JP5066386B2 - Image processing apparatus, method, and program thereof - Google Patents

Description

  The present invention relates to an image generation apparatus, method, and program for performing image processing on three-dimensional volume data.

  2. Description of the Related Art In recent years, three-dimensional image display devices that generate and visualize an image obtained by projecting three-dimensional volume data have become widespread with the improvement of computer processing capability. As a technique for visualizing and displaying three-dimensional volume data, volume rendering is well known. In this volume rendering, there is a method called a ray casting method. In this method, an image is generated based on a projection plane on the observation side.

  In the ray casting, as shown in FIG. 9, a virtual projection plane P is arranged in a three-dimensional space, and a virtual ray called a ray l is emitted from each pixel of the projection plane P. This is a method of generating a projection image that sees through the three-dimensional structure inside the volume data 100 on the projection plane P by forming an image of virtual reflected light from the image. Specifically, simulation is performed on light that is irradiated from the projection plane P and reflected, attenuated, or absorbed by the volume data 100 expressed by voxel values. Using such volume rendering, an object structure can be drawn from volume data. In the case of volume data 100 representing the structure of a human body in which tissues such as bones and internal organs are complicated and complicated, these tissues are separated by adjusting the transmittance (adjusting the opacity). Can be drawn. That is, it is possible to generate a projection image that can increase the opacity of each voxel that constitutes the portion of the portion to be seen through, while reducing the opacity of the portion that is not seen through and can observe only the portion to be observed. For example, if the opacity of the epidermis is set low, each organ such as blood vessels, bones and internal organs can be observed through.

  However, such 3D volume data has an enormous amount of data, and the 3D image display apparatus has problems in the capacity of the magnetic disk, the transfer time from the magnetic disk to the memory, the required amount of memory, and the like. Therefore, by compressing the image data and holding it as compressed image data, and creating the three-dimensional image directly from the compressed image data while partially decompressing the voxel data specified by the three-dimensional image processing sequentially, the above-mentioned There has been proposed a method for dealing with the problem (for example, Patent Document 1).

As described above, when observing the three-dimensional volume data, only the structure (object) of interest is extracted and observed among the structures included in the three-dimensional volume data. If the extracted result is stored once, there is no need to perform the extraction process every time. Therefore, in order to identify each object, as shown in FIG. 10, identification volume data 110 that indicates which object of blood vessel, bone, or internal organ each voxel of the three-dimensional volume data 100 representing an image corresponds to. I have it. For example, when the pixel value of each voxel is represented by 2 bytes, each voxel has 1-byte identification data separately from the 2 bytes, and an object is associated with each bit, and each bit is turned on / off. Is switched to indicate which object each voxel corresponds to.
JP 2006-338630 A

  However, if one voxel has 1 byte of identification data as described above, the amount of data further increases, causing problems in the magnetic disk capacity, the transfer time from the magnetic disk to the memory, and the required amount of memory. Come. Further, 1-byte identification data can be divided only into a maximum of 8 objects.

  Therefore, an object of the present invention is to provide an image processing apparatus, method, and program for managing a plurality of objects included in volume data without increasing the amount of data.

An image processing apparatus according to the present invention is an image processing apparatus that performs image processing by identifying each object of three-dimensional volume data including a plurality of objects.
A pixel value range calculation means for obtaining, for each object, a range of pixel values having a high appearance frequency among pixel values appearing in each object, from pixel values of voxels constituting each object extracted from the three-dimensional volume data;
Pixel value replacement means for replacing a pixel value of a voxel whose pixel value exceeds the range of each object among the voxels constituting each object with a value within the range;
Pixels that collectively shift the pixel values of all voxels constituting each object replaced by the pixel value replacement means for each object so that the ranges of pixel values of voxels constituting different objects do not overlap each other Value shifting means,
The pixel value replacement means replaces the pixel value of a voxel whose pixel value exceeds the maximum value of the range of each object among the voxels constituting each object with the maximum value or a value in the vicinity of the maximum value, Among the voxels to be configured, the pixel value of a voxel whose pixel value is lower than the minimum value of the range of each object is replaced with the minimum value or a value near the minimum value.

In addition, the program of the present invention provides a computer of an image processing apparatus that performs image processing by identifying each object of three-dimensional volume data including a plurality of objects.
A pixel value range calculation means for obtaining, for each object, a range of pixel values having a high appearance frequency among pixel values appearing in each object, from pixel values of voxels constituting each object extracted from the three-dimensional volume data;
Pixel value replacement means for replacing a pixel value of a voxel whose pixel value exceeds the range of each object among the voxels constituting each object with a value within the range;
Pixels that collectively shift the pixel values of all voxels constituting each object replaced by the pixel value replacement means for each object so that the ranges of pixel values of voxels constituting different objects do not overlap each other It is a program that functions as a value shift means,
The pixel value replacement means replaces the pixel value of a voxel whose pixel value is lower than the maximum value of the range of each object among the voxels constituting each object with the maximum value or a value near the maximum value, Among the voxels to be configured, the pixel value of a voxel whose pixel value exceeds the minimum value of the range of each object is replaced with the minimum value or a value near the minimum value.

The operation method of the image processing apparatus according to the present invention includes a pixel value range calculation unit that obtains a range of pixel values for each object from three-dimensional volume data including a plurality of objects, and a pixel that replaces a pixel value constituting each object. An operation method of an image processing apparatus comprising value replacement means and pixel value shift means for collectively shifting pixel values for each object ,
From the pixel values of voxels constituting each object extracted from the three-dimensional volume data by the pixel value range calculation means, a range of pixel values having a high appearance frequency among pixel values appearing in each object is determined for each object. A pixel value range calculation step to be obtained;
A pixel value replacement step of replacing, by the pixel value replacement means, a pixel value of a voxel whose pixel value exceeds the range of each object among the voxels constituting each object, with a value in the range;
The pixel value shifting means replaces the pixel values of all the voxels constituting each object replaced by the pixel value replacing step for each object so that the pixel value ranges of the voxels constituting different objects do not overlap each other. And a pixel value shift step for collectively shifting to
The pixel value replacing step replaces a pixel value of a voxel whose pixel value exceeds a maximum value of the range of each object among voxels constituting each object with the maximum value or a value near the maximum value, and Among the voxels to be configured, the pixel value of a voxel whose pixel value is lower than the minimum value of the range of each object is replaced with the minimum value or a value near the minimum value.

  “A value near the maximum value” means a value within a range that does not interfere with the maximum value when performing various image processing such as the generation of a projection image, and “a value near the minimum value” means a value near the minimum value. A value in a range that does not interfere with the minimum value when performing various image processing such as generation. The “value in the vicinity of the maximum value” and the “value in the vicinity of the minimum value” may be changed according to the type of the object and the image processing.

  An “object” refers to a structure that constitutes three-dimensional volume data to be subjected to image processing. For example, when the three-dimensional volume data is image data obtained by photographing a human body, each organ, tissue, Or each area which divided each organ according to anatomical structure, a function, etc. can be made into an object.

  According to the present invention, a range of pixel values is obtained for each object, and the pixel values of all voxels constituting each object are determined for each object so that the pixel value ranges of different objects do not overlap each other. By shifting in a batch, it is possible to identify which object each voxel represents depending on which range the pixel value of the voxel is in, and it is not necessary to refer to different identification data, so that The amount of data can be saved.

  Embodiments of an image processing apparatus according to the present invention will be described below in detail with reference to the drawings. FIG. 1 is a schematic configuration diagram of an image processing apparatus according to the present invention. The configuration of the image processing apparatus 1 as shown in FIG. 1 is realized by executing an image processing apparatus program read into the auxiliary storage device on a computer. At this time, the image processing apparatus program is stored in a storage medium such as a CD-ROM or distributed via a network such as the Internet and installed in a computer.

  The image processing apparatus 1 includes a storage unit 10 that stores the three-dimensional volume data 100, an extraction unit 20 that extracts each object of the three-dimensional volume data, and a voxel that constitutes each object extracted from the three-dimensional volume data 100. A pixel value range calculation means 30 for obtaining, for each object, a range of pixel values having a high appearance frequency among the pixel values appearing in each object from the pixel value, and the pixel value of the voxel constituting each object The pixel value replacement means 40 that replaces the pixel value of the voxel that exceeds the range with the value within the range, and the pixel value range of all the voxels that constitute each object do not overlap each other. As described above, the pixel value shifting means 50 that shifts each object collectively Equipped with a.

  The storage means 10 is a large-capacity storage device such as a hard disk. The storage means stores three-dimensional volume data 100.

  The three-dimensional volume data 100 is represented by a set of voxels obtained by finely dividing the three-dimensional space, and represents the density and density distribution of the image in the three-dimensional space by a three-dimensional data array. Specifically, the volume data 100 is obtained by stacking two-dimensional tomographic image data sequentially obtained along a direction perpendicular to the tomographic plane of the target object, and is captured by a CT apparatus, an MRI apparatus, or the like. It is generated by superimposing a plurality of tomographic images. For example, the volume data 100 obtained using the CT apparatus is obtained by storing the X-ray absorption amount for each voxel, and one pixel value for each voxel (the X-ray absorption in the CT apparatus). (Value indicating quantity) is given data.

  The three-dimensional volume data 100 is composed of a plurality of objects. For example, in the case of volume data generated from a tomographic image obtained by imaging a human body with a CT apparatus, the object refers to an organ or tissue such as a heart, kidney, bone, or blood vessel. Further, each area obtained by dividing an organ by an anatomical function or position is defined as one object. For example, the liver is anatomically divided into two sections D1 and D2 on the left and right sides of a cord-like structure called a liver sickle-like mesenchyme M (see FIG. 2), for vascular control and bile duct running Based on the Cantlie (country) line C (line connecting the bottom of the gallbladder and the inferior vena cava of the liver), it is divided into two left and right lobe areas D3 and D4 (see FIG. 3). Alternatively, it can be divided into several areas in order to clarify the positional relationship. For example, the classification method of Healey & Schrony (see Fig. 4) and Couinaud (see Fig. 5) is widely used. The leaves are combined and divided into 5 areas, and the latter is divided into 8 sub-areas (S1-8) by dividing the outer area, the front area, and the rear area up and down. Each area divided in this way is defined as one object.

  The extraction means 20 uses a CAD (Computer Aided Diagnose) function to extract each organ based on the pixel value and shape appearing in each organ and make it an object. The extracted organ may be further extracted as an object by dividing each organ into a plurality of sections based on the function, shape, blood vessel running, positional relationship with other organs, and the like.

  The pixel value range calculation means 30 obtains a range of pixel values having a high appearance frequency in each object from a statistical point of view, from the pixel values of voxels constituting each extracted object. If the range from the maximum value to the minimum value of the voxel pixel values included in the extracted object is the pixel value range, the pixel value of each object is always included in the range, but it can be obtained by shooting the subject. Since the obtained pixel value includes voxels having outliers exceeding the range of pixel values appearing in each organ, the range of pixel values becomes larger than necessary. Therefore, a range of pixel values having a high appearance frequency in the extracted object is obtained as a range of pixel values of each object. Specifically, for example, a range of pixel values of an object is a ± n × σ (n is a constant) from an average a of pixel values of all voxels constituting one object and its standard deviation σ. Usually, n = 2 is desirable.

The pixel value replacement unit 40 replaces a pixel having a pixel value exceeding the pixel value range of each object determined by the pixel value range calculation unit 30 so as to fall within the pixel value range of each object. The pixel values in each object include outliers that exceed the normal range of pixel values that appear in the object, but are clipped so that they fall within the range of pixel values of each object. Specifically, among the voxels constituting each object, the pixel value of the voxel whose pixel value exceeds the maximum value of the range of each object is replaced with the maximum value or a value near the maximum value, and the pixel value of the voxel below the minimum value Is replaced with the minimum value or a value near the minimum value. Even if such outliers are replaced with the maximum value or the minimum value, when the projection image is created and displayed, no difference appears in the image, and there is no serious problem in observation.

  The pixel value shifting means 50 collectively shifts the pixel values of all the voxels constituting each object for each object so that the pixel value ranges of the voxels constituting the different objects do not overlap each other.

  When the pixel value of each voxel is represented by 2 bytes, the range of the pixel value is 0 to 65535, but the pixel value included in the three-dimensional voxel data obtained by actually photographing the human body with the CT apparatus is usually O to 4096. Expressed in range. Therefore, as shown in FIG. 6, when a numerical value 0 to 65535 represented by 2 bytes is divided by a numerical value width of 4096, 16 sections of O to 4096, 4097 to 8192, 8193 to 10292,... Can be divided into Therefore, if one section is assigned to each object, the pixel values of 15 objects are mutually exclusive except for the range of all pixel values O to 4096 included in the original (original image) volume data 100. It can be memorized so as not to overlap.

  By the way, looking at the range in which the pixel value of each organ appears for each organ, the width is only about several hundreds. In view of this, according to the pixel value range of each object determined by the above-described pixel value range calculation means 30, the numerical values 0 to 65535 represented by 2 bytes can be divided into 16 or more sections. As shown in FIG. 7, the pixel values of all voxels included in each object are collectively shifted so that the pixel value ranges of each object do not overlap each other in a plurality of sections. Convert to the pixel value of the section.

  Next, a method for converting the pixel value of the volume data in the image processing apparatus of the present invention will be specifically described with reference to the flowchart of FIG.

  Volume data 100 is generated from a tomographic image taken with a CT apparatus or the like and stored in the hard disk (storage means 10) (S100).

  First, the extraction means 20 extracts N objects using a CAD (Computer Aided Diagnose) function (S101).

Next, the pixel value range calculation unit 30, each object _{object i (i = 0,1,2, ···} ) extracted for each, the mean value _{a i} and the standard of the pixel values of the voxels in the object object _i a deviation sigma _i, to limit the scope of the pixel values of each object object _i in a _i ± n * _{σ i.} n is a constant and may be determined so as to include most pixel values appearing in each object object _i . First, when the average value of the pixel values of the region extracted as the first object object ₀ (i = 0) is a ₀ and the standard deviation is σ ₀ , the pixel value range is a ₀ ± n * σ _0. (S102, S103). Within this range, it clips the pixel value in the object _{object 0, a 0 -n * σ} 0 following pixel values in the object object ₀ is in _{a 0 -n * σ 0, a} 0 + n * σ 0 The pixel value is replaced by the pixel value replacement means 40 so that the above pixel value becomes a ₀ + n * σ ₀ (S104).

Further, the pixel value is shifted for each object so that the range of pixel values of voxels included in each object object _i does not overlap. For example, as shown in FIG. 7, when the maximum pixel value of the original (original image) volume data 100 is Vmax, the pixel value shift means 50 changes the pixel value V in the object object ₀ to Vmax + (V− (a ₀ -n * σ ₀ ), and the pixel values are collectively shifted, whereby the pixel value of the object object ₀ is converted into a range of Vmax to Vmax + 2 * n * σ ₀ (S105). ).

Next, similarly for the second object object ₁ (i = 1) (S107), when the average value of the pixel values of the region extracted as the object object ₁ is a ₁ and the standard deviation is σ ₁ , Similarly, when the pixel value is replaced so that the pixel value in the object object ₁ falls within the range of a ₁ ± n * σ ₁ and the pixel value is shifted, the original volume data and the first object object are changed. as not suffer the pixel value with the _{0 Vmax + 2 * n * σ} 0 + (V- (a 1 -n * σ 1)) to convert (S102 to S105).

  By repeating this calculation for N objects (S103 to S107), it is possible to store the pixel values of a plurality of objects so that the pixel value ranges do not overlap by shifting the pixel values.

Using the voxel data thus converted, the image processing apparatus 1 performs various image processing such as generation of a projection image.

  In the conventional method of dividing an object using 8-bit (1 byte) identification mask data, the object can be divided into only the number of bits. Therefore, in order to divide into 8 or more objects, it is necessary to have identification data of 16 bits (2 bytes) or more for one voxel. However, in the current computer environment, 16 bits and 32 bits of identification data are provided for each voxel. It is difficult to have.

  Therefore, as described above, by using such an average value and standard deviation, the range of the pixel value of the object is determined, and the pixel value of each object is shifted within the range, so that one voxel pixel A 16-bit number representing a value can be divided into several sections to manage more than 15 objects.

  In the above description, the method of automatically extracting objects using the CAD function has been described. However, the operator can manually extract an organ while viewing the image, or divide the image into a plurality of areas and extract each as an object. You may do it.

  In the above description, the pixel value shift is performed at a time after extracting all the objects. However, the pixel value shift process may be performed every time the object is extracted.

  Alternatively, the extraction process is performed again on the object whose pixel value has already been shifted, the object whose pixel value is shifted is further divided into a plurality of objects, and the pixel value is shifted again for each divided object. You may make it perform.

Schematic configuration diagram of a projection image generation apparatus of the present invention Illustration for explaining the area of the liver (Part 1) Illustration for explaining the liver area (Part 2) Illustration for explaining the liver area (Part 3) Illustration for explaining the area of the liver (Part 4) Diagram for explaining the relationship between an object and a range where pixel values are separated The figure for demonstrating the relationship of the range of an object and the pixel value which appears in each object Flowchart for explaining a processing flow when converting a pixel value of volume data in an image processing apparatus Diagram for explaining volume rendering Diagram for explaining volume data for identification

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Memory | storage means 20 Extraction means 30 Pixel value range calculation means 40 Pixel value replacement means 50 Pixel value shift means 100 Three-dimensional volume data

Claims (3)

In an image processing apparatus that performs image processing by identifying each object of three-dimensional volume data including a plurality of objects,
A pixel value range calculation means for obtaining, for each object, a range of pixel values having a high appearance frequency among pixel values appearing in each object, from pixel values of voxels constituting each object extracted from the three-dimensional volume data;
Pixel value replacement means for replacing a pixel value of a voxel whose pixel value exceeds the range of each object among the voxels constituting each object with a value within the range;
Pixels that collectively shift the pixel values of all voxels constituting each object replaced by the pixel value replacement means for each object so that the ranges of pixel values of voxels constituting different objects do not overlap each other Value shifting means,
The pixel value replacement means replaces the pixel value of a voxel whose pixel value exceeds the maximum value of the range of each object among the voxels constituting each object with the maximum value or a value in the vicinity of the maximum value, An image processing apparatus that replaces a pixel value of a voxel whose pixel value is lower than a minimum value of the range of each object among constituent voxels with a value near the minimum value or the minimum value. A computer of an image processing apparatus that performs image processing by identifying each object of three-dimensional volume data including a plurality of objects,
A pixel value range calculation means for obtaining, for each object, a range of pixel values having a high appearance frequency among pixel values appearing in each object, from pixel values of voxels constituting each object extracted from the three-dimensional volume data;
Pixel value replacement means for replacing a pixel value of a voxel whose pixel value exceeds the range of each object among the voxels constituting each object with a value within the range;
Pixels that collectively shift the pixel values of all voxels constituting each object replaced by the pixel value replacement means for each object so that the ranges of pixel values of voxels constituting different objects do not overlap each other It is a program that functions as a value shift means,
The pixel value replacement means replaces the pixel value of a voxel whose pixel value exceeds the maximum value of the range of each object among the voxels constituting each object with the maximum value or a value in the vicinity of the maximum value, A program characterized by replacing a pixel value of a voxel whose pixel value is lower than a minimum value of the range of each object among constituent voxels with a value near the minimum value or the minimum value. Pixel value range calculation means for obtaining a range of pixel values for each object from three-dimensional volume data including a plurality of objects, pixel value replacement means for replacing pixel values constituting each object, and pixel values for each object collectively And an operation method of an image processing apparatus provided with a pixel value shifting means for shifting ,
From the pixel values of voxels constituting each object extracted from the three-dimensional volume data by the pixel value range calculation means, a range of pixel values having a high appearance frequency among pixel values appearing in each object is determined for each object. A pixel value range calculation step to be obtained;
A pixel value replacement step of replacing, by the pixel value replacement means, a pixel value of a voxel whose pixel value exceeds the range of each object among the voxels constituting each object, with a value in the range;
The pixel value shifting means replaces the pixel values of all the voxels constituting each object replaced by the pixel value replacing step for each object so that the pixel value ranges of the voxels constituting different objects do not overlap each other. And a pixel value shift step for collectively shifting to
The pixel value replacing step replaces a pixel value of a voxel whose pixel value exceeds a maximum value of the range of each object among voxels constituting each object with the maximum value or a value near the maximum value, and An operation method of an image processing apparatus, wherein a pixel value of a voxel whose pixel value is lower than a minimum value of the range of each object among the voxels to be configured is replaced with a value near the minimum value or the minimum value.

Images