KR100332072B1 - An image processing method for a liver and a spleen from tomographical image - Google Patents

Abstract

Preprocessing to remove images of background and muscles that are not part of the body from image data provided by tomography equipment, organ search process to estimate the intensity and location of organs to be extracted from the preprocessed images In the completed image, the unnecessary components attached to the outlines of organs are removed, the trimming and filling process of searching for the parts of organs in the internal pixels of the organ image, the information provided from the tomography apparatus, and the extracted organs An organ extraction and volume calculation method from a tomography image, characterized in that it comprises a long-term volume calculation process for calculating the volume of the extracted organ using the image information.

Description

An image processing method for a liver and a spleen from tomographical image

The present invention relates to a method for extracting liver and spleen images from tomography images and calculating their volume.

The liver is one of the organs that have a lot of medical attention because the function of the organs of the human body is very important and the Asians have a much higher incidence of diseases for the liver. The spleen is responsible for some of the functions of the liver and is known as an important organ to replace the function of the liver if it is not desired.

To date, the distinction and recognition of these organs is dependent on the visual judgment of the medical professional, and there is no way to find out the volume of each organ. In such an environment, it may not be easy to visually recognize or read each organ according to various conditions of the image, and it is difficult to diagnose a patient condition due to liver enlargement by quantitatively calculating the size of the liver. Thus, until now, the administration of a substance such as contrast enhancement medium has been performed to observe a particular organ highlighted by it, and a method of calculating the volume of an organ through autopsy of a corpse has been used.

An object of the present invention is to substantially eliminate the defects and defects of the prior art described above, and automatically extracts the organs by automatically estimating the contrast values of the organs from the tomography image data without using a contrast enhancer. It is.

Another object of the present invention is to improve the accuracy of organ extraction by removing the noise attached to the outline of the extracted organ.

Still another object of the present invention is to calculate the volume of an organ using header information provided from a tomography apparatus and information provided from an extracted organ image.

It is another object of the present invention to reduce the error of volume calculation by replenishing missing pixels inside the extracted organ.

A feature of the present invention for achieving the above object is a pre-processing to remove the image of the background and muscles that are not part of the body from the image data provided from the tomography apparatus, and the contrast of the organ to be extracted from the image of the pre-processing is completed Organ search process for estimating the value and position, trimming and filling process to search for the parts of the internal pixels of the organ image in the long-term image In this regard, the present invention comprises a long-term volume calculation process for calculating the volume of the extracted organ using information provided from the tomography apparatus and the extracted organ image information.

1 is a control block diagram showing a processing relationship of image processing according to the present invention;

2 is an exemplary raw CT image used for extraction of an organ;

3 is a histogram indicating the distribution of contrast values of each pixel in the image of FIG. 2;

4 is an exemplary CT image of an organ distinguished by a specific contrast value;

5 is a flowchart illustrating a method of extracting an organ according to the present invention;

6 is an exemplary view of a background removed from the raw CT image of FIG. 1;

7 is an exemplary view showing coordinates for indicating a position of a torso in the image of FIG. 6;

8 is an exemplary diagram of an image obtained by leveling an image of FIG. 7;

9 is an exemplary view of an image in which fat and bone components are removed from the image shown in FIG. 8;

10 is an exemplary view of an image in which muscle layer components are removed from the image of FIG. 9;

11 is an exemplary view showing a sample sampled for estimating a contrast value of an organ;

12 is a diagram illustrating an image in which a location search line for location search of an organ is set;

13 is an example of an image matrix of an algorithm for tracking pixels constituting an organ;

14 is an exemplary view of an image in which the liver and spleen are tracked;

15 is an exemplary view of the liver and spleen separated from the image of FIG. 14;

16 is an example of an image matrix of an algorithm of a smoothing process according to the present invention;

17 is an exemplary view showing the liver and spleen through the smoothing process according to the present invention;

18 is an exemplary view of an image matrix of an algorithm of a filling process according to the present invention;

19 is an exemplary view of the liver and spleen after the erosion process according to the present invention,

20 is an exemplary view of the liver and spleen after the dilation process according to the present invention.

Hereinafter, the configuration and operation of the present invention will be described in detail. 1 is a control block diagram illustrating a processing relationship of image processing according to the present invention. The image file provided from the tomography apparatus 10 is transferred to the image processing unit 30 through the interface card 20. At this time, the image processing unit 30 stores the transferred image information in the main memory device 40 and loads a program for extracting the organs stored in the auxiliary memory device 50 to estimate and extract the organs. Is output to the extracted image output device 60.

The tomography apparatus 10 includes a computed tomography apparatus (hereinafter referred to as CT) or a magnetic resonance tomography apparatus (MRI), and the image file in the present invention means a CT image. In addition, the extracted image output device 60 is a function of displaying the processed information so that the user can recognize, and may include a printer or a monitor.

The size of one CT image slice provided from the tomography apparatus 10 is 512 x 512 in size and contains an image in which each pixel can be represented by a contrast value of 0 to 4095. One pixel consists of two bytes, so one CT image slice is 512 x 512 x 2 = 520 kbytes. In addition, the header of the file contains various types of information, from information on the composition and format of the image to information on the patient.

In the present invention, the image slice information is converted into an 8-bit pgm file, and then processed for extraction of an organ. In this process, each pixel of the CT image file is transformed into an image having a contrast value of 0 to 255, which can be represented by 8 bits. In order to extract an organ, various characteristic information of an organ to be extracted may be used. There are various kinds of information such as the intensity value, location, shape, and size of each organ. In the present invention, the intensity value of pixels constituting the liver and spleen and location information of each are used.

In general, each organ included in one CT image of a normal person has a certain range of contrast values that are distinguished from other organs, and specific organs can be extracted using the distribution of the contrast values. However, some organs may have the same or similar contrast values, in which case additional information about the location of each organ is used.

2 is an exemplary diagram of a raw CT image used for extracting an organ, in which a liver appears relatively large among several organs. FIG. 3 shows a histogram based on contrast values of each pixel in such an image. As can be seen on the histogram, the intensity range shows the highest peak in the range of 90-92, and the portion with the intensity is the liver, spleen and muscle.

On the other hand, the contrast values of all the pixels that make up a particular organ are not necessarily in the range of a given value. Other parts included in the organ, such as microscopic blood vessels, components such as fat or water, may cause portions of the organ to have different contrast values. However, if this part is a small part of the organ from the overall perspective of each organ, it is treated as part of the organ even if it has a different contrast value.

In the present invention, the CT image used for the extraction of liver or spleen is limited only to a normal person who does not have a special lesion in the organ, and the contrast values of the liver and spleen to be extracted are each CT without using an estimated value. We started with the assumption that each image could be different.

FIG. 4 is a process in which pixels corresponding to a given specific range of contrast values are displayed to be distinguished from other portions, and parts of white color inside the body are pixels corresponding to a given range of contrast values. 4A shows the stomach, FIG. 4B shows the liver and the spleen, and FIG. 4C shows the bones. As can be seen from FIG. You can see the concentration. This means that each organ in the CT image has a range of contrast values that are distinct from other organs, and this information can be used to separate a particular organ from the entire image. Accordingly, the present invention is to automatically determine the contrast value of the organ to be extracted from a given CT image.

CT images contain images of various organs, depending on the area where they are taken, and organs such as liver, spleen, heart, stomach, and kidneys appear in multiple slices. Usually, one person is photographed from as few as 14 slices to as many as 20 slices, and the number of slices and the interval between slices may vary depending on the diagnosis purpose. However, not all slices taken contain liver and spleen. When 20 slices are taken, the liver usually contains 15 to 18 slices and the spleen contains 6 to 8 slices.

5 is a flowchart illustrating a method of extracting an organ according to the present invention. The S1 process removes the background that is not part of the body from the image data provided by the tomography apparatus, the standardization process that emphasizes the difference in contrast distribution between organs (S2 process), and the constant image in the image where the background is removed and only the body remains. S3 process of removing fat and ribs out of the contrast value, and setting the coordinates of the pixel that is removed from the innermost side around the body, and comparing these coordinates with the coordinates of the immediate surroundings. The S4 process of removing and treating the muscle as a muscle, and setting a large number of square constant sample areas in the slice in which the image of the organ is the largest among the slices from which the muscle is removed, S5 process of calculating the value, and the location search line of the organ by the predetermined distance in the image from which the muscle layer is removed S6 process of comparing the contrast values of the pixels, searching for the positions of pixels that match the contrast value of the previously investigated organs above the threshold, and determining the position as coordinates corresponding to the inside of the organs; S7 process of extracting each lung region of the liver and spleen by tracking them based on the contrast value estimated by 4-connectivity, and another lung around the position coordinates not included in the extracted lung region. S8 process of extracting the lung region separated by the extraction of the area, and a small outward protrusion where a given SE does not enter while scanning the set Structuring Element from the inside of the organ along the boundary line Trimming to remove noise on the outline of the extracted organ by removing the Filled with erosion and dilation processes, which are treated as tubules and converted into parts of the organ, and the holes with a size exceeding 6 pixels in diameter are repeated three times in order to judge other tissues inside the organ. Process (S10 process) is made. Hereinafter, the operation will be described in detail with reference to the accompanying drawings.

The entire CT image, consisting of 512 x 512 pixels, has a black background with almost zero contrast in the background except for the torso of the patient in the center of the CT image. Since this part is not the part corresponding to the actual body, it has nothing to do with extraction of the liver or spleen. Therefore, first remove these parts (transform them to white with a contrast value of 255). This process can be accomplished by scanning the entire image of 512 X 512 and changing the pixel to white when the contrast value is 70 or less and not part of the body. FIG. 6 is an image from which a black background of a background is removed from an input CT image converted into an 8-bit pgm format. The process of extracting the liver and spleen begins with these images.

On the other hand, the remaining part of the body after removing the background shows a slight difference depending on the shooting conditions of the entire 512 x 512 sized image. It becomes visible. Accordingly, the detailed process for extracting the liver and spleen is performed after the body position coordinates are determined within the entire 512 x 512 image. If the CT image is processed based on these coordinates, the time efficiency can be very improved compared to the process performed by reading the entire image of 512 x 512 size. FIG. 7 represents eight coordinates indicating the position of the torso in a given CT image of 512 × 512 size after the background is removed.

An equalization process is performed on the CT image from which the background is removed. Here, the leveling process is a processing process for increasing the ease of visual discrimination for the CT image. In the original CT image, each organ has its own distribution of contrast values, but the difference is not so severe that it is difficult to distinguish between organs at the boundary of organs. Therefore, it is possible to make the visual distinction easier by emphasizing the difference between the contrast values between the organs. The processing for this is the leveling process. However, this process is a process for improving the ease of image identification regardless of the extraction of organs. FIG. 8 is an image showing a result of performing a leveling process on an image from which a background is removed as shown in FIG. 6.

Fatty areas in the darkest part of the outermost area (less than 60) are found. Also inside is a muscle layer with almost the same intensity as the liver or spleen. Due to the fact that these parts have the same intensity values as the liver, slices that are attached to the liver or spleen in some locations, along with the pixels that make up the real organ, are mistakenly recognized as part of the organ when tracking the pixels corresponding to the liver and spleen. Can be extracted. In this case, the muscle layer is very irregular in shape, so it is very difficult to trim it later. Therefore, if the muscle part is not completely removed from the organ to be extracted, not only accurate extraction of liver and spleen is made but also the volume of the finally calculated organ has a small error.

Accordingly, in the present invention, such muscle parts are removed in advance, and the process is as follows. First, if you search for enough thickness along the left, upper, and right sides of the torso in the image where the background has been removed, the fat (contrast value below 60), muscle (contrast value 88-92), and bone (contrast value (115-178) At this time, the fat and bone except for the liver, the spleen and the muscle to be extracted are first removed using these contrast values.

When fat and bone are removed, irregular bands of muscle appear around the body. These muscles have almost the same intensity values as the liver or spleen, and some of them are attached to the liver or spleen inside. Therefore, the muscle part is searched for and removed from the liver and spleen inside the body. 10 shows an image in which the muscle layer is removed.

As mentioned above, the contrast values of liver and spleen are almost constant in the CT scan without contrast medium. When 90 to 92 pixels, which are irradiated with the liver's contrast value, are extracted from the image with only the background removed, other parts such as the spleen and muscle are also selected. Since the muscle part of these has already been removed in the process up to the S4 process and the liver and spleen are located on the opposite side, it is possible to extract the parts corresponding to the liver and spleen by using such position information. However, these contrast values are only statistical values, and therefore, they cannot be regarded as equally applicable to everyone's CT images. Therefore, in order to solve this problem, the present invention was able to find out in advance what distribution the contrast values of the liver correspond to for each individual CT image, and the process is as follows.

Among 20 slices of CT scans for a person, the slices containing the liver are 15 to 18 slices. Among the 8th to 12th slices, the liver was the largest, and its position was consistently observed from the top left to the bottom. This selects three parts of these slices that are considered part of the liver, and selects 50 x 50 pixels at each location. The intensity ranges of the pixels constituting the sampled portion were examined and the intensity ranges of the pixels constituting the top 95% were determined as the intensity values of the liver and spleen in the entire CT image of the patient. In addition, in order to eliminate the error of the decision was determined by applying all the 8th to 12th slice.

FIG. 11 illustrates a sample of pixels for estimating contrast values of organs in slices with large liver and spleen. In reality, the CT image from which the background and muscle layers have been removed is used. However, for the sake of simplicity, the image is expressed using a leveling image. Sampling to estimate the contrast value of the liver is as follows. The first sample of size 50 X 50 is taken from the position of the upper left coordinate (P, Q) of the muscle removed image 20 pixels away from the left diagonal. Again, sampling is performed for the second region of the same size in the region shifted 25 pixels left below the first region, and the third region is sampled in the same way.

The spleen is as follows. Sampling is performed for the first area of size 20 X 20 at a position shifted 10 pixels upward from the left diagonal coordinates (L, M) of the removed muscle image, and again in the right diagonal direction of the first area. Sampling of the second area, the third area in the same way, continues.

When applying this process to actual CT images, there are some differences depending on the images.However, in normal people, the liver and spleen occupy 95% or more of the pixels that make up the organs. It was found that the method of organ extraction using the light and dark value estimation increased the accuracy of the organs finally extracted.

As a next step, the initial values of the extracted organs can be obtained by tracking pixels corresponding to the liver and spleen using the intensity values obtained by this method, and extracting the traced pixels from a given image. However, determination of the location of each organ must first be made before organ tracking. The purpose of this positioning is to speed up the tracking of organs and, in some cases, to ensure correct navigation even when each organ is separated. It is also to prevent the liver or spleen from being processed if the initial search position is not inside the organ.

12 is an exemplary view illustrating an image in which a location search line for organ location search is set. The liver is always near the top left in one slice, and the spleen is located in the bottom right corner of the body (the location of each organ is almost fixed). You can find it by searching. Most organs are fixed in position, but each person or slice may have slightly different positions as the size and shape of the organs vary. Also, in some slices, organs appear in separate forms. Therefore, in the present invention, a position search line is set and used for accurate extraction of such organs. The location search line used was set to identify the existence of the liver, spleen regardless of size, shape, or separation in the image. The exact position of the position search line and the process of setting the position coordinate using the same are as follows.

In the case of the liver, the position search line is set at the center wall on the left side of the body from which the muscle layer has been removed and is set to the position 2/3 of the left side of the body. Moving this position search line by 20 pixels to the right and comparing the contrast values of the pixels in the upward direction, if a pixel falls within the range of the contrast value of the already investigated organ, eight pixels around that pixel are also examined. If it is within the contrast range of, the position is maintained at the coordinates corresponding to the inside of the liver. Another search line for the liver is located at the top of the center of the body. This is a search line to find these isolated livers, since they are usually located at the top of the center of the body.

The search line for the search of the spleen starts from the center of the right side of the body and to the 2/3 position of the right side. The spleen search line is moved by 10 pixels from right to left to search for the location of the spleen in the downward direction and remember the location. Through the exploration, a number of coordinates identified as internal to each organ are stored and maintained in the liver and spleen. The liver and spleen can be extracted by extracting the lung region containing each coordinate using these coordinates in the next process of organ extraction. Can be. The important fact here is to maintain multiple location search coordinates for each organ and to track the organs using all of these coordinates to detect the separated organs.

The position of the liver is tracked to keep track of the connected pixels that make up the liver using a set of position coordinates. This separates only pixels connected by 4-connection from a given coordinate. 13 is an example of an image matrix of an algorithm for tracking pixels constituting an organ. Starting from the initial coordinates of the portion indicated by B in FIG. 13C, the contrast values of the pixels fall within a given range. This initial coordinate may be a pixel located at any point in the area to be selected. If the pixel of the initial coordinate is included in the range of the given contrast value, it is changed to another contrast value (= 0), and then compared with respect to the pixel of (x-1, y) shown in FIG. 13A. If this pixel is also included in the range of examined contrast values, it is considered connected by 4-connectivity and recursively calls the tracking routine whose initial coordinate is (x-1, y). If (x-1, y) is not connected, each direction is compared in the order shown in FIG. 13B, and if connected, recursion is similarly performed. At this time, the pixel irradiated as connected changes the contrast value to 0 and is not irradiated again. Starting from a point in the closed region as such, all pixels connected as shown in FIG. 13D are displayed. For the image of FIG. 13C, only the portion where the contrast value is changed to 0 in FIG. 13D is left and the remaining portion is removed, leaving only the connected regions as shown in FIG. 13E. If this is done for all retained position coordinates, no new search is made if the coordinates found in the same closed area are within the range searched first for the previous position coordinates. In addition, if the search is performed using a position coordinate that is not included in the previous search range, it means a part of the separated organ. This allows tracking and extraction of liver and spleen within CT images.

14 is an exemplary view of an image in which the liver and spleen are tracked when this process is applied to a CT image. Each organ was traced and extracted from the original CT image so that it can be compared from the leveling image. The CT image of the muscle layer removed shows the dark and dark values of the liver and the spleen, and shows the pixels traced as part of the liver and spleen in black. FIG. 15A is an exemplary view of the liver separated from the image of FIG. 14, and FIG. 15B is an exemplary view of the spleen separated from the image of FIG. 14.

As can be seen through Figure 15a and 15b, looking at the liver and spleen images extracted from the muscle removed image, it can be seen that there is an uneven protrusion on the outer portion of the image. These parts were extracted with the liver or spleen without separation of other tissues around each organ, such as muscles that could not be removed or parts of other organs closely related to the organ, or evenly distributed contrast values in these areas. Because it does not. The trimming process removes these noises so that each finally extracted organ has a smooth outline.

The algorithm works as if the ball of the appropriate size was rolled along the boundary line inside the liver and spleen, and the outward protrusion where the ball did not enter was removed. More precisely, it uses a morphological filter that scans a structured element (SE) of any size over the entire image, with outward projections with a thickness less than the given SE. Will be removed.

In practice, the algorithm implemented uses a SE of 12 * 12, but for the sake of convenience, the SE of 3 × 3 will be described as an example. For the image of FIG. 16B, the SE of FIG. 16A is compared with the image to be applied while moving one pixel from the upper left one by one. The pixel is selected only when all pixels (9) are equal to SE. Thus, while the portion indicated by B in FIG. 16C and its periphery remain, the region of thickness smaller than SE, such as the portion indicated by A, is removed. However, although the region of B 'is a terminal portion of the liver, the following method is taken to prevent damage due to inconsistent with SE. This compares with SE and counts the number of matches so that only seven pixels, not nine, are selected. This slight tolerance allows the inclusion of any irregular shape at the boundary, thus reducing the loss of the original image. In the case of SE in the form of a circle, the irregular boundary part becomes the curve of the circle, but by using this error tolerance method, the irregular shape can be kept as it is. In FIG. 16C, a portion of the relatively thick muscle remains, but only the lung region is extracted based on the coordinates obtained through the location search of each organ, thereby removing the part not belonging to the organ, thereby obtaining the result of FIG. 16D.

In the present invention, since the size of the SE is fixed to 12 x 12, no effect is obtained when the object to be removed has a thickness of 12 pixels or more. In this case, a larger SE size can be used to remove thicker protrusions. However, if the SE size is too large, the execution time becomes longer. Therefore, the size of SE is adjusted and applied according to the characteristics of the image to be processed.

The resultant image obtained by trimming the image tracked and separated from the original image by the above process is shown in FIG. 17. 17A shows an image of the liver that has undergone a smoothing process, and FIG. 17B shows an image of the spleen that has undergone a smoothing process. It can be seen that it has a much smoother outline than the images of FIGS. 15A and 15B.

The edges were trimmed by trimming the extracted liver and spleen images, but there are still tiny dots or small holes that appear white with contrast values of 255. These parts are not within the range of the organ's contrast values estimated at the previous stage of estimating the contrast values, but some blood vessels or other microscopic parts that pass through the organ's interior, which, depending on their size, need to be included in the organs. It is a judgment. However, inside the organs currently extracted, these parts are treated as white with a contrast value of 255 and are not considered to be included in the organs. Therefore, in obtaining the volume of an organ, it may be treated as not part of the organ, thereby making an error in obtaining an accurate organ volume. Therefore, these parts should be treated with the same intensity value as each organ to obtain a volume.

Erosion and dilation are used to convert small holes in the liver and spleen into parts of the organ according to their size. When the erosion process is applied to the image, the white part is reduced and the black part is enlarged. Applying this algorithm to CT images, the white part in the image background of the extracted organs decreases with the execution of erosion, which in turn results in an increase in the size of the entire organs. In addition, the white parts inside each organ gradually decrease as the size of the surrounding black parts increases, and as a result, they appear white in the liver because they have different values from the extracted liver contrast values. The parts become smaller in size as erosion is applied, and the diameter of the white hole is reduced by two pixels by one erosion.

Dilation refers to an operation that has a concept opposite to erosion. By dilation, the size of the white area is gradually increased and the size of the black area is gradually reduced. That is, the total size of the organs grown by erosion can be restored to the original size by dilation. However, if the white area that existed inside the extracted organ is completely black by the previous erosion, it is not restored again.

The important point here is that, as already mentioned, not all white parts inside the liver are treated as liver. Depending on the size of the white part, some may be other tissues that pass inside the liver. Accordingly, the determination of the number of executions of the erosion has become an important problem, and in the present invention, the number of executions of the erosion is limited to three times, so that a hole smaller than 6 pixels in diameter is treated as a small vessel and converted into a part of the liver. I was. More than that, however, it was considered part of another relatively large tissue passing through the liver and restored to its original size. If the size of the white area is more than 6 pixels, the size of the erosion process is reduced even after 3 erosions, but the white area may remain. Erosion removes all white parts, so even if dilation is performed, the original shape cannot be restored.

In conclusion, by performing dilation after erosion, the size of the extracted whole organs can be transformed as a part of organs. This decision was based on the judgment of a medical professional in the field of diagnostic radiology.

18 is a diagram illustrating an image matrix of an algorithm of a filling process according to the present invention. FIG. 18A illustrates an original image matrix, and FIG. 18B illustrates a result of performing one erosion, and FIG. 18C illustrates an image matrix according to a result of one dilation.

FIG. 19A illustrates an image of the liver after performing erosion from the image of FIG. 17A, and FIG. 19B illustrates an image of the spleen after performing erosion from the image of FIG. 17B. FIG. 20A is a diagram illustrating an image of the liver after dilation is performed from the image of FIG. 19A, and FIG. 20B is a diagram illustrating an image of the spleen according to the result of performing dilatinization on the image of FIG. 19B.

The header portion of the CT image file provided by the tomography apparatus holds various information about the CT image. Various environmental information, such as the sex and age of the patient and the size of the CT image taken, is maintained. In particular, among various environmental information, information about the interval between slices and information about the size of pixels constituting the image is useful for calculating the volume of the extracted organ. The number of slices of a CT image photographed for a person may vary depending on a photographing purpose, but is usually about 20 slices. In this case, the distance between the CT image slices is about 10 mm, and the size of one pixel is about 0.65 mm. However, not all CT images have a constant interval and size, so this information should be analyzed in the header portion of each CT image.

The volume of each organ can be calculated by the following equation.

Among the symbols used in the formula, 'N' means the number of slices containing the extracted organ (liver or spleen), 'S _i ' is the slice number, 'D' is the interval between slices, and 'L _p ' is the extraction The number of pixels constituting the old organ (liver or spleen), 'X' means the horizontal length of one pixel, 'Y' means the vertical length of one pixel.

Algorithms for extracting organs and calculating volume mentioned above were applied to a total of 30 CT images. From these results, it can be seen that the accuracy of the extracted organs was significantly accurate when compared with images separated by manual methods by a specialist in diagnostic radiology. This is supported by the fact that the difference between the volume obtained using the extracted organ image and the volume obtained from the manual image is within 5%.

Table 1 below compares the average value of the volume obtained from the liver and spleen images extracted by the method proposed in the present invention with the average value of the volume obtained by a manual method such as an autopsy.

Volume according to the present invention Volume by manual error Liver 1.214 liter 1.172 liter 3.58% Spleen 0.225 liter 0.216 liter 4.17%

Although there are some differences depending on the CT images, the spleen shows a slightly larger error than the liver. This is because the size of the spleen is relatively small. In conclusion, the extraction of liver and spleen according to the present invention and the volume calculation of organs through it can be said to be relatively accurate.

As described above, the present invention has an effect of automatically estimating the contrast value of the tomography image slice without using a contrast enhancer to track and extract the organ. In addition, it is possible to increase the accuracy of the extraction by removing the noise attached to the outline of the extracted organ, it is possible to calculate the volume of the organ through the information provided from the tomography apparatus and the area information of the extracted organ.

Claims (5)

A method for extracting images and volume values of liver and spleen by using a computer device from organ image data provided from a tomography apparatus; Pre-processing to read image data of organs stored in computer readable form through tomography device to create image data from which images of background and muscle are not part of body Process, A long-term search process for estimating the intensity and location of the organ (liver or spleen) to be extracted from the preprocessed image; A smoothing and filling process of removing unnecessary components attached to the outline of organs in the organ search process and searching for a part corresponding to an organ part among internal pixels of the organ image; And a long-term volume calculation process of calculating a volume of the extracted organ by using the information provided from the tomography apparatus and the extracted long-term image information. The method of claim 1; The pretreatment process Converting the image data provided from the tomography apparatus into an 8-bit pgm format, Searching and removing the background image which is not directly related to the extraction of organs by row and column for the whole image, Standardization process that emphasizes the difference in contrast between each organ; The process of removing fat and ribs out of a certain contrast value in the image where the background is removed and only the body remains Set up the coordinates of the pixel that is being removed from the innermost with respect to the periphery of the body, and compares these coordinates with the coordinates of the immediate surroundings and removes the difference of 3 pixels or more as a muscle Long-term extraction and volume calculation method from photographed image. The method of claim 1; The long-term search process Calculating the intensity of the liver from the distribution of the intensity values of pixels in the region by setting a large number of square sample areas in the slice in which the muscle among the slices from which the muscle is shown is the largest; Calculating a spleen's contrast value from the distribution of the contrast values of pixels in the region by setting a large number of square sample areas in the slice in which the muscle of the slice from which the muscle is removed appears the largest; Moves the location search line of the organ by a predetermined distance from the center left wall of the image where the muscle layer is removed, compares the intensity values of the pixels, searches for the position of the pixels that match the intensity and the threshold of the previously irradiated liver. Judging by internal coordinates, A process of searching for separated liver images that may appear in some cases by comparing the contrast values of pixels by moving a predetermined distance by using a separate search line set at ¾ position below the body of the upper middle part of the muscle layer, Move the location search line of the organ by a certain distance starting from the opposite side to the start position of the liver and compare the contrast values of the pixels and search for the position of the pixels that match the contrast value of the spleen that is already irradiated more than the threshold. Judging by coordinates corresponding to the inside of the spleen, Extracting each lung region of the liver and spleen by including the coordinates by tracking the pixels determined to be the liver and spleen based on the contrast value estimated by 4-connectivity; And extracting a lung region separated from the liver by extracting another lung region based on the positional coordinates not included in the extracted lung region. The method of claim 1; The trimming and filling process Trimming to remove the noise attached to the outline of the extracted organ by scanning the constructed structuring element along the boundary line from the inside of the organ to the outside while removing the small outward protrusion that does not enter the given SE. Trimming process, Small holes with a diameter of 6 pixels or less inside the organs are treated as microvessels and converted into parts of the organs. Holes larger than 6 pixels in diameter are repeated three times to determine other tissues inside the organs. And organ filling and volumetric calculation method from tomography images, characterized in that the filling process consisting of a dilation process. The method of claim 1; The long-term volumetric calculation process The interval of the image slice is calculated from the information included in the header portion of the image slice provided by the tomography apparatus, and the area of the organ is calculated from the horizontal length and the vertical length of each pixel of the extracted organ. N: number of slices containing the extracted organs S _i : Slice number D: spacing between slices L _p : Number of constituent pixels of the extracted organ X: width of one pixel Y: height of one pixel An organ extraction and volume calculation method from a tomography image, characterized in that the volume of the organ is obtained by