JP2011224275A - Image processing apparatus, method and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image processing method, apparatus and program for improving accuracy of extracting a specific organ by automatically or semiautomatically setting a proper ROI when automatically extracting the specific organ from contrast medical image volume data.SOLUTION: The image processing apparatus estimates a region of a specific organ on the basis of a three-dimensional position and size of a blood vessel supplying blood to an organ, as an ROI, to extract the specific organ using the ROI.

Description

The present invention relates to an image processing method, an image processing apparatus, and a program for improving extraction accuracy when a specific region is extracted from a three-dimensional medical image.

In diagnosis using a medical image inspection apparatus represented by an X-ray CT (X-ray Computed Tomography) apparatus or an MRI (Magnetic Resonance Imaging) apparatus, a three-dimensional medical image (hereinafter referred to as `` medical image volume data '') , Also referred to as “volume data”), a visualization process may be performed using a plurality of medical image processing algorithms to assist diagnosis. The medical image processing algorithm used here has a plurality of methods depending on the case, such as extracting and displaying only a necessary tissue from input volume data, or displaying an image with an emphasized edge.

  Further, in recent years, medical image capturing apparatuses have accelerated the number of tomographic images (hereinafter also referred to as “slice images”) that can be captured at a time due to technological advances. The amount is also huge. When displaying an image by automatically applying a medical image processing algorithm, it is very difficult to apply all the processing automatically in consideration of safety and accuracy. Checks and corrections by engineers and engineers are essential, but the more manual corrections, the higher the burden and human costs on doctors and engineers. If the volume of volume data is enormous, these costs are further increased. It will be expensive. In order to reduce this burden, it is required to increase the accuracy and validity of the part of the processing that is automatically performed as much as possible.

  Here, attention is paid to a technique for extracting a specific organ from a contrast-enhanced medical image in medical image processing. A contrast agent flows into each organ or diseased part according to the flow of blood, changes the value of voxels in those areas, thereby improving the visibility of medical images and facilitating diagnosis using images. is there. Contrast-enhanced medical images not only improve the visibility of blood vessels but also organs and diseased sites because of the nature of contrast medium flowing along the bloodstream. Patent Document 1 is a technique that utilizes the high visibility of blood vessels, and in automatic liver region recognition, after extracting blood vessels of the entire liver, the blood vessel skeleton is extracted from the blood vessel skeleton in the extracted liver region. It is a technique that is used for surgical simulation by segmenting the region where blood vessels are controlled.

JP 2004-337257 (Japanese Patent Application No. 2003-135122)

On the other hand, when an organ itself is extracted, it is often difficult to separate the organ from its surrounding tissues and blood vessels, depending on the degree of contrast and the characteristics of the patient, even if it has already been contrasted. Therefore, even if a specific organ is automatically extracted after contrast enhancement, it is extracted not only in the organ itself but also in the surrounding surrounding tissue, and its volume is too large or not. In some cases, the volume of the extraction portion may be too small due to the occurrence of an extraction area. With regard to Patent Document 1, it is necessary to extract the entire liver before separating the liver section. However, if this extraction result is too large or too small than the actual result, manual correction or the like is necessary. End up. Therefore, the accuracy of extraction for extracting the organ itself needs to be as high as possible. The present invention solves the above-described conventional problems, and provides an image processing method, an image processing apparatus, and a program for improving extraction accuracy in extraction processing when a specific region is extracted from a three-dimensional medical image. There is to do.

In order to solve the above problems, the first aspect of the present invention provides an automatic area extraction unit (21) for automatically extracting a specific area from medical image volume data to be processed, and information on the automatically extracted area. And an image processing algorithm execution unit (22) for displaying the original volume data on a display device, and an image processing device (11) for performing image processing on the medical image volume data using a predetermined image processing algorithm. The automatic region extraction unit (21) is identified from the blood vessel extraction processing unit (30) that performs extraction processing of contrasted blood vessels such as arteries, veins, and portal veins, and the position and size of the extracted blood vessels. Region of Interest when performing extraction processing on the organ of the subject, that is, ROI determination unit (31) for determining a range or region of interest (hereinafter referred to as ROI) to which the extracted image processing is applied And an organ extraction process execution unit (32) that executes an organ extraction process using an organ extraction algorithm corresponding to each organ using the determined ROI.

  In order to solve the above problems, the first aspect of the present invention provides an automatic region extraction step for automatically extracting a specific region from medical image volume data to be processed, information on the automatically extracted region and the original information. An image processing algorithm execution step of displaying volume data on a display device, and performing image processing on the medical image volume data using a predetermined image processing algorithm, wherein the automatic region extraction step includes: ROI determination for determining the ROI when performing extraction processing for a specific organ from the blood vessel extraction processing step for performing extraction processing of contrasted blood vessels such as arteries, veins, and portal veins, and the position and size of the extracted blood vessels Using the determined ROI and the organ extraction process that executes the organ extraction process using the organ extraction algorithm corresponding to each organ A step, and further comprising a.

According to the present invention, it is possible to provide an image processing method, an image processing apparatus, and a program that improve the extraction accuracy in extraction processing when a specific region is extracted from a three-dimensional medical image.

1 is a system configuration diagram illustrating a configuration of an image processing apparatus as an example of an embodiment. It is a flowchart which shows an example of the automatic area | region extraction process in the automatic area | region extraction part. It is the figure which represented the extracted blood vessel (upper figure) and ROI (lower figure) determined from it in 1 slice in volume data in a simulated manner. It is the figure which represented the extracted coronary artery (upper figure), and the three-dimensional ROI (lower figure) determined from it in a simulated manner. It is a figure which compares the three-dimensional shape of a coronary artery (upper figure) and a portal vein (lower figure) in simulation. It is an example of the extraction likelihood estimation table which compares the characteristic of a general shape with a coronary artery, the extracted blood vessel, the determined ROI, and the shape of the extracted specific organ, and estimates extraction likelihood.

Hereinafter, an image processing apparatus, an image processing method, and a program according to an embodiment of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a system configuration diagram illustrating a configuration of an image processing apparatus as an example of the present embodiment.
As shown in FIG. 1, an image processing device 11 that performs image processing on volume data captured by X-ray CT, MRI, or the like, an external storage device 10 that stores volume data, and instructions for starting each processing are input. An input device 12 for displaying the image, and a display device 13 for displaying the image on which the image processing has been performed.
The image processing apparatus 11 includes an internal memory 20, an automatic area extraction unit 21, and an image processing algorithm execution unit 22. The automatic region extraction unit 21 executes a blood vessel extraction processing unit 30 that executes a contrast-enhanced blood vessel extraction process, a ROI determination unit 31 that determines an ROI for performing a specific organ extraction process, and an extraction algorithm for the organ. An organ extraction process execution unit 32. The automatic area extraction unit 21 and the image processing algorithm execution unit 22 are realized by program processing or a dedicated circuit by a CPU (Central Processing Unit) provided in the image processing apparatus 11. Furthermore, a storage unit (not shown) for storing programs for realizing the functions of the internal memory 20 and the image processing device 21 includes a RAM (Random Access Memory), a ROM (Read Only Memory), a HDD (Hard Disk Drive), It is composed of a storage medium such as a flash memory.

The process until the automatic area extraction result is displayed on the display device 13 will be described.
First, in response to an input from the input device 12, volume data stored in the external storage device 10 is transferred to the internal memory 20. Next, the blood vessel extraction processing unit 30 performs blood vessel extraction from the volume data stored in the internal memory 20 in accordance with the instruction for starting the automatic region extraction processing from the input device 12 and the instruction related to the blood vessel extraction processing. Information such as a three-dimensional position and size is transferred to the ROI determination unit 31. The ROI determination unit 31 determines an ROI to be used for extraction processing of a specific organ from the extracted three-dimensional position and size of the blood vessel, and the three-dimensional position and size of the ROI are extracted from the organ. Transfer to the execution unit 32. The organ extraction processing execution unit 32 performs automatic organ extraction processing on the volume data, then transfers the automatic region extraction processing result to the image processing algorithm execution unit 22, and uses the display device 13 to display the ROI and the image processing result. Is displayed.

  FIG. 2 is a flowchart showing an example of automatic organ extraction processing in the automatic region extraction unit 11. First, the volume data stored in the external storage device is input to the internal memory 20 in the image processing device 11 (step S201). Then, the blood vessel extraction processing unit 30 performs blood vessel extraction processing on the input volume data (step S202). The extracted information on the three-dimensional position and size of the blood vessel is transferred to the ROI determination unit 31, and the ROI determination unit 31 uses the extracted information on the blood vessel to perform processing for extracting a specific organ. A determination is made (step S203). In the organ extraction process 32, an organ extraction algorithm corresponding to a specific organ is executed using the determined ROI (step S204), the region extraction result is output (step S205), and the process ends (S206).

  Step S202 performed in the blood vessel extraction processing unit 30, that is, the blood vessel extraction processing is executed automatically, semi-automatically or manually. When performing the extraction process automatically and semi-automatically, for example, the RegionGrowing method can be used. In Region Growing, one or several points called seeds are set automatically or manually, the pixel values of neighboring pixels are referenced from the seed, and if the pixel value meets the set conditions, the area is expanded to that pixel. It is a way to go.

  Next, an example of ROI determination processing performed by the ROI determination unit 31 will be specifically described with reference to the drawings. FIG. 3 schematically shows one slice in the volume data. It is assumed that the hatching of the vertical line is a part of the blood vessel extracted by the blood vessel extraction unit 30, the periphery thereof is a specific organ, and the line surrounding it is the boundary of the specific organ. The purpose of the ROI determination processing unit 31 is to estimate the range in which the extracted blood vessel supplies blood, assume that it is similar to the existence range of the specific organ, and to include the ROI that includes it. Is to set. Therefore, for example, Dilation processing (processing that expands the region by expanding several pixels from the point that forms the contour of the region to the outside of the region) is performed on the extracted blood vessel region, The region is expanded to a region slightly wider than the region where blood is supplied, and the region is set as the ROI when extracting a specific organ. In the lower diagram of FIG. 3, the range of the right-slanting diagonal line expanded from the blood vessel is ROI. FIG. 3 shows an example in which dilation is performed in units of slices. In addition to this, there is a method in which region expansion is performed two-dimensionally within a plane by setting a plane perpendicular to the blood vessel traveling direction. In addition, it may be performed three-dimensionally for volume data. FIG. 4 shows an example in which the ROI for extracting the atrial ventricle is determined from the coronary artery. The upper diagram of FIG. 4 represents a coronary artery from which a thick line has been extracted, the surrounding area is the atrial ventricle, and the solid line surrounding it represents the boundary of the atrial ventricle. The result of determining the ROI by three-dimensionally expanding the coronary artery is the lower diagram of FIG.

  The organ extraction algorithm executed in step S204 uses, for example, an extraction algorithm that uses the difference between the luminance of the pixel of the organ to be extracted and the luminance of the pixels of the other region for the region inside the ROI. Is effective.

  One example of a simple organ extraction algorithm is threshold processing. When extraction is performed by threshold processing, a set of pixels having values within two thresholds within the ROI is set as an extraction result region. In this case, the threshold value can be set in accordance with, for example, a histogram in the ROI and the shape of the histogram.

  As another extraction algorithm, it is also useful to apply a region extraction method such as Water Shed or Region Growing.

  Water Shed is one of area division methods, and is a technique for dividing an area using a luminance gradient of an image. As a result of area division using Water Shed, overdivision is often a problem, but here, the number of organ areas in the three-dimensional ROI is small, and the volume ratio in the ROI is very high. Therefore, setting of conditions becomes easy.

  When using Region Growing, the conditions set here are generally the difference in luminance between the target pixel and the extracted adjacent pixels, and the range in which the luminance difference between the target pixel and the seed pixel does not exceed a certain threshold. Often to do. Here, for example, by using the blood vessel extracted in step S202 as a seed and setting parameters used for Growing by using the average value and variance of the pixels in the ROI, it is possible to extract a region compared to the case where no ROI is set. It is conceivable that the accuracy of this will improve. In determining the ROI, not only the dilation is simply performed, but also, for example, the existence probability of the organ is determined according to the distance from the blood vessel, and the ROI with the probability distribution can be set. In general, it can be said that the farther from the blood vessel, the lower the probability. However, for example, even if the distance from one blood vessel is the same, if the distance is within the same distance from a plurality of blood vessels, the probability is considered to increase. Specifically, for example, in the heart region, as shown in FIG. 4, the coronary artery is structured to cover the atrial ventricular region. In this case, it can be said that the portion near the center or center of gravity of the entire ROI has a higher probability of being an atrial ventricular region than the portion near the contour of the ROI. In this case, the number of coronary arteries existing at the same distance from the pixels near the center of the ROI is equal to the number of coronary arteries existing at the same distance from the pixels near the ROI contour. It can be thought that there are many compared.

  For example, in addition to the condition based on the luminance as described above, the condition that the region growing region is easily expanded to a pixel having a high organ existence probability and difficult to expand to a pixel having a low probability. In addition, the extraction accuracy can be expected to improve. When the ROI with probability distribution set in this way is used in the organ extraction processing execution unit, extraction with high accuracy is possible by changing parameters of the extraction algorithm according to the probability distribution.

  Here, the extraction accuracy of the blood vessel extraction processing unit 30 can be presented as the extraction likelihood simultaneously with the display of the extraction result of the specific organ. FIG. 5 is a comparison of the shape of the coronary artery and the portal vein. As shown here, the coronary artery is elongated in the body axis direction, and the thickest blood vessel is at the upper part of the entire blood vessel region. There are features. On the other hand, it can be said that the entire shape of the portal vein extends perpendicular to the body axis direction, and the thickest blood vessel is located below the entire blood vessel region. FIG. 6 shows an example of the table showing these characteristics. Judgment conditions are set using the maximum vessel diameter, maximum blood vessel position, ROI major axis direction, ROI major axis / minor axis ratio, ROI volume, and the like, and the likelihood score is calculated based on whether these conditions are met. The ROI determination unit transfers these total values to the image processing algorithm execution unit 22, and the image processing algorithm execution unit 22 presents the extraction likelihood simultaneously with the extraction result of the specific organ as the extraction likelihood. In addition, for example, a means such as an alphabet or a message can be used.

  In addition, it is considered that the shape of a blood vessel formed by such an organ is used when ROI is determined. For example, as shown in FIG. 5, a blood vessel runs around the heart to surround it, and in the case of the liver, a portal vein exists so as to stretch the entire liver region. Using such characteristics, the ROI determination algorithm can be changed for each organ. As described above, the image processing method, the image processing apparatus, and the program for reducing the manual processing part by acquiring the time information from the medical image volume data in the extraction process when extracting a specific region from the three-dimensional medical image. Can be provided.

DESCRIPTION OF SYMBOLS 10 External storage device 11 Image processing apparatus 12 Input apparatus 13 Display apparatus 20 Internal memory 21 Automatic area extraction part 22 Image processing algorithm execution part 30 Blood vessel extraction process part 31 ROI determination part 32 Organ extraction process execution part

Claims (10)

An image processing apparatus that applies a specific image processing algorithm to medical image volume data,
An automatic region extraction step;
An image processing algorithm execution step,
The automatic region extraction step includes:
A blood vessel extraction processing step in which the computer performs blood vessel region extraction processing from the input volume data;
ROI determination step for determining ROI for extraction of a specific organ from the three-dimensional position and size of the blood vessel extracted in the blood vessel extraction processing step;
An image processing apparatus comprising: an organ extraction process execution step for executing an organ extraction algorithm corresponding to a specific organ using the ROI determined in the ROI determination step. The image processing apparatus according to claim 1,
The ROI determination unit
An image processing apparatus further comprising a ROI determination algorithm changing step for changing a ROI determination algorithm according to an organ to be extracted. The image processing apparatus according to claim 1, wherein:
The ROI determination step further includes a ROI determination step with existence probability for determining an ROI to which an existence probability of an organ is added according to a two-dimensional or three-dimensional distance from the blood vessel extracted in the blood vessel extraction processing step. A featured image processing apparatus. The image processing apparatus according to claim 3,
The organ extraction process execution step includes:
An image processing apparatus, further comprising: an extraction algorithm correction step that changes parameters of the extraction algorithm in accordance with the ROI with presence probability determined in the ROI determination step with presence probability. The image processing apparatus according to claim 1, wherein:
The ROI determination step further includes an extraction likelihood estimation step of comparing the extracted blood vessel and the determined three-dimensional feature of the ROI with the general three-dimensional features of blood vessels and organs and estimating the extraction likelihood from the matching degree. When,
The image processing algorithm execution step further includes:
An image processing apparatus comprising: an extraction likelihood fixed time step that sets the extraction likelihood estimated in the extraction likelihood estimation step simultaneously with a specific organ extraction processing result. An image processing method for applying a specific image processing algorithm to medical image volume data,
An automatic region extraction step;
An image processing algorithm execution step,
The automatic region extraction step includes:
A blood vessel extraction processing step in which the computer performs blood vessel region extraction processing from the input volume data;
ROI determination step for determining ROI for extraction of a specific organ from the three-dimensional position and size of the blood vessel extracted in the blood vessel extraction processing step;
An image processing method comprising: an organ extraction process execution step of executing an organ extraction algorithm corresponding to a specific organ with respect to a specific organ using the ROI determined in the ROI determination step. The image processing method according to claim 6,
The ROI determination unit
An image processing method further comprising a ROI determination algorithm changing step of changing a ROI determination algorithm according to an organ to be extracted. The image processing method according to claim 6 or 7,
The ROI determination step further includes a ROI determination step with existence probability for determining an ROI to which an existence probability of an organ is added according to a two-dimensional or three-dimensional distance from the blood vessel extracted in the blood vessel extraction processing step. A featured image processing method. The image processing method according to claim 8, comprising:
The organ extraction process execution step includes:
An image processing method, further comprising an extraction algorithm correction step for changing parameters of the extraction algorithm in accordance with the ROI with presence probability determined in the ROI determination step with presence probability. The image processing method according to claim 6, wherein:
The ROI determination step further includes an extraction likelihood estimation step of comparing the extracted blood vessel and the determined three-dimensional feature of the ROI with the general three-dimensional features of blood vessels and organs and estimating the extraction likelihood from the matching degree. When,
The image processing algorithm execution step further includes:
An image processing method, comprising: an extraction likelihood fixed time step that sets the extraction likelihood estimated in the extraction likelihood estimation step simultaneously with a specific organ extraction processing result.

Images