JPH11250263A - Method for automatically collating slice picture of chest three-dimensional tomographic picture, and record medium recorded with the program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for automatically collating the slice picture of a chest three-dimensional tomographic by which the slice pictures at the same position of a body are automatically extracted from the chest three- dimensional tomographic pictures of two pairs of the same persons, and to provide a recording medium in which the program for the method is recorded. SOLUTION: A diagnostic picture and a comparison picture are inputted to a computer (S1 and S2), the comparison picture is matched with the diagnostic picture by the parameter change (parallel movement, enlargement and reduction, and rotation) of the comparison picture (S3 and S4), the mean value of the absolute value of the difference of the concentration of each picture element is obtained as a defference Diff for the AND area of the both pictures (S5 and S6),the minimum difference Dm ,n of the defference Diff is decided by changing the parameter (S7-S9), and the comparison picture having the minimum value among the minimum defference Dm ,n is decided for each comparison picture (S10-S12). The comparison picture having the minimum value is decided also for each diagnostic picture, and a corresponding table is prepared (S13-S15).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for automatically comparing slice images when comparing and interpreting a chest three-dimensional tomographic image, such as a computed tomography (CT) image, taken at different times. is there.

[0002]

2. Description of the Related Art The comparative interpretation of past and present chest CT images is a clue for determining whether or not a shadow appearing in an image is a tumor, whether or not the shadow is malignant.

Conventionally, when comparing images having different imaging timings, there has been a method of automatically extracting a difference (Th).
Follow-up observation algorithm for 3D chest images by in Section CT: Gaku Tahara, Yoshiki Kawata, Noboru Niki et al .: The 4th Collection of Chest CT Screening Workshop Vol. 4, NO
3) There was no technology to directly assist doctors in performing comparative reading.

[0004]

Therefore, the doctor needs to
When trying to make a diagnosis by comparing the T image with the past, the doctor manually corrects the shift of the slice image due to the difference in the respiratory state of the examinee and the imaging position, so that it takes time to interpret the image, There was a problem that the burden was large. In particular, this burden becomes a serious problem when performing large-scale image reading such as a mass examination.

It is an object of the present invention to provide two sets of chests 3 of the same person.
An object of the present invention is to provide a method for automatically checking a slice image of a chest three-dimensional tomographic image capable of automatically extracting a slice image at the same position of the body from a three-dimensional tomographic image, and a recording medium on which a program for the method is recorded.

[0006]

SUMMARY OF THE INVENTION In order to solve the above-mentioned problems, the present invention is directed to a second embodiment having an image sequence having a slice plane in which the body section of the chest is in the X and Y directions and a body axis in the Z direction. 1
In collating one slice image of the three-dimensional tomographic image with one slice image of the second three-dimensional tomographic image photographed at another time, the slice image of the first three-dimensional tomographic image Inputting one diagnostic image extracted from the array and one comparative image extracted from the slice image array of the second three-dimensional tomographic image; and modifying the comparative image to match the diagnostic image in the slice plane direction. A parallel movement amount α in the X-axis direction, a parallel movement amount β in the Y-axis direction, scaling by a magnification γ in the X-axis direction, a magnification δ in the Y-axis direction, and a rotational movement by an angle θ. Extracting, by labeling, a diagnostic image of slice image number m and a comparison image of slice image number n deformed by the above process, and extracting an AND region of both images;
Calculating the average value of the absolute values of the differences between the densities of the pixels in the AND region of the comparison image and the diagnostic image as the difference D _iff , and calculating the difference D _{iff using} the translation amounts α, β, A step of selecting the enlargement / reduction ratios γ, δ and the rotation angle θ; and a minimum difference D _m , _{n of the} difference D _Diff obtained in the selection step, for each slice image of the second three-dimensional tomographic image. The method further comprises a step of calculating and a step of determining a comparison image having the minimum value among the minimum dissimilarities D _m , _n of the respective comparison images as a slice image collation position for comparative reading with the diagnostic image.

Further, according to the present invention, in the process of determining the comparison image having the minimum degree of difference D _m , _n as the slice image collation position, the body axis direction is set as the Z axis, and The value sequence of the minimum dissimilarity D _m , _n of the comparison image is interpolated by a quadratic function, and the Z-axis coordinate at which the minimum value of the quadratic function is obtained is a slice image collating position corresponding to the accuracy of the interval of the slice image sequence or less. It is characterized by having the process of.

Further, the present invention provides one slice from a file of a first three-dimensional tomographic image having a slice plane in which the body section of the chest is in the X and Y axis directions and an image sequence in which the body axis is in the Z axis direction. A procedure for extracting an image as a diagnostic image and extracting one slice image as a comparative image from a file of a second three-dimensional tomographic image captured at another time, and a slice plane as a modification for matching the comparative image with the diagnostic image In the X-axis direction, the translation amount β in the Y-axis direction, the magnification is reduced in the X-axis direction by the magnification γ, the magnification is reduced in the Y-axis direction by the magnification δ, and the rotation is performed by the angle θ. Extracting the AND area between the diagnostic image of the slice image number m and the comparison image of the slice image number n deformed by the above process by labeling; The average value of the absolute values of the differences between the densities of the pixels in the AND region of the cut image is calculated as the difference D _iff
, A step of selecting the translation amounts α and β, the enlargement / reduction ratios γ and δ, and the rotation angle θ in the calculation of the difference D _Diff , and a step of selecting the difference D obtained in the selection process.
calculating the minimum dissimilarity D _m , _n of the _iff for each slice image of the second three-dimensional tomographic image; and comparing the comparison image having the minimum value among the minimum dissimilarities D _m , _n of the comparative images. A program for causing a computer to execute the diagnosis image and a procedure of determining a slice image comparison position to be compared and interpreted is recorded on a computer-readable recording medium.

Further, according to the present invention, in the step of determining a comparison image having the minimum dissimilarity D _m , _n as a slice image collation position, the body axis direction is defined as a Z axis, and each of the plurality of images corresponding to the Z axis direction is determined. The value sequence of the minimum dissimilarity D _m , _n of the comparison image is interpolated by a quadratic function, and the Z-axis coordinate at which the minimum value of the quadratic function is obtained is a slice image collating position corresponding to the accuracy of the interval of the slice image sequence or less. Is recorded on a computer-readable recording medium as a program for causing a computer to execute the procedure described above.

According to the above-described automatic matching method and recording medium, an appropriate comparison image is automatically determined from the difference in the degree of difference with respect to the diagnostic image, and a correspondence table is created by the automatic matching or the matching result. If so, when a request from a doctor is made, a corresponding comparison image is automatically searched and displayed, thereby reducing the burden on the doctor and shortening the image interpretation time.

[0011]

Embodiments of the present invention will be described below with reference to the drawings.

It is assumed that a current and past images are compared for a chest X-ray CT image sequence for lung cancer screening having a slice thickness of 1 cm. A chest X-ray CT image for lung cancer screening is captured using spiral CT, and one slice image having a body section of the chest in the X and Y axis directions is 1 cm in the Z axis (body axis) direction.
Minute information is included.

FIG. 2 shows a chest CT showing an embodiment of the present invention.
FIG. 2 is a configuration diagram of an image comparison reading apparatus. Reference numeral 1 denotes a file in which a three-dimensional tomographic image sequence of the lung taken in the previous year (N) is stored as a slice image, and 2 denotes a three-dimensional tomographic image sequence of the lung taken in the current year (M). This is a file saved as a slice image. The CT image sequences stored in the two files 1 and 2 are different due to a difference in an imaging position and a deformation of a lung due to respiration. Reference numeral 3 denotes a file for storing, as a three-dimensional tomographic image, a slice image sequence as a result of automatically matching each slice image of the file 1 with each slice image of the file 2 by computer processing.

Reference numeral 4 denotes an automatic collation function for storing a file 3A obtained as a result of automatically collating each slice image of the file 1 with each slice image of the file 2 as a slice image sequence in the file 3. File 2
This is a computer system having a comparative reading support function of taking out a slice image 3A at the same position as one slice image 2A taken out of the display device and displaying the slice image 3A on the display device 4A in parallel.

The computer system 4 is equipped with software for realizing the above-mentioned automatic slice collation function and comparative reading support function. Among these, the automatic collation processing is performed for one slice image (diagnosis image) 2A extracted from the file 2 captured in the current fiscal year with respect to the slice image (comparison image) sequentially extracted from the previous year's slice image sequence in the file 1. In order to extract a slice image at the same collation position by transforming 1A, the degree of difference is calculated. As the degree of difference, the average value of the absolute values of the differences between the densities of the respective parts of the diagnostic image 2A and the comparative image 1A is used.

Since the density of the image is an X-ray absorption value, the value differs depending on the body tissue. Therefore, a difference in concentration means a difference in body tissue. Looking at the tomogram of the lung area, the shape of the lung and the shape of the vertical section differ depending on the position of the Z-axis of the body. At different positions, the composition ratio of the tissue changes and the difference in density increases. Find the degree of difference.

FIG. 1 is a flowchart of a process (automatic collation function) for calculating the degree of difference and determining a comparative image to be interpreted based on the difference by software installed in a computer in order to support comparative image interpretation. is there.

In FIG. 1, first, image data of a diagnostic image 2A and any one comparative image 1A are input (steps S1 and S2).

Next, the comparative image 1 is matched with the diagnostic image 2A.
In order to apply a transformation to A, a transformation parameter having an appropriate value is input (step S3), and the comparative image 1A is transformed (step S4).

This deformation involves enlargement / reduction of the comparative image 1A in the X-axis and Y-axis directions in order to absorb a change in lung volume due to thorax in lung deformation due to respiration. Further, in order to absorb a change in the volume of the lung due to the elevation of the diaphragm, the image sequence is moved up and down. Furthermore, rotation movement is performed to absorb a difference in body orientation when lying on the bed, and parallel movement is performed to absorb a difference in imaging position.

Next, the diagnostic image 2A and the deformed comparative image 1
For each of A, a body region is extracted by labeling, and as shown in FIG. 3, a body region a of the diagnostic image 2A,
An AND set that is a partial region included in both regions of the body region b of the comparative image 1A is obtained (step S5).

Next, the calculation of the dissimilarity is limited within AND set, the area of the AND set A of the body region of the diagnostic image 2A and the comparative image 1A which is deformed by the deformation parameters and S _A, diagnostic images 2A And the comparison image 1A, the average value of the absolute values of the difference values of the densities of the corresponding pixels is calculated to calculate the degree of difference _Diff (step S6). This difference D
_iff is defined as follows.

[0023]

(Equation 1)

Where α and β are the degrees of difference _Diff are the parallel movement amounts in the X and Y axes directions, γ and δ are the scaling factors in the X and Y axes directions,
θ is the rotation angle on the X, Y axis plane, f _m (x, y) is a function representing the density of the pixel at the coordinates x, y in the m-th diagnostic image, and gn (γx-α, δy-β, θ) is a function representing the density of the pixel rotated by the angle θ at the deformed coordinates γx-α, δy-β position in the n-th comparative image.

These deformation parameters α, β, γ, δ,
θ is discretely changed (steps S7 and S8), and the degree of difference _Diff is calculated in consideration of the degree of lung respiration,
The minimum dissimilarity D _m , _{n having the} minimum dissimilarity D _iff
Is obtained (step S9).

According to the above procedure, N comparative images are sequentially switched and extracted from the file 1 for the m-th diagnostic image 2A as shown in FIG. Is calculated for each of the minimum differences D _m and _n (S10, S11).

[0027] To determine the comparative image 1A _n with the smallest value among these minimum dissimilarity D _{m, n} (step S12). Determining the comparison image 1A _n as collation subject to image diagnostic images 2A _m.

[0028] Figure 5, N of one with respect to the diagnostic image 2A _m
It is an example of a graph of the minimum dissimilarity _Dm , _n of two comparative images.
In this example, the second slice image of the diagnostic image (m = 2)
, The third slice image of the comparison image (n = 3)
Indicates a corresponding image.

According to the above procedure, a comparison image in which the minimum difference D _m , _n is obtained for each diagnosis image is determined (steps S13 and S14), and a correspondence table of the comparison image to the diagnosis image is created (S15). ).

In this way, when there is a request for comparison with a past image during a doctor's interpretation, a corresponding comparison image is searched from the correspondence table, and the diagnostic image and the comparison image are displayed in parallel on the display device 4A. By displaying, comparative image reading becomes possible.

In the above embodiment, the use of the correspondence table for comparative reading is only an example, and other comparative reading can be used.

Further, in selecting the value of the deformation parameter, the amount of calculation can be reduced by limiting the amount of translation, the rate of enlargement / reduction, and the rate of change of the rotation parameter within the range of the degree of lung deformation due to respiration. . For example, the amount of movement of the diaphragm due to breathing is 0.8 cm to 8.
A value of 1 cm is disclosed in the document "Measured values to know for diagnostic imaging: Kenichi Kodera and Kyoichi Hiramatsu: Medical Shoin". Considering this value, narrowing the upper and lower limits of parameters Can reduce the amount of calculation.

In the calculation of the degree of difference _Diff , the amount of calculation can be reduced by calculating the degree of difference by lowering the resolution.

In automatic slice image comparison between the diagnostic image and the comparison image, the slice images are discrete with respect to the Z-axis. However, as shown in FIG. 6, the minimum difference D _m , _n between slice images corresponding to the Z-axis direction. Is interpolated by a quadratic function, and a Z-axis coordinate value that takes the minimum value of the quadratic function can be obtained.

The processing procedure for this is as shown in FIG. 7, and step S16 is performed instead of step S12 in FIG.
Is added, and the value sequence of D _m , _n of each comparative image is interpolated by a quadratic function, and the Z-axis position of the minimum value is determined as the collation position.

The collation position is not always an integer value indicating the number n of each slice image sequence of the comparison image, and may indicate a position between the numbers. Automatic collation can be performed as the collation position. Further, the interpolation in the Z-axis direction is effective because it is also used for display at the time of independent interpretation.

The above description is for the case of collation using a chest CT image. Instead, another three-dimensional tomographic image of the chest, for example, an MRI (Magnetic Resona) is used.
(Non-Imaging) image can be used to perform similar image interpretation.

The present invention provides a computer-readable recording medium, such as a floppy disk, for executing a program for causing a computer or the like to execute a procedure or a calculation algorithm for realizing the processing shown in FIGS. It can be recorded on a memory card, MO, CD, DVD or the like and distributed.

[0039]

As described above, according to the present invention, the chest 3
When comparing a 3D tomographic image with a chest 3D tomographic image photographed at different times, the comparative image is matched with the diagnostic image by changing deformation parameters to determine a comparative image for the diagnostic image, and each pixel of both images is changed. Since the comparison image in which the degree of difference in density is the minimum value or the minimum value obtained by interpolating the minimum value with a quadratic function is automatically extracted as the position of the image to be interpreted, the correspondence table is used when a request from the doctor is made. By searching for and displaying the corresponding comparative image from, for example, there is an effect that the burden on the doctor can be reduced and the image interpretation time can be reduced.

[Brief description of the drawings]

FIG. 1 is a processing procedure showing an embodiment of the present invention.

FIG. 2 is a configuration diagram of a comparative image reading apparatus for chest CT images according to the present invention.

FIG. 3 is an explanatory diagram of an AND set in the embodiment.

FIG. 4 is an exemplary view for explaining a method of determining a comparison image in the embodiment.

FIG. 5 is a characteristic example of a minimum difference D _m , _n of each comparative image in the embodiment.

FIG. 6 is an example of characteristics obtained by interpolating a minimum difference D _m , _{n according} to another invention with a quadratic function.

FIG. 7 shows a processing procedure in another invention.

[Explanation of symbols]

1… File of previous year 2… File of this year 2A… Diagnostic image extracted from file of this year. 3. File matching previous year's file with this year's file 3A. Comparative image extracted from previous year's file 4. Computer system 4A. Display device

Claims (4)

[Claims] 1. A first section having an image sequence in which a body section of a chest is a slice plane in X and Y axis directions and a body axis is in a Z axis direction.
In collating one slice image of the three-dimensional tomographic image with one slice image of the second three-dimensional tomographic image photographed at another time, the slice image of the first three-dimensional tomographic image Inputting one diagnostic image extracted from the array and one comparative image extracted from the slice image array of the second three-dimensional tomographic image; and transforming the comparative image to the diagnostic image in the slice plane direction. Is translated in the X-axis direction and the translation amount β in the Y-axis direction.
The process of enlarging / reducing at the magnification δ in the Y-axis direction and rotating and moving by the angle θ, the diagnostic image of the slice image number m, and the comparison image of the slice image number n deformed by the above process are labeled, and the two images are compared. Extracting an AND area, calculating an average value of absolute values of differences in density between pixels in the AND area of the comparison image and the diagnostic image as a difference D _iff , and calculating the difference D _iff A step of selecting the translation amounts α and β, the enlargement / reduction rates γ and δ, and the rotation angle θ; and the minimum difference D _m and _{n of the} difference D _Diff obtained in the selection step is the second three-dimensional value. determining a process of calculating respectively for each slice image of the tomographic image, as the minimum dissimilarity D _m, a slice image matching position comparative reading with the diagnostic image comparison image having the minimum value of _n of the each comparative image Slice image automatic collation method chest three-dimensional tomographic image, comprising: the extent, the. 2. The method according to claim 1, wherein, in the step of determining a comparison image having the minimum difference D _m , _n as a slice image collation position, the body axis direction is set to a Z axis, and the Z axis direction corresponds to the Z axis direction. A value sequence of the minimum dissimilarity D _m , _n of each comparison image is interpolated by a quadratic function, and a Z-axis coordinate having a minimum value of the quadratic function is compared with a slice image corresponding to an accuracy equal to or less than the interval of the slice image sequence. A method for automatically matching slice images of a chest three-dimensional tomographic image, the method comprising a step of determining a position. 3. A first plane having an image sequence in which a body section of a chest is a slice plane in X and Y axis directions and a body axis is in a Z axis direction.
One slice image is extracted from the file of the three-dimensional tomographic image of the above as a diagnostic image, and the second slice image is taken at another time.
A procedure for extracting one slice image as a comparison image from the file of the three-dimensional tomographic image described above, and as a modification to match the comparison image with the diagnosis image, a translation amount α in the X-axis direction in the slice plane direction and a Y-axis direction Translates parallel translation amount β, magnification γ in the X-axis direction,
A procedure of performing scaling by a magnification δ in the Y-axis direction and rotating and moving by an angle θ, a diagnostic image of a slice image number m, and a comparison image of a slice image number n deformed by the above-described process, by labeling of both images by labeling a step of extracting an aND region, a step of calculating an average value of the absolute value of the difference between the concentration of each pixel in the aND region of the comparison image and the diagnostic image as dissimilarity D _iff, the calculation of the dissimilarity D _iff A step of selecting the translation amounts α and β, the enlargement / reduction rates γ and δ, and the rotation angle θ; and the minimum difference D _m and _{n of the} difference D _Diff obtained in the selection process as the second three-dimensional determining a procedure for calculating each for each slice image of the tomographic image, as the minimum dissimilarity D _m, a slice image matching position comparative reading with the diagnostic image comparison image having the minimum value of _n of the each comparative image And forward, as a program causing a computer to execute a recording medium having recorded thereon a program slice image automatic collation method chest three-dimensional tomographic images in which the computer is recorded in a recording medium readable, it is characterized. 4. The method according to claim 3, wherein in the step of determining a comparison image having the minimum degree of difference D _m , _n as a slice image collation position, the body axis direction is a Z axis, and the Z axis direction corresponds to the Z axis direction. A value sequence of the minimum dissimilarity D _m , _n of each comparison image is interpolated by a quadratic function, and a Z-axis coordinate having a minimum value of the quadratic function is compared with a slice image corresponding to an accuracy equal to or less than the interval of the slice image sequence. A recording medium on which a program for a method for automatically checking a slice image of a chest three-dimensional tomographic image is recorded, the program having a procedure for determining a position.