projection data', the number of projection data of the cut-off projection image is'cut-off projection data number, and the number of projection data of the uncut projection image is' A projection data acquisition step of using the number of non-cut projection data;
The operation processing unit includes: a sinogram acquisition step of obtaining a sinogram of the total number of projection data from the projection data of the total number of projection data acquired in the projection data acquisition step;
The calculation processing unit includes a non-cut sinogram extraction step of extracting a non-cut sinogram (SNT) of the number of non-cut projected data from the sinogram obtained in the sinogram acquisition step;
The operation processing unit includes: an image reconstruction step using a POCS algorithm for obtaining a reconstructed image by applying a projection onto convex sets (POCS) algorithm to the uncut sinogram (SNT) obtained in the uncut sinogram extraction step;
The operation processing unit performs forward projection of the reconstructed image obtained in the image reconstruction step by the POCS algorithm using a virtual CT projector, and restores the projection data of the total number of projected data, which is a virtual sinogram (SOUT) including external information. , Projection data recovery step by forward projection;
It characterized in that it comprises a, POCS-based external information restoration method in the field modulation CT acquisition system. The method of claim 1,
A cut sinogram extraction step of extracting a cut sinogram (ST) of the number of cut projection data from the sinogram obtained in the sinogram acquisition step;
A first summing step of summing the uncut sinogram (SNT) extracted in the uncut sinogram extraction step and the cut sinogram (ST) extracted in the cut sinogram extraction step;
It characterized in that it further comprises, POCS-based external information restoration method in the field modulation CT acquisition system. The method of claim 2,
The operation processing unit, on the summation result of the uncut sinogram (SNT) and the cut sinogram (ST) output in the first summing step, the virtual sinogram output in the projection data recovery step by forward projection ( SOUT) to obtain projection data of the total number of projection data;
It characterized in that it further comprises, POCS-based external information restoration method in the field modulation CT acquisition system. The method of claim 1,
The cut-out projection image is an image photographed with only the region of interest as an irradiation field, and the non-cut projection image is an image photographed with the entire view provided by the projection data acquisition unit as an irradiation field. POCS-based external information restoration method in CT acquisition system. The method of claim 3,
Filtered back projection (FBP) on the summation result of the uncut sinogram (SNT), cut sinogram (ST), and virtual sinogram (SOUT) output in the second summation step ) Performing reconstruction to obtain a reconstructed CT image, obtaining an FBP reconstructed CT image;
It characterized in that it further comprises, POCS-based external information restoration method in the field modulation CT acquisition system. The method of claim 5,
Filtered inverse projection (FBP) reconstruction is characterized in that the summation result output in the second summation step is corrected with a correction filter, and back-projection is performed using a virtual CT projector to obtain a reconstructed CT image. A POCS-based external information restoration method in a field modulation CT acquisition system. A recording medium storing a computer program source for a POCS-based external information restoration method in the field modulated CT acquisition system according to any one of claims 1 to 6. The method of claim 1,
The total number of projected data, the number of cut projected data, and the number of non-cut projected data are values set by a user before the projected data acquisition step, wherein the POCS-based external information restoration method in a field modulated CT acquisition system. The method of claim 1,
The total number of projected data is 360, the number of cut projected data is 72, and the number of uncut projected data is 288. POCS-based external information restoration method in a field modulated CT acquisition system. At a 360-degree angle centered on the subject, the subject is irradiated with X-rays to obtain projected data, but by adjusting the field of view, the projected data of the cut-out projected image and the projected data of the uncut-off projected image are alternately acquired. , Projection data acquisition unit;
Obtain a sinogram from the projection data received from the projection data acquisition unit, extract an uncut sinogram (SNT) and a cut sinogram (ST) from the obtained sinogram,
A reconstructed image is obtained by applying a POCS algorithm to an uncut sinogram (SNT), and a virtual sinogram (SOUT) including external information is obtained by forward projecting the obtained reconstructed image,
A CT image by performing filtered back projection (FBP) reconstruction on the projection data obtained by summing the uncut sinogram (SNT), the cut sinogram (ST) and the virtual sinogram (SOUT). An operation processing unit to obtain
Field modulation CT acquisition system comprising a. The method of claim 10,
The projection data acquisition unit,
Projection data is obtained by irradiating X-rays to the subject at a 360-degree angle centered on the subject, and the projected data of the cut-off projection image and the projection data of the non-cut-off projection image are acquired alternately, and the acquired projection data It is characterized in that the number of'total projected data' is the number of projected data, the number of projected data of the cut-off projection image is'cut-off projected data number, and the number of projected data of the uncut projected image is'non-cut projection data number Field Modulated CT Acquisition System. The method of claim 11, wherein the operation processing unit,
A sinogram acquisition unit that obtains a sinogram of the total number of projection data from projection data of the total number of projection data acquired by the projection data acquisition unit;
A non-cut sinogram extraction unit for extracting a non-cut sinogram (SNT) of the number of non-cut projection data from the sinogram obtained by the sinogram acquisition unit;
A POCS algorithm processing unit for obtaining a reconstructed image by applying a POCS algorithm to the non-cutting sinogram (SNT) obtained by the non-cutting sinogram extraction unit;
Projecting the reconstructed image obtained from the POCS algorithm processing unit in the forward direction using a virtual CT projector to obtain projection data of the total number of projection data, which is a virtual sinogram (SOUT) including external information. A projection data recovery unit by forward projection;
It characterized in that it comprises a, field modulation CT acquisition system. The method of claim 12, wherein the operation processing unit,
A cut sinogram extraction unit for extracting a cut sinogram (ST) of the number of cut projection data from the sinogram obtained by the sinogram acquisition unit;
A first adder for summing the uncut sinogram (SNT) extracted from the uncut sinogram extraction unit and the cut sinogram (ST) extracted from the cut sinogram extraction unit;
To the sum result of the uncut sinogram (SNT) and the cut sinogram (ST) output from the first adder, a virtual sinogram (SOUT) output from the projection data recovery unit by forward projection is added, A second adder for obtaining projection data of the total number of projection data;
Filtered back projection (FBP) on the sum result of the uncut sinogram (SNT), cut sinogram (ST), and virtual sinogram (SOUT) output from the second adder An FBP reconstructed CT image acquisition unit that performs reconstruction to obtain a reconstructed CT image;
It characterized in that it further comprises, field modulation CT acquisition system. The method of claim 13,
A field modulated CT acquisition system, characterized in that the total number of projected data, the number of cut projected data, and the number of non-cut projected data are values set by a user before the projected data acquisition step. The method of claim 10,
The cut-out projection image is an image photographed with only the region of interest as an irradiation field, and the non-cut projection image is an image photographed with the entire view provided by the projection data acquisition unit as an irradiation field. CT acquisition system.

Images