KR20160131493A - Method for quality control and accuracy upgrade of patient-specific CT-FFR using computer simulation of contrast media distribution - Google Patents

Abstract

The present invention relates to a new computerized tomography (CT) contrast media distribution simulation for determining whether patient-specific fractional flow reserve (FFR) in coronary arteries has reliability. According to the present invention, when a CT-FFR verification method is used, it can be determined whether a specific CT-FFR result can be used. That is, a screening basis on whether a doctor uses a result of calculation of CT-FFR to a corresponding patient is suggested. Therefore, a medical accident by a doctor or filing of a lawsuit, which is caused by misdiagnosis due to inaccuracy in a calculation result, can be prevented. As a result, the usefulness and validity of a CT-FFR technique is increased, thereby allowing the CT-FFR technique to be used in a further wider range.

Description

[0001] The present invention relates to a computer-simulated CT-FFR, and more particularly, to a CT-FFR,

The present invention relates to a novel CT contrast agent distribution simulation for determining the reliability of a fractional flow reserve (FFR) for each patient.

In general, myocardial perfusion reserve is widely used as a clinical index to assess the degree of functioning of coronary artery stenosis. This is a non-invasive method of calculating the FFR of individual patients by creating a shape model based on the CT data of the patient, performing computations using a numerical analysis method and a computer (Fig. 1). Contrast agent simulation is a technique for predicting the spatial distribution characteristics of contrast agents in coronary arteries through computer and numerical methods.

Patent Application No. 10-2014-0114553, entitled " New Patient Customized Model for Calculation of Myocardial Blood Flow Reserve in Coronary Arteries "

 Miyoshi T, Osawa K, Ito H, Kanazawa S, Kimura T, Shiomi H, Kuribayashi S, Jinzaki M, Kawamura A, Bezerra H, Achenbach S, Nrgaard BL. Non-invasive computed fractional flow reserve (CT) for diagnosis of coronary artery disease. 2015, Circ J. 79 (2): 406-12.  Choi JH, Min JK, Labounty TM, Lin FY, Mendoza DD, Shin DH, Ariaratnam NS, Koduru S, Granada JF, Gerber TC, Oh JK, Gwon HC, Choe YH. Intracoronary transluminal attenuation gradient in coronary angiography for determining coronary artery stenosis. JACC Cardiovasc Imaging. 2011 Nov; 4 (11): 1149-57.  Kim, H. J., Vignon-Clementel, I. E., Coogan, J. S., Figueroa, C. A., Jansen, K. E., Taylor, C. A., 2010. Patient-specific modeling of blood flow and pressure in human coronary arteries. Ann. Biomed. Eng. 38 (10), pp. 3195-3209.  K. T. Bae, J. R Heiken, and J. A. Brink. 1998. Aortic and Hepatic Contrast Medium Enhancement at CT; Part I. Prediction with a Computer Model, Radiology, v. 207, No. 3, pp. 647-655.

The present invention relates to a method for calculating a basis for prediction accuracy of a CT-FFR for each patient and determining from the result that the simulation result is reliable enough to be applicable to a patient. In addition, when reliability of the CT-FFR results is low and it is difficult to apply to the patient, a repetitive simulation method for improving the accuracy is also proposed. Currently, the output process of CT-FFR is performed through the simulation process after extracting coronary artery blood vessel shape through segmentation of CT image. However, if the quality of the CT image is poor or excessive calcium deposition is present here, coronary artery blood vessel morphology can not be correctly extracted, resulting in inaccurate CT-FFR results. However, there is no way to confirm whether the simulation process is fully reflected in the segmentation process. Although there is a TAG (transluminal attenuation gradient) method that utilizes the gradient of the brightness of the image for difficult-to-identify blood vessel regions, this only serves to roughly predict the degree of clogging of the blood vessel, and CT- It is difficult to utilize.

After computation of CT-FFR, there is also a method of verifying by invasive 2-dimensional X-ray angiography image. However, the introduction of additional invasive validation procedures undermines the CT-FFR's primary advantage of non-invasively predicting FFR. This disadvantage is an important factor that makes the physician reluctant to apply the patient's CT-FFR results directly to the patient. In other words, there is no objective and scientific basis for judging whether the quality of CT-FFR results is reliable. Because of this, the accuracy of the CT-FFR measurement is up to 80% (HeartFlow, Kyoto Miyoshi, 2015, Japan, 2015), but the CT-based FFR calculations alone Clinical applications have not yet been reported. Therefore, if there is a scientific basis for the validity of the calculated CT-FFR results, it would be a powerful method to ensure the reliability of CT-FFR (Fig. 2).

The present invention focuses on the fact that the concentration of the contrast agent administered during the CT imaging of the blood vessel is correlated with the brightness of the CT image. Therefore, the distribution of the contrast agent is simulated for the three-dimensional shape used in the CT-FFR calculation, and this is compared with the brightness distribution of the CT image. If the two distributions are similar in character, the image segmentation is correct and accurate CT-FFR is derived. In other words, through the contrast agent simulation for such CT images, validity of the segmentation data of the blood vessel images used to calculate the CT-FFR is confirmed.

When the CT-FFR verification method according to the present invention is used, it is possible to determine whether or not a specific CT-FFR result can be used. In other words, the physician provides a screening basis for whether to use the results for the patient who computed the CT-FFR. Therefore, it can prevent a medical accident or a lawsuit from being caused due to an error due to the inaccuracy of a calculation result by a doctor. This will ultimately increase the utility and validity of the CT-FFR technology and make it more widely available in clinical practice.

FIG. 1 is a conceptual diagram of a prediction method by simulation of current FFR. Here, a three-dimensional vessel shape model of a patient that can have a decisive influence on the simulation accuracy is created in a segmentation process of extracting a coronary artery shape of a patient from CT data .
FIG. 2 is a problem of the present FFR simulation. In the process of extracting the three-dimensional shape of the coronary artery from the CT image, an error may occur. This may lead to inaccurate calculations, so there is a need for judgment or screening methods to exclude or recalculate them.
FIG. 3 is a flowchart illustrating a procedure of a quality control method using a molding agent simulation according to the present invention.
FIG. 4 is a conceptual diagram of administration of a contrast agent during CT imaging. When a CT image is actually taken, the contrast agent is injected through the arm vein as shown in the figure for high contrast of the blood vessel. Because the contrast agent is expressed at a higher intensity than the blood, it allows a clearer identification of the shape of the blood vessel to be diagnosed. However, when the contrast medium flows with blood, the distribution of the contrast agent may be disturbed when the blood flow is changed in the stenotic region, and the concentration may be different from the existing blood vessel because the concentration is decreased toward the back of the blood vessel.
FIG. 5 is a graph showing the gradient distribution of the CT brightness according to the stenosis and the distribution of the contrast agent according to the blood flow in the blood vessel. Here, the gradient of the CT brightness is quantified as TAG (transluminal attenuation gradient) Therefore, it has a gradually steep gradient.
FIG. 6 shows the simulation result of the difference of the contrast agent distribution according to the stenosis in the coronary artery blood vessel. As the stenosis becomes more severe in the same model, the contrast agent concentration difference before and after stenosis becomes more severe.
FIG. 7 is a schematic diagram of the entire human body system for the distribution simulation of the contrast agent presented by Bae et al., Wherein the round ellipsis indicates the vascular system and the numbers therein indicate the volume of the blood vessel. And the number on the line between the ellipse and the ellipse represents the blood flow (unit: ml) between the vascular systems. And the number of squares means the volume of the tissue (unit: ml). In this patent, the coronary artery is replaced with the CFD model shown on the right.

Other objects, specific advantages and novel features of the present invention will become more apparent from the following detailed description and preferred embodiments with reference to the accompanying drawings.

Hereinafter, the CT-FFR quality control and the accuracy improvement based on the new contrast agent simulation according to the present invention will be described in detail with reference to the accompanying drawings.

Pijls et al. Introduced the fractional blood flow reserve (FFR), which is the pressure ratio before and after stenosis of fully expanded microvascular vessels as a clinical indicator of coronary artery disease. The Pijls group used the guidewire technique to measure FFR, which became a technique for evaluating the degree of coronary stenosis.

Kim et al. Proposed a non-invasive simulation method using CT images and patient information in evaluating FFR values. This is an integrated method of computational fluid dynamics techniques for the hemodynamic computation of aortic and coronary vessels and a centralized parameter model of the whole cardiovascular system. The validity and usefulness of this simulation model has been verified through several clinical studies. However, no method has been proposed to confirm or verify the validity of the results of this technique.

Shim et al. Developed a simplified method for calculating FFR as compared to Kim et al. (Fig. 1). This method is simple and efficient because it models only the microvascular vessels of the coronary arteries and associated coronary circulation systems except for the aorta, which reduces the number of calculation grids compared to Kim et al. And does not include the cardiac and somatic models connected to the aorta . However, this technique did not provide a way to verify the quality of the results.

In this technique, we proposed a method based on contrast agent simulation to confirm coronary artery hemodynamics and accuracy of CT-FFR calculations. In this method, two prior art simulations are first required and the distribution of the contrast agent is predicted based on this. First, the temporal distribution of the contrast agent in the aortic and cardiac peripheral blood vessels is predicted using the central parameter method of the patient-specific contrast agent. Second, the temporal change of the blood flow velocity is obtained in the same way as the conventional FFR calculation for the three-dimensional shape obtained by segmenting the CT image of the actual patient. After obtaining these two conditions, the partial differential equation is solved in the 3D shape obtained from the CT image to obtain the contrast agent distribution. And compares it with the HU distribution of the CT image to evaluate the accuracy and validity of the calculation.

This method is performed through a procedure as shown in FIG. First, calculate the FFR by extracting the patient's 3-dimensional coronary artery from the CT image. The velocity distribution obtained from this calculation is used for the simulation of the contrast agent to obtain the concentration distribution in the 3D coronary artery shape of the contrast agent. Then, it is determined whether or not the pattern of the concentration distribution matches the brightness pattern in the CT image to determine whether the pattern is accurate or not. Based on this, we decide whether to use FFR calculation results in diagnosis or not. In the case of exclusion, there are two ways to recompute by re-segmentation and give up the case.

The first procedure among the procedures shown in FIG. 3 is to extract the shape of the coronary artery from the CT image of the patient. Here we use the well-known active contour method to obtain the surface of the coronary artery lumen. And the entire volume of this vessel lumen is filled with tetrahedral grid generation. Then, a hemodynamic simulation is performed in which a computational fluid dynamics (CFD) as described by Shim et al. And a lumped parameter model (LPM) for peripheral blood vessels are combined. At this time, the flow condition is made for the normal state, not the hyperemia as in the FFR calculation. Here, flow conditions such as speed with time are calculated. These values are then used in the contrast agent simulation calculation. The simulation of the contrast distribution is an analysis of the following equation, which is solved using the finite element method.

In equation (1), the first term on the left is the time variation, the second term is the convection term, and the right term is the variable representing the concentration of the contrast agent, t is the time, x, y, z is the cartesian coordinate , U, v, and w denote the velocities in the x, y, and z directions, respectively. D represents the diffusion coefficient of the contrast agent. The Euler method is applied to the time derivative of the above equation, and the following equation is obtained using the previously calculated velocity.

In the above equation (2), superscripts n + 1 and n represent a time step index, and t represents an interval between time steps. Using the weighted residual method and the Galerkin method to derive the finite element equations for Equation (2), the following matrix equation is derived and the time / spatial distribution of the contrast agent can be determined by solving the equation for each time.

In equation (3) above, the subscript notation follows the normal tensor format. In Equation (3), A _ij is a coefficient matrix and F _i is a forcing vector, all of which are expressed as a function that varies according to the value of the previous time step c _i .

In the above finite element equations, N _i denotes the interpolation function at the ith node.

4, the intensity of a CT image may vary depending on the distribution of a contrast agent during a CT scan. In fact, when performing a CT scan, the contrast agent is injected through the arm vein as shown in the figure for a high contrast of the blood vessel. Because the contrast agent is expressed at a higher intensity than the blood, it allows a clearer identification of the shape of the blood vessel to be diagnosed. However, because the contrast medium flows with the blood, the distribution of the contrast agent may decrease when the blood flow rate decreases due to the stenosis site, and the concentration of the contrast agent becomes thinner toward the back of the blood vessel, (Figure 5).

In fact, according to a study by Choi et al., The greater the stenosis, the lower the concentration of contrast agent after stenosis, which increases the TAG (transluminal attenuation gradient) value that quantifies the gradient of CT brightness. Choi et al. Also proposed a technique to infer the degree of stenosis from the TAG value of CT. Fig. 6 is an example of calculating the contrast agent distribution by the finite element method. In this example, the outlet is interpreted in a very simple way with a certain value. However, in order to calculate the actual contrast distribution, it should be combined with a lumped parameter model (LPM) for contrast concentration of the whole body. That is, as suggested by Kim et al. And Shim et al., For the FFR simulation, the CFD model for the 3-dimensional coronary artery and the LPM model for the cardiac peripheral blood vessel were required. Similarly, for an accurate interpretation of the contrast agent distribution, the three-dimensional model for the contrast agent and the LPM model for the contrast agent distribution in the peripheral blood vessel should be combined.

In order to solve the finite element of the contrast agent distribution expressed in Equation (3), a boundary condition is required at the entrance and exit of the 3-dimensional coronary artery. This requires an LPM model of contrast distribution for the whole body. This is based on the method presented by Bae et al. This model can describe the contrast agent behavior in the entire circulatory system according to individual characteristics (height, weight) and is a single & two compartment model. Here, parameter tuning such as permeability factor (PS) and conversion factor for enhancement (factor that converts the concentration of the contrast agent to HU, which is the unit of CT brightness) is needed to describe the existing experimental data well. Do. If the parameter tuning is successfully performed, the coronary is modeled according to the actual vessel structure of the patient so that the analysis of the regional flow of the coronary can be performed, and then the outlet boundary condition required for the equation (3) is obtained.

FIG. 7 is a schematic diagram of an LPM model of contrast agent distribution in the human whole body as presented by Bae et al. Here, the round ellipse refers to the blood vessel system, and the numbers in it indicate the volume of the blood vessel. And the number on the line between the ellipse and the ellipse represents the blood flow (unit: ml) between the vascular systems. And the number of squares means the volume of the tissue (unit: ml). In this patent, the coronary artery is replaced with the CFD model shown on the right. That is, the LPM model shown in FIG. 7 for simulating the 3D contrast agent model and the exit boundary condition thereof is combined and analyzed in a manner similar to the FFR simulation. In other words, the contrast agent is obtained through this contrast agent + LPM linkage simulation, and it is converted into the CT brightness (HU: Hounsfield unit) in the coronary artery branch and compared with the brightness of the actual CT image. Through this, it is judged whether or not the segmented image properly reflects the actual shape. If it is judged to be different from the actual one, it is recalculated again or processed as unreliable data and excluded.

The embodiments of the present invention described above and shown in the drawings should not be construed as limiting the technical idea of the present invention. The scope of protection of the present invention is limited only by the matters described in the claims, and those skilled in the art will be able to modify the technical idea of the present invention in various forms. Accordingly, such improvements and modifications will fall within the scope of the present invention as long as they are obvious to those skilled in the art.

Claims (2)

Inputting image data including a plurality of coronary artery images of a patient to a computer system,
Processing the input image data using the computer system to generate a three-dimensional model representing an anatomical structure including at least a portion of the plurality of coronary arteries;
Performing hemodynamic simulation on the anatomical structure of the three-dimensional model representing the generated anatomical structure using the computer system to obtain blood flow information in the three-dimensional model;
Obtaining a virtual agent distribution of a contrast agent for a three-dimensional model based on the blood flow information using the computer system;
Comparing the actual contrast agent distribution of the contrast agent virtual distribution and the coronary artery image data for the three-dimensional model using the computer system;
Using the computer system to determine whether the difference between the virtual distribution of the contrast agent and the actual distribution of the contrast agent is within a predetermined range. The method according to claim 1,
Wherein the step of comparing the virtual distribution of the contrast agent with the actual distribution of the contrast agent comprises the step of converting the virtual contrast agent distribution to the CT brightness and wherein the comparison of the virtual contrast agent distribution and the actual contrast agent distribution comprises comparing the CT brightness corresponding to each distribution, Three dimensional modeling verification method of arteries.

Images