JP4575821B2 - Image processing apparatus, image processing method and image diagnosis support system for visualizing peristaltic movement of uterus, computer program therefor, and recording medium recording the same - Google Patents

Description

  The present invention relates to an image processing apparatus, an image processing method and an image diagnosis support system for analyzing and visualizing peristaltic motion of a uterus based on a plurality of image data obtained by photographing a uterus according to a time series, and for the same The present invention relates to a computer program and a recording medium on which the computer program is recorded.

  In diagnostic imaging of gynecological areas, an ultrasound image, an X-ray image, and an MR (Magnetic Resonance) image are used. Ultrasonic images have blind spots, low resolution, and lack of tissue specificity, and X-ray imaging has the problem of X-ray exposure. On the other hand, diagnosis using MR images does not use X-rays, so it can be safely taken for pregnant patients, and functional and morphological changes can be achieved with excellent tissue contrast. It can draw remarkably.

Conventional MRI (Magnetic Resonance Imaging) does not capture uterine peristaltic movements such as repeated contraction movements such as the endometrium undulating and repeated movements of the muscle layer outside the endometrium, with a long imaging time. It was possible. Therefore, uterine peristalsis is not considered at all in diagnostic imaging, and the uterus has been treated as if it were a stationary organ.
However, in recent years, high-speed imaging methods such as HASTE (half-Fourier-acquisition single-shot turbo spin-echo) have been developed, and the time required to take MR images has been shortened. Dynamic imaging that directly draws peristaltic movement has become possible (Non-Patent Documents 1 and 2).

The female genitalia is affected by age, menstrual cycle and hormones. Therefore, unlike other organs, the shape of the uterus is not constant even when normal. That is, the uterus is considered to perform exercises having different properties during the ovulation period, menstrual period, and luteal phase. Therefore, it is expected that analysis of the uterine peristaltic movement will lead to the discovery of a treatment suitable for each patient's symptoms related to menstruation and fertility.
Takashi Ueda et al., “Anatomy and Physiological Functions Necessary for Imaging the Pelvis (Female)”, INNERVISION (18/9), 2003 Kaori Togashi "Gynecological imaging diagnosis by MRI", Journal of Japanese Society of Radiological Technology, Vol.59, No.8, August 2003, pp895-903

  However, the analysis of uterine peristalsis by current MR images relies on visual observation of raw MR images with the naked eye. However, the uterine peristalsis appearing in the raw MR image is only observed as a movement of a minute wavy projection appearing in the endometrium or a movement of a subtle shadow (low luminance part) in the myometrium. . More specifically, in the MR image obtained by photographing the pelvis, active movements of surrounding organs appear, and the shape of the uterus constantly changes under the influence of the movements of the surrounding organs. Yes. In an image of such a situation, the uterine peristalsis is only observed as a weak movement close to noise, and requires very careful observation by a skilled specialist.

Therefore, the evaluation of the uterine peristalsis movement based on the MR image must be subjective, and there is a problem that even a specialist doctor needs considerable skill for proper diagnosis.
Therefore, an object of the present invention is to make it possible to visualize the peristaltic motion of the uterus in an easy-to-understand manner, and thereby to support an image diagnosis of the uterus, an image processing method, an image diagnosis support system, and the same. It is to provide a computer program and a medium on which it is recorded.

  In order to achieve the above object, an invention according to claim 1 is an image processing apparatus for analyzing and visualizing a peristaltic movement of a uterus, wherein the period is shorter than a peristaltic period (for example, 2 to 3). A plurality of uterine MR image data corresponding to a plurality of uterine MR (magnetic resonance) images taken in time series at intervals of seconds) (particularly composed of luminance data of individual pixels. In particular, a so-called sagittal cross-sectional image) And a plurality of uterine MR image data stored in the image data storage means are subjected to Fourier transform in the time axis direction, and a plurality of frequencies for each pixel in the uterine MR image are stored. Fourier transform means for generating phase data of components (frequency components of luminance change over time) and phase data of at least one frequency component generated by the Fourier transform means are used. Then, a phase gradient in each part of the uterine MR image with respect to a predetermined direction of the frequency component (difference value of phase data with a pixel adjacent to the predetermined direction, a differential value) is calculated, and phase gradient data representing the phase gradient is generated. Gradient data generation means and phase gradient distribution image data representing a phase gradient distribution image in which each part of the image is colored in accordance with the phase gradient based on the phase gradient data generated by the phase gradient data generation means represents peristalsis of the uterus. An image processing apparatus for visualizing uterine peristalsis characterized by comprising phase gradient distribution image data generating means for generating image data (representing the direction and / or speed of peristalsis).

  Uterine peristalsis is roughly divided into cyclic movements of the wavy microprojections in the endometrium at an almost constant speed and periodic movements of local state change sites in the myometrium at an almost constant speed. The In general, during the ovulation phase, the peristaltic direction in the endometrium and myometrium is the direction from the cervix to the body for sperm transport. During menstruation, the direction of peristalsis in the endometrium and myometrium is the direction from the uterine body to the cervix for menstrual blood drainage. Furthermore, in the luteal phase (secretory phase), it is considered that the intima is not peristaltic because of the implantation of the pregnant egg and the retention of the initial pregnant egg.

  In the present invention, time-series Fourier transform is performed on a plurality of time-series uterine MR image data. Thereby, for each pixel constituting the uterine MR image, the temporal change of the pixel value (usually the luminance value) is decomposed into a plurality of frequency components, and phase data of the individual frequency components is generated. As described above, since the uterine peristalsis is a periodic motion of the endometrium and myometrium at a substantially constant speed, the uterine peristalsis is analyzed by focusing on the frequency component corresponding to this peristalsis. be able to.

  Therefore, in the present invention, phase data of at least one frequency component generated by Fourier transformation is used, and phase gradient data representing a phase gradient in a predetermined direction of the frequency component in each part of the uterine MR image is calculated. When a gradient (phase gradient) in a predetermined direction is calculated with respect to phase data of frequency components corresponding to the uterine peristalsis, the phase gradient represents the motion speed (speed and direction of peristalsis) of each part of the uterus. When a phase gradient distribution image is formed with color data corresponding to such phase gradient data as pixel values, the movement (direction and / or speed) of each part of the uterus can be grasped at a glance. In the present invention, phase gradient distribution image data representing such a phase gradient distribution image is generated.

Therefore, by displaying the phase gradient distribution image on the display device using the phase gradient distribution image data corresponding to the phase data of the frequency component corresponding to the uterine peristalsis, it is possible to provide an image in which the uterine peristalsis is clearly visualized. it can. Thereby, the image diagnosis regarding the uterus can be effectively supported.
The direction for obtaining the phase gradient may be designated by the operator, or may be set in advance in a predetermined direction.

  The image processing device uses the phase data of at least one frequency component generated by the Fourier transform means to generate phase distribution data (phase image data) representing the phase distribution of the frequency component of each part of the uterine MR image. Phase distribution data generation means (phase image data generation means) may be included. In this case, the phase gradient data generation means calculates the phase gradient of each part of the uterine MR image based on the phase distribution data (phase data of each pixel) generated by the phase distribution data generation means. Also good.

  According to the second aspect of the present invention, a plurality of uterine MR image data stored in the image data storage means are subjected to geometric transformation for compensating for variations in the position and shape of the uterine image portion. Geometric transformation means for generating uterine MR image data after the geometric transformation processing is further included, and the Fourier transformation means performs Fourier transformation on the uterine MR image data after the geometric transformation processing by the geometric transformation means. The image processing apparatus according to claim 1, wherein the image processing apparatus is an image processing apparatus.

  In an image taken by MRI, the position and shape of the uterus vary with time. In view of this, in the present invention, geometric transformation for matching the position and shape of the uterine image portion is performed on a plurality of uterine MR image data, and then Fourier transformation is performed. Thereby, the change of the image corresponding to the uterine peristalsis can be extracted more reliably.

  According to a third aspect of the present invention, the Fourier transform unit generates amplitude data of a plurality of frequency components for each pixel of the uterine MR image in addition to the phase data, and the Fourier transform unit generates the Fourier transform unit. Further comprising frequency component selection means for selecting at least one frequency component based on the amplitude data, wherein the phase gradient data generation means relates to the phase gradient data of each part of the uterine MR image with respect to the frequency component selected by the frequency component selection means. The image processing apparatus according to claim 1, wherein the image processing apparatus generates an image.

  In this configuration, phase data and amplitude data relating to a plurality of frequency components are generated for each pixel of the uterine MR image by Fourier transform. The movements found in the uterine MR image include not only uterine peristalsis but also noisy movements such as movement and deformation due to influences from surrounding organs. Some of the plurality of frequency components decomposed by the Fourier transform reflect such a noise-like motion.

  When attention is paid to the image region including the uterus, the frequency component corresponding to the uterine peristalsis has a larger amplitude than the frequency component corresponding to the noise-like motion. Thus, by paying attention to the amplitude data of individual frequency components obtained by Fourier transform, it is possible to select frequency components representing uterine peristalsis. Then, if the phase gradient data of each part of the uterine MR image is obtained with respect to the frequency component thus selected, and the phase gradient distribution image data representing the distribution is created, the pseudo color image corresponding to the phase gradient distribution image data is created. Can be displayed on the display device so that the uterine peristalsis can be easily visualized.

When the phase distribution data generation means (phase image data generation means) is provided, the phase distribution data generation means generates phase distribution data relating to the frequency component selected by the frequency component selection means. Preferably there is.
According to a fourth aspect of the present invention, the frequency component selecting unit extracts a maximum amplitude frequency component corresponding to the maximum amplitude data for each pixel of the uterine MR image, and the maximum amplitude frequency extracting unit. The maximum amplitude frequency distribution image data representing the distribution of the maximum amplitude frequency component extracted for each pixel in the uterine MR image (for example, image data having color data corresponding to the maximum frequency component as a pixel value) is generated. The image processing apparatus according to claim 3, further comprising a maximum amplitude frequency distribution image data generation unit.

  According to this configuration, if the maximum amplitude frequency distribution image (for example, a pseudo color image representing the distribution of the maximum frequency component) is displayed on the display device based on the maximum amplitude frequency distribution image data, the maximum in the uterine MR image can be obtained. The distribution of amplitude frequency components can be visually grasped. Based on this maximum amplitude frequency distribution image, the frequency component corresponding to the uterine peristalsis can be specified.

More specifically, as described in claim 5, the frequency component selection unit may further include a frequency component selection input reception unit that receives a frequency component selection input by an operator.
In the diagnostic imaging support system to which the image processing device is applied, for example, an input device that can be input by an operator, and a display that displays an image of the maximum amplitude frequency distribution image data generated by the maximum amplitude frequency distribution image data generation unit. And a device. The operator makes a judgment input from the maximum amplitude frequency distribution image displayed on the display device, and selects one of the specific frequency components from the input device. This input is received by the frequency component selection input receiving means. The frequency component is selected.

More specifically, if there is a frequency component in the maximum amplitude frequency distribution image that forms a shape that approximates part or all of the uterus in the uterine MR image, such frequency component is There is a high possibility that the corresponding frequency component. Therefore, the operator selects such frequency components and inputs them from the input device.
In medical imaging, it is often preferred to leave room for manual operation by an operator (typically a specialist) rather than a fully automated system. This is because the images used as the basis of diagnosis are a variety of images that depend on the individual patient, and also vary depending on the patient's condition, so automatic image analysis is not always successful. This is because it is possible to perform image diagnosis with high reliability and to obtain a sense of security with respect to the diagnosis result by providing a step for entrusting to this determination. Furthermore, by leaving some steps of the image analysis to the operator's judgment, there is an effect that the calculation load of the image processing apparatus is remarkably reduced.

However, it is not impossible to automatically select frequency components corresponding to uterine peristalsis, and it is possible to automate frequency component selection processing as necessary.
Specifically, as described in claim 6, the frequency component selecting unit further includes the maximum amplitude frequency distribution based on the maximum amplitude frequency distribution image data generated by the maximum amplitude frequency distribution image data generation unit. An area rank reference selection unit that selects at least one frequency component based on the rank of the area of each frequency component in the image corresponding to the distribution image data may be included.

  In the maximum amplitude frequency distribution image, there is a high possibility that the frequency component region corresponding to uterine peristalsis becomes dominant. Therefore, it is highly possible that the frequency component having the highest area (for example, only the first place, first place to third place) occupied in the maximum amplitude frequency distribution image is a frequency component corresponding to uterine peristalsis. Therefore, with the above-described configuration that automatically selects the frequency component having the higher occupied area in the maximum amplitude frequency distribution image, it is possible to appropriately select the frequency component corresponding to the uterine peristalsis without requiring judgment by the operator. .

This configuration is particularly effective when the uterine MR image to be processed after Fourier transformation is an image obtained by cutting out a region near the uterus.
In addition, as described in claim 7, when the frequency component selection unit selects a frequency component corresponding to maximum amplitude data equal to or greater than a predetermined threshold value, the frequency corresponding to uterine peristalsis is used. Automatic selection of ingredients is possible. That is, since the frequency component corresponding to the uterine peristalsis is considered to exhibit a certain level of amplitude, the frequency component corresponding to the maximum amplitude data less than the predetermined threshold value can be regarded as noise. Also, for frequency components that do not have an amplitude above a certain level, the analysis itself based on the phase data is not very meaningful, so the maximum amplitude data is used as the threshold value in order to eliminate meaningless image analysis processing. The frequency component that does not reach is preferably excluded from processing.

  In the invention according to claim 8, the Fourier transform means generates amplitude data of a plurality of frequency components for each pixel of the uterine MR image in addition to the phase data, and is generated by the Fourier transform means. And further comprising amplitude image data generating means for generating amplitude image data representing an amplitude image representing an amplitude distribution of the frequency component of each part of the uterine MR image using the amplitude data of at least one frequency component. Item 8. The image processing device according to any one of Items 1 to 7.

  According to this configuration, an amplitude image representing the amplitude distribution of individual frequency components is provided to the operator by performing image display on the display device based on the amplitude image data generated by the amplitude image data generation means. it can. For example, the amplitude image may be obtained by giving luminance corresponding to the amplitude data of the frequency component to each pixel. In the amplitude image of the frequency component corresponding to the uterine peristalsis, it is highly possible that pixels having a large amplitude form part or all of the shape of the uterus. Therefore, by providing such an amplitude image, the operator can easily specify and select the frequency component corresponding to the uterine peristalsis.

The invention according to claim 9 uses the phase data of at least one frequency component generated by the Fourier transform means to generate phase image data representing a phase image representing the phase distribution of the frequency component of each part of the uterine MR image. The image processing apparatus according to claim 1, further comprising phase image data generation means for performing the processing.
According to this configuration, by displaying an image on the display device based on the phase image data generated by the phase image data generation means, a phase image representing the phase distribution of each frequency component is provided to the operator. can do. For example, this phase image may be obtained by giving brightness corresponding to the phase of the frequency component of the uterine MR image to each pixel.

When a phase image as described above is created for the frequency component corresponding to the uterine peristalsis, an image portion showing a phase change appears in this phase image according to a part or all of the shape of the uterus. Therefore, providing the phase image can assist the operator in the process of specifying and selecting the frequency component corresponding to the uterine peristalsis.
When creating a phase image of a certain frequency component, it is preferable that the pixel value is fixed (for example, fixed to “0”) for a pixel whose amplitude of the frequency component is less than a predetermined threshold value. As a result, the image of the part that is considered to be completely unrelated to the uterine peristalsis can be crushed, and the determination by the operator becomes easier.

  According to the tenth aspect of the present invention, uterine MR image data (a representative one selected from a plurality of uterine MR image data or a plurality of uterine MR images stored in the image data storage means may be used. Image overlay processing for generating overlay image data by overlaying the phase gradient distribution image data generated by the phase gradient distribution image data generating means to image data obtained by taking an average value of image data, etc. 10. The image processing apparatus according to claim 1, further comprising means.

According to this configuration, when an image is displayed on the display device based on the overlay image data, a colored image representing the phase gradient distribution is displayed in an overlay manner on the uterine MR image. As a result, it is possible to provide an image that can grasp at a glance the correspondence between each part of the uterus and the site where the peristalsis occurs, and to easily determine what peristalsis occurs in which part. .
The invention according to claim 11 calculates the phase difference between the two pixels based on the phase data of at least two spaced apart pixels among the phase data of one frequency component generated by the Fourier transform means. The phase difference calculating means and the peristaltic speed calculating means for calculating the peristaltic speed of the uterus based on the phase difference calculated by the phase difference calculating means. This is an image processing apparatus.

  When the phase difference between the two pixels is obtained for the frequency component corresponding to the uterine peristalsis, the speed of the uterine peristalsis can be calculated based on this phase difference. By providing the calculated speed of uterine peristalsis, for example, by displaying it on a display device, it is possible to provide more information about uterine peristalsis and to support image diagnosis more effectively. More specifically, the speed of uterine peristalsis can be calculated based on the phase difference between two pixels and the distance between the two pixels. The distance between two pixels can be calculated from position information (coordinates) of the two pixels.

The invention according to claim 12 further includes means for receiving an instruction input at two points in the image by an operator, wherein the phase difference calculation means is the phase data (same frequency) of the two points of the pixel by the received instruction input. 12. The image processing apparatus according to claim 11, wherein a phase difference between the two pixels is calculated based on component phase data.
According to this configuration, since it is possible to instruct the two points where the speed is to be measured from the input device, it is possible to provide information on the peristaltic speed between any two points desired by the operator.

A thirteenth aspect of the invention includes the image processing apparatus according to any one of the first to twelfth aspects, and a display device connected to the image processing apparatus and displaying the phase gradient distribution image. This is an image diagnosis support system.
With this configuration, the phase gradient distribution image corresponding to the phase gradient distribution image data can be displayed on the display device, and this image information can be provided to the user. As a result, it is possible to provide an image that clearly visualizes the state of peristalsis of the uterus and can favorably support image diagnosis.

It is preferable that the display device can display the maximum amplitude frequency distribution image, the amplitude image, and the phase image.
The invention according to claim 14 further includes an input device operable by an operator, connected to the image processing device, and for inputting an instruction input from the operator to the image processing device. The image diagnosis support system according to claim 13.

With this configuration, it is possible to instruct the processing content of the image processing apparatus by an instruction input by the user.
The invention according to claim 15 is an image processing method for analyzing and visualizing peristaltic movement of the uterus, wherein a plurality of uterine MR (magnetic resonance) images taken in time series with a period shorter than the period of peristaltic movement of the uterus. ) A step of preparing a plurality of uterine MR image data corresponding to an image, and performing a Fourier transform in the time axis direction on the plurality of uterine MR image data, and a plurality of frequency components for each pixel in the uterine MR image A phase transformation of the frequency component in each part of the uterine MR image with respect to a predetermined direction is calculated using a Fourier transform step for generating the phase data of the uterine MR image, and phase data of at least one frequency component generated by the Fourier transform step. A phase gradient data generation step for generating phase gradient data representing the phase gradient, and the phase gradient data generation step A phase gradient distribution image data generation step for generating phase gradient distribution image data representing a phase gradient distribution image in which each part of the image is colored according to the phase gradient based on the generated phase gradient data as image data representing uterine peristalsis; and Is an image processing method for visualizing uterine peristalsis. By this method, the same effect as that described in relation to the invention of claim 1 can be obtained.

  According to a sixteenth aspect of the present invention, the phase difference between the two pixels is calculated based on the phase data of at least two spaced apart pixels among the phase data of one frequency component generated by the Fourier transform step. 16. The image processing method according to claim 15, further comprising: a phase difference calculating step; and a peristaltic speed calculating step for calculating a peristaltic speed based on the phase difference calculated in the phase difference calculating step. is there. By this method, the same effect as that of the 11th invention can be obtained.

Of course, the invention of the image processing method can be modified and improved similarly to the invention of the image processing apparatus.
According to the seventeenth aspect of the present invention, a plurality of uterine MR image data corresponding to a plurality of uterine MR (magnetic resonance) images photographed in time series in a cycle shorter than the cycle of uterine peristalsis is processed, and uterine peristalsis is performed. A computer program for operating a computer as an image processing device for analyzing and visualizing motion, the computer performing time-direction Fourier transform on the plurality of uterine MR image data, Fourier transform means for generating phase data of a plurality of frequency components for each pixel in the MR image, and using the phase data of at least one frequency component generated by the Fourier transform means, the frequency component in each part of the uterine MR image Phase gradient data for calculating a phase gradient for a given direction and generating phase gradient data representing the phase gradient Generation means, and phase gradient distribution image data representing a phase gradient distribution image obtained by coloring each part of the image according to the phase gradient based on the phase gradient data generated by the phase gradient data generation means, and image data representing uterine peristalsis It is a computer program for visualizing uterine peristalsis characterized by functioning as phase gradient distribution image data generating means for generating a By causing the computer program to be executed by a computer, the computer can function as the image processing apparatus according to claim 1.

  According to an eighteenth aspect of the present invention, the computer is further configured based on the phase data of at least two separated pixels among the phase data of one frequency component generated by the Fourier transform unit. 18. A phase difference calculating means for calculating a phase difference between the uterus and a peristaltic speed calculating means for calculating a peristaltic speed based on the phase difference calculated by the phase difference calculating means. It is a computer program. By causing the computer program to execute the computer program, the computer can be caused to function as the image processing apparatus according to claim 11.

Of course, the computer program may be modified or improved to cause the computer to function as the above-described various function processing means described in relation to the invention of the image processing apparatus.
The invention described in claim 19 is a recording medium that records the computer program as described above and is readable by a computer. Examples of such recording media include optical disks represented by CD-ROM and DVD-ROM, magnetic optical disks represented by MO disks, and magnetic disks represented by hard disk drives and flexible disks.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
[Basic principle of uterine peristalsis analysis]
First, the basic principle of uterine MR image analysis by the image processing apparatus according to an embodiment of the present invention will be described.
FIG. 1 shows an example of a uterine MR image (sagittal section). The uterus has a “body” shown in region a and a cervix shown in region b. The wall of the uterus is composed of three layers. The innermost layer is the intima c, and the muscle layer d exists so as to cover the intima, and the boundary between the intima c and the muscle layer d is a junctional zone.

  The uterine MR image shown in FIG. 1 is an example of an image taken by the fast spin echo method. In the spin echo method, it is possible to capture a T1-weighted image and a T2-weighted image. However, in the T1-weighted image, the myometrium and the intima have a uniform low signal, so that they cannot be distinguished and photographed. Therefore, in MRI of the uterus, imaging with a T2-weighted image is mainly performed, and the uterine MR image shown in FIG. 1 is also a T2-weighted image.

  In the analysis of uterine peristalsis, data of a plurality of uterine MR images taken in time series at a time interval (for example, every 2 seconds) shorter than the period of peristaltic uterus is used. The image size is, for example, 1.2 mm wide and 1.4 mm long. Since the luminance value of the first image is higher than that of the other images due to the nature of photographing, it is preferable to perform analysis using the images from the second image except for the first image. For the fast Fourier transform described later, a power of 2 (for example, 64) uterine MR images are required. Therefore, for example, when 60 uterine MR image data are to be processed and image analysis is performed on 59 uterine MR image data after the second, this is converted into 64 images by interpolation processing. It is converted into uterine MR image data for use.

  When multiple uterine MR images are displayed continuously on a display and carefully observed, minute wavy projections move along the inner lining of the uterus along the inner side at a substantially constant speed so that the uterine intima undulates. Observed in motion. Along with the movement of the minute wavy projections, the luminance value of each pixel constituting the image portion of the intima shows a periodic time change. In the muscle layer, it can be observed that something like a black shadow is moving. The muscular layer is composed of thick muscles, and is thought to be in periodic motion, similar to the intima.

  In the uterine MR image shown in FIG. 2, FIG. 3 shows an example in which changes in luminance values in the time axis direction are plotted for three points “1”, “2”, and “3” on the endometrium. The horizontal axis is a time axis represented by the numbers of a plurality of images according to time series. From FIG. 3, it can be inferred that the peristaltic motion of the endometrium can be expressed as a combination of periodic waves of changes in density (luminance) when focusing on a fixed point. The same applies to the peristaltic movement of the myometrium.

Therefore, Fourier transform processing is performed on the time-series data of each pixel of the uterine MR image (luminance data string of pixels at the same position in a plurality of uterine MR images taken according to the time series). Actually, as shown in FIG. 2, Fourier transform processing is performed using a uterine MR image trimmed into a rectangular region so as to exclude an image portion other than the uterine portion as a target image.
However, the position and shape of the image portion of the uterus fluctuate under the influence of the movement of surrounding organs. Therefore, in the Fourier transform in which the analysis is performed with the processing region fixed, if the position of the uterus changes, the target position of the motion analysis is shifted, and an appropriate analysis cannot be performed. Therefore, prior to Fourier transformation, geometric transformation processing is performed to compensate for uterine position and shape variations between a plurality of uterine MR images. Specifically, an affine transformation parameter that approximates deformation between images is obtained using nonlinear optimization, and the position and shape of the uterus are adjusted by affine transformation using the parameter. More specifically, after the translational movement component is first determined, an affine transformation parameter is obtained using the translational movement parameter as an initial value. The original image is deformed using the affine transformation parameters, an image sequence in which the corresponding region is cut out (trimmed) is created, and Fourier transformation is performed on the image sequence.

Fast Fourier transform (FFT), which is a form of discrete Fourier transform, is applied to the Fourier transform. Thereby, the amplitude and phase of main frequency components in the luminance change of each pixel are obtained. Based on this, the direction and speed of peristalsis and endometrial peristalsis can be analyzed.
More specifically, N time-series data (image data representing luminance values) for one pixel at the same position in N uterine MR images are represented by x ₀ , x ₁ , x ₂ ,. When _expressed as x _n ,..., x _N−1 , the discrete Fourier transform X _k for the frequency ω _k = 2πk / N (k = 0, 1, 2, 3,..., N−1) is given by Given in (1). However, j is an imaginary unit.

The amplitude data | C _k | and the phase data φ _k of the frequency component of the frequency ω _k are given by the following equations (2) and (3), respectively.
| C _k | = {X _kRe ² + X _kIm ² } ^1/2 (2)
φ _k = tan ⁻¹ (−X _kIm / X _kRe ) (3)
However, X _kRe and X _kIm are the real part and the imaginary part of X _k , respectively.

These amplitude data | C _k | and phase data φ _k are determined for the frequency components of all frequencies ω _k that can be calculated for individual pixels of the uterine MR image.
In order to perform the fast Fourier transform, as described above, power-of-two time series data is required. Therefore, when there are 59 effective uterine MR images, data obtained by interpolating luminance values of individual pixels of 59 uterine MR images (images cut out after affine transformation) from 59 to 64 is created in advance. Then, a fast Fourier transform is performed on the 64 pieces of data.

Based on the amplitude data obtained by the fast Fourier transform, a frequency having a maximum amplitude other than the frequency “0” (DC component) is obtained for each pixel. Then, a maximum amplitude frequency distribution image representing the frequency component of the maximum amplitude of each pixel is displayed on the display.
Specifically, as shown in FIG. 4A, different colors are assigned to a plurality of frequency components in advance, and each pixel is colored according to the frequency component having the maximum amplitude of the pixel. Thus, the frequency component distribution with the maximum amplitude is displayed in pseudo color on the display.

  Since the peristaltic motion of the uterus appears in the low frequency component, the example of FIG. 4A shows an example in which a distribution image for the frequency component of the first frequency to the tenth frequency (k = 1 to 10) is obtained. Yes. The first frequency is a frequency component having one cycle with 64 sheets, and the second frequency is a frequency component having one cycle with 32 (= 64/2) sheets. The same applies to frequency components equal to or higher than the third frequency component.

For reference, FIG. 4 (b) shows an image in which the color portion of the first frequency in the color image of FIG. 4 (a) is emphasized, and FIG. 4 (c) shows the color image of FIG. 4 (a). 4D shows an image in which the color portion of the second frequency is emphasized, and FIG. 4D shows an image in which the color portion of the sixth frequency in the color image in FIG. 4A is emphasized.
Since the peristalsis of the intima part or the muscle layer part is a regular periodic motion, it is expected that the pixel values (frequency components having the maximum amplitude) of the intima part or the muscle layer part are the same. Actually, in the example of FIG. 4, a plurality of pixels having the sixth frequency component as the maximum amplitude frequency component form an image portion having a shape approximating the endometrium.

Accordingly, by viewing the maximum amplitude frequency distribution image, the user (operator, generally a specialist) can easily specify the frequency component corresponding to the uterine peristalsis.
Amplitude image data and phase image data are created as necessary for the purpose of assisting in identifying frequency components corresponding to uterine peristalsis, and the corresponding amplitude image and phase image are respectively displayed on the display.

  The amplitude image data is amplitude distribution data in which the amplitude data of each pixel is a pixel value (luminance value) for one frequency component. An amplitude image corresponding to such amplitude image data is displayed brighter as a pixel has a larger amplitude of the frequency component. As described above, the uterine peristalsis is a regular periodic motion, so the amplitude image of the frequency component corresponding to the uterine peristalsis forms an image part with a shape that approximates the intima or muscle layer part of the uterus. It is expected. Therefore, it is possible to specify a frequency component that forms an image portion that approximates the shape of the uterus in the amplitude image as corresponding to uterine peristalsis. An example of an amplitude image is shown in FIG.

  The phase image data is phase distribution data in which the phase data of each pixel is a pixel value (luminance) for one frequency component. However, the pixel value is fixed to “0” for a pixel whose amplitude data for the frequency component is less than a predetermined threshold value. Such phase image data is displayed brighter as the pixel has a larger phase. Since the pixel value of the pixel below the threshold value is set to “0”, it can be expected that an image portion corresponding to the shape of the uterus is formed in the phase image of the frequency component corresponding to the uterine peristalsis, Furthermore, this image portion is expected to show a luminance change corresponding to the uterine peristaltic direction and speed. Therefore, the frequency component in which the image portion having such a characteristic is seen in the phase image can be specified as corresponding to uterine peristalsis. An example of a phase image is shown in FIG.

  When the frequency component corresponding to uterine peristalsis is specified in this way, the density gradient (luminance gradient) of the phase image is calculated. That is, for each pixel, the difference (density gradient) of the pixel value from the adjacent pixel is obtained in a predetermined direction. This difference is nothing but phase gradient data representing the phase gradient. The direction for obtaining the phase gradient is specified on the phase image displayed on the display by the operator, for example.

  Furthermore, color information is assigned according to the value of the phase gradient data. For example, if the phase gradient data is positive, a first color (for example, blue) is assigned to the pixel, and if the phase gradient data is negative, a second color (for example, red) that is different from the first color. Is assigned to the pixel. Thereby, a phase gradient distribution image in which the direction of uterine peristalsis is represented by color information is formed. Examples of phase gradient distribution images are shown in FIG. 22, FIG. 28 and FIG.

Of course, the pixel value (luminance) may be varied according to the value of the phase gradient. Thereby, the speed of uterine peristalsis can be displayed together.
The speed of uterine peristalsis can be determined by specifying any two points in the image. Speed = distance ÷ time, time = period × phase difference ÷ 2π. “Period” is a period of one cycle, and is represented by a period = number of images × 2 (seconds) ÷ frequency. The frequency is the frequency of the frequency component specified to correspond to uterine peristalsis.

Therefore, if the phase difference is calculated using the two specified phase data for the frequency component specified to correspond to the uterine peristalsis, the speed of the uterine peristalsis can be obtained based on this phase difference.
[Configuration of diagnostic imaging support system]
FIG. 5 is a block diagram for explaining a basic hardware configuration of the diagnostic imaging support system according to the embodiment of the present invention. This diagnostic imaging support system has a basic configuration as a computer system. That is, the diagnostic imaging support system includes a computer main body 10, a keyboard 11A and a mouse (an example of a pointing device) 11B as an input device 11 connected to the computer main body 10, and a display ( Display device) 13 and a printer 14 connected to the computer main body 10.

  The computer main body 10 includes a CPU 15 as an arithmetic processing unit, a ROM 16 and a RAM 17 as main storage devices, a hard disk drive (HDD) 18 and a CD-ROM drive 19 as auxiliary storage devices, an input control unit 20, and a display. An interface 21 and a printer interface 22 are provided, and these are connected to each other by a bus 23. The input control unit 20 processes an input signal from the input device 11 and inputs it to the CPU 15. The display interface 21 has a function of receiving input of image data and displaying an image corresponding to the image data on the display 13. The printer interface 22 is for connection with the printer 14.

  The CD-ROM drive 19 can be loaded with a CD-ROM (disc) 25 as a recording medium on which a computer program is recorded. The computer main body 10 is loaded with the CD-ROM 25 on which the image processing program is recorded and the computer main body 10 executes the image processing program after performing a necessary installation operation. The computer system can function as an image diagnosis support system.

  Of course, the computer program need not be provided in a form recorded on a recording medium, and may be provided via a network. That is, the computer main body 10 is provided with a network interface, and the computer main body 10 can also function as an image processing apparatus by downloading a computer program to the computer main body 10 via the network interface.

  On the other hand, the uterine MR image data to be processed is stored in the hard disk drive 18, for example. This uterine MR image data may also be stored in the hard disk drive 18 via the CD-ROM drive 19, or may be acquired via a network and stored in the hard disk drive 18. The uterine MR image data is data representing a plurality of uterine MR images taken in time series at a time interval (for example, 2 seconds) shorter than the cycle of the uterine peristalsis.

  FIG. 6 is a block diagram for explaining a functional configuration of the computer main body 10 functioning as the image processing apparatus. The computer main body 10 substantially has the function processing unit shown in FIG. 6 by causing the CPU 15 to execute the image processing program. That is, the computer main body 10 includes an image processing calculation unit 30, a data storage unit 50, and an image display control unit 70. The image processing calculation unit 30 and the image display control unit 70 are function processing units realized mainly by calculation processing by the CPU 15. The data storage unit 50 is composed of storage areas of the RAM 17 and the hard disk drive 18.

  The image processing calculation unit 30 includes an affine transformation processing unit 31, a trimming processing unit 32, a Fourier transformation processing unit 33, a maximum amplitude frequency component extraction processing unit 34, a maximum amplitude frequency distribution image data generation processing unit 35, and a frequency. Component selection input reception unit 36, amplitude image data generation processing unit 37, phase image data generation processing unit 38, phase gradient calculation direction input reception unit 39, phase gradient data generation processing unit 40, and phase gradient distribution image data A generation processing unit 41, an image overlay processing unit 42, a speed measurement point input receiving unit 43, a phase difference calculation unit 44, and a peristaltic speed calculation unit 45 are provided.

  Further, the data storage unit 50 includes an atomic-miya MR image data storage unit 51, a post-affine transformation uterine MR image data storage unit 52, a target uterine MR image data storage unit 53, a Fourier transform data storage unit 54, and a maximum amplitude. Frequency component data storage unit 55, maximum amplitude frequency distribution image data storage unit 56, amplitude image data storage unit 57, phase image data storage unit 58, phase gradient data storage unit 59, and phase gradient distribution image data storage unit 60, an overlay image data storage unit 61, a phase difference data storage unit 62, and a peristaltic speed data storage unit 63.

The atomic-miya MR image data storage unit 51 stores a plurality of uterine MR images (original data. However, if necessary, the images are taken in time series by MRI at a time interval (for example, 2 seconds) shorter than the cycle of the uterine peristalsis. The result of interpolation processing is stored.
The affine transformation processing unit 31 reads the uterine MR image data from the atomic-miya MR image data storage unit 51, performs affine transformation processing on the uterine MR image data, and converts the uterine MR image data after affine transformation into the uterine MR image after affine transformation. The data is stored in the data storage unit 52.

The trimming processing unit 32 reads the uterine MR image data after the affine transformation processing from the uterine MR image data storage unit 52 after the affine transformation, performs a trimming processing for excluding images other than the uterine portion from the uterine MR image data, and performs this processing. The later trimmed uterine MR image data is stored in the target uterine MR image data storage unit 53.
The Fourier transform processing unit 33 reads the uterine MR image data processed by the trimming processing unit 32 from the target uterine MR image data storage unit 53, and performs Fourier transform (fast Fourier transform) on the uterine MR image data. More specifically, the Fourier transform processing unit 33 performs Fourier transform in the time axis direction on a plurality of uterine MR image data trimmed through affine transformation. Thereby, amplitude data and phase data are obtained for a plurality of frequency components for each pixel of the uterine MR image. That is, the temporal change of the luminance value of each pixel of the uterine MR image is decomposed into a plurality of frequency components, and amplitude data representing the amplitude and phase data representing the phase are obtained for each frequency component. These data are stored in the Fourier transform data storage unit 54.

The maximum amplitude frequency component extraction processing unit 34, based on the amplitude data for a plurality of frequency components of each pixel stored in the Fourier transform data storage unit 54, the frequency having the maximum amplitude data for each pixel of the uterine MR image. Extract ingredients. Thus, the data representing the frequency component extracted for each pixel is stored in the maximum amplitude frequency component data storage unit 55.
The maximum amplitude frequency distribution image data generation processing unit 35 represents an image (maximum amplitude frequency distribution image) representing the distribution of the extracted maximum amplitude frequency component based on the data stored in the maximum amplitude frequency component data storage unit 55. Maximum amplitude frequency distribution image data is generated and stored in the maximum amplitude frequency distribution image data storage unit 56. Specifically, the maximum amplitude frequency distribution image data generation processing unit 35 assigns different color data to a plurality of frequency components, and generates maximum amplitude frequency distribution image data using the color data as pixel values. The image display control unit 70 reads the maximum amplitude frequency distribution image data and displays the maximum amplitude frequency distribution image on the display 13 (see FIG. 5). As a result, a pseudo color display image representing the distribution of the maximum amplitude frequency component in the uterine MR image is displayed on the display 13.

  The frequency component selection input receiving unit 36 receives a frequency component selection input from the input device 11. The operator determines from the maximum amplitude frequency distribution image displayed on the display 13, specifies the frequency component corresponding to the uterine peristalsis, and inputs an instruction to select this frequency component via the input device 11. This selection instruction is received by the frequency component selection input receiving unit 36.

  The amplitude image data generation processing unit 37 generates amplitude image data for displaying an amplitude image for the frequency component corresponding to the selection instruction received by the frequency component selection input receiving unit 36. Specifically, the amplitude image data generation processing unit 37 reads the amplitude data of each pixel for the selected frequency component from the Fourier transform data storage unit 54, and uses the amplitude data as a pixel value ( Amplitude distribution data) is created, and this amplitude image data is stored in the amplitude image data storage unit 57. The image display control unit 70 reads the amplitude image data and performs display control of the display 13. Thereby, an amplitude image using the amplitude data for the selected frequency component as the luminance value of each pixel is displayed on the display 13.

  On the other hand, the phase image data generation processing unit 38 generates phase image data for displaying the phase image for the frequency component corresponding to the selection instruction received by the frequency component selection input receiving unit 36. Specifically, the phase image data generation processing unit 38 reads out the phase data of each pixel for the selected frequency component from the Fourier transform data storage unit 54, and uses the phase data as a pixel value. Phase distribution data) is created, and this phase image data is stored in the phase image data storage unit 58. However, the phase image data generation processing unit 38 refers to the amplitude image data stored in the amplitude image data storage unit 57, and the value thereof (that is, the value of the amplitude data) does not reach a predetermined threshold value. The phase image data is fixed to “0”. The image display control unit 70 reads the phase image data from the phase image data storage unit 58 and performs display control of the display 13. As a result, for the pixels other than the pixels whose phase image data is fixed to “0”, a phase image having the phase data for the selected frequency component as the luminance value of each pixel is displayed on the display 13. Become.

The phase gradient calculation direction input receiving unit 39 receives a phase gradient calculation direction instruction input from the input device 11. The operator determines the direction of uterine peristalsis from the amplitude image and / or phase image displayed on the display 13 and instructs the direction from the input device 11. This instruction input is received by the phase gradient calculation direction input receiving unit 39.
The phase gradient data generation processing unit 40 reads the phase image data stored in the phase image data storage unit 58 and is adjacent to the phase gradient calculation direction received by the phase gradient calculation direction input receiving unit 39 for each pixel. A difference value (differential value) from the pixel is obtained, and the difference value is generated as phase gradient data and stored in the phase gradient data storage unit 59. The phase gradient data thus represents the speed (direction and speed) of movement in each part of the uterine MR image. When phase image data of frequency components corresponding to uterine peristalsis is stored in the phase image data storage unit 58, the phase gradient data represents the peristaltic speed (direction and speed) of each part of the uterus.

  The phase gradient distribution image data generation processing unit 41 reads the phase gradient data from the phase gradient data storage unit 59 and generates phase gradient distribution image data for displaying the phase gradient distribution in color. Specifically, the phase gradient distribution image data generation processing unit 41 assigns different color data in advance for each value of the phase gradient data, and sets the image data having such color data as the pixel value as the phase gradient distribution image data. And stored in the phase gradient distribution image data storage unit 60. The image display control unit 70 reads the phase gradient distribution image data from the phase gradient distribution image data storage unit 60 and executes display control processing of the display 13. As a result, a phase gradient distribution image in which the phase gradient distribution is displayed in a pseudo color is displayed on the display 13.

  For example, if the value of the phase gradient data is positive, the phase gradient distribution image data generation processing unit 41 sets the first color data (for example, blue data) as the phase gradient distribution image data, and the value of the phase gradient data is negative. For example, second color data (for example, red data) different from the first color data may be used as the phase gradient distribution image data. In this case, if the direction from the cervix to the uterine body (or the opposite direction) is designated as the phase gradient calculation direction from the input device 11, the peristalsis from the cervix to the body and the body to the neck Percussion can be displayed in different colors.

  The phase gradient distribution image data generation processing unit 41 assigns the first color data (for example, blue data) if the value of the phase gradient data is positive, and the first color if the value of the phase gradient data is negative. While assigning second color data (for example, red data) different from the data, phase gradient distribution image data may be generated in which the absolute value of the phase gradient data is a luminance value (pixel value). If it does in this way, while the direction of the peristaltic of each part can be expressed in different colors, the speed of peristaltic movement of each part can be expressed by luminance.

  The image overlay processing unit 42 reads the uterine MR image data (for example, one image used as a reference in the affine transformation process) from the atomic-miya MR image data storage unit 51, and the phase gradient distribution image data storage unit 60 reads the phase. The gradient distribution image data is read out, overlay image data obtained by overlaying these is generated, and stored in the overlay image data storage unit 61. That is, the image overlay processing unit 42 superimposes the phase gradient distribution image data on the region corresponding to the rectangular region cut out by the trimming processing unit 32 in the uterine MR image data. The image display control unit 70 reads the overlay image data from the overlay image data storage unit 61 and displays the image on the display 13. As a result, the phase gradient distribution image is superimposed and displayed on the uterine MR image (original image), so that the state of the uterine peristalsis can be grasped at a glance together with the position thereof.

Of course, the image display control unit 70 can also cause the display 13 to display an image of the image data stored in the phase gradient distribution image data storage unit 60. At this time, only the phase gradient distribution image is displayed on the display 13.
The speed measurement point input receiving unit 43 receives input of two points in the image from the input device 11 in a state where the phase image, the phase gradient distribution image, or the overlay image is displayed on the display 13. This input is accepted as a speed measurement point at which the speed between the two points is to be measured.

The phase difference calculation unit 44 reads the phase data of the two pixels received by the velocity measurement point input reception unit 43 from the phase image data storage unit 58, calculates the difference between them as phase difference data, and outputs the phase difference data. Store in the storage unit 62.
The peristaltic speed calculation unit 45 is based on the distance between the two speed measurement points received by the speed measurement point input receiving unit 43 and the phase difference data stored in the phase difference data storage unit 62. The speed is obtained as a peristaltic speed, and peristaltic speed data representing this peristaltic speed is stored in the peristaltic speed data storage unit 63.

The image display control unit 70 displays the peristaltic speed data stored in the peristaltic speed data storage unit 63 on the display 13 (for example, numerical display).
FIG. 7 is a flowchart collectively showing a processing flow for processing uterine MR image data to visualize uterine peristalsis. A plurality of uterine MR image data obtained by MRI is subjected to an affine transformation process, which is a geometric transformation for compensating for variations in the position and shape of the uterus between images (step S1). Trimming is performed to cut out the partial image (step S2). Next, the following processing is performed by using a plurality of trimmed uterine MR image data as target image data.

  That is, Fourier transform processing (FFT) in the time axis direction is performed on a plurality of uterine MR image data, and amplitude data and phase data relating to a plurality of frequency components are generated for each pixel (step S3). . Next, for each pixel, a frequency component having the maximum amplitude data (maximum amplitude frequency component) is extracted (step S4), and a maximum color image that is a pseudo color image with color data corresponding to the maximum amplitude frequency component as a pixel value. Amplitude frequency distribution image data is created, and an image corresponding to this (maximum amplitude frequency distribution image) is displayed on the display 13 (step S5).

An operator (typically a specialist who performs image diagnosis) selects a frequency component that is considered to correspond to uterine peristalsis based on the maximum amplitude frequency distribution image displayed on the display 13, and inputs a selection instruction to the input device. 11 (step S6).
As a result, amplitude image data and phase image data of the selected frequency component are created, and amplitude images and phase images corresponding to these are displayed on the display 13 (steps S7 and S8). The operator determines whether or not the selected frequency component corresponds to uterine peristalsis from the displayed amplitude image and phase image (step S9). When it is considered that the selected frequency reflects uterine peristalsis, the operator instructs the execution of the subsequent processing from the input device 11 (step S10). Thereby, the process after step S11 is performed. On the other hand, when it is considered that the selected frequency component does not reflect uterine peristalsis, the operator inputs an instruction to select another frequency component from the input device 11 (step S6). Thereby, amplitude image data and phase image data are created for the newly selected other frequency component, and the corresponding amplitude image and phase image are respectively displayed on the display 13 (steps S7 and S8).

  In step S11, the operator designates the calculation direction of the phase gradient by designating, from the input device 11, two points on the phase image of the frequency component determined to reflect the uterine peristalsis. Thereby, the brightness | luminance gradient between the pixels in a phase image is calculated | required as phase gradient data along the said direction (step S12). Further, phase gradient distribution image data having two color data corresponding to the sign of the phase gradient data as pixel values is created, and the corresponding phase gradient distribution image is displayed on the display 13 (step S13). Further, as necessary, the user instructs the input device 11 to display an overlay image of the uterine MR image data and the phase image distribution image (step S14). In response to this, overlay image data is created by overlaying both images, and the corresponding overlay image is displayed on the display 13 (step S15).

When obtaining the speed of uterine peristalsis, the operator can select two points in the image from the input device 11 in either the phase image (step S8), the phase gradient distribution image (step S13), or the overlay image (step S15). Is input as a speed measurement point (step S16). In response to this, a difference (phase difference) in phase data between the two input velocity measurement points is obtained (step S17), and the velocity of the peristaltic motion is obtained using this phase difference (step S18). Then, the calculated speed of the peristaltic movement is displayed on the display 13 (step S19).
[Simulation experiment]
An image in which intimal motion was artificially added to the uterine MR image was created, and the above-described image processing was performed on the image.

  FIG. 8 is an image example for explaining 64 time-series uterine MR images created for a simulation experiment. A black dot is added to the position A on one target image (uterine MR image) to form the first uterine MR image, and the second and subsequent images are sequentially directed to the position B on the same target image. A black dot was also added while shifting the position, and the black dot reached position B on the eighth sheet. Similarly, for the ninth and subsequent images, the same target image was used, and black dots were added to the image so that eight images moved from position A to position B. In this way, 64 uterine MR image data were created by adding simulated peristaltic motion, which is one cycle of 8 images.

  FIG. 9 shows the temporal change in the luminance values of the pixels at positions A and B. The horizontal axis represents time, but is represented by the image numbers from the first image to the 64th image. From this figure, it can be observed that the luminance values of the pixels at positions A and B change as the black dots move. Since the black dots are located at the positions A and B at different times, the two curves (broken lines) corresponding to the pixels at the positions A and B are shifted in the time axis direction.

  In the example of FIG. 8, the coordinates of the position A are (119, 87), and the coordinates of the position B are (91, 111). In this case, the distance between the two points is 44.65 (mm). That is, since 44.65 (mm) is moved at (8-1) × 2 (sec), the speed of the motion created as simulation data here is 44.65 (mm) / 14 (sec) = 3. 19 (mm / sec).

  The 64 uterine MR image data as described above were subjected to Fourier transform in the time axis direction to obtain the maximum amplitude frequency component of each pixel, and a maximum amplitude frequency distribution image was created. This is shown in FIG. 10 ((a) is a color image, and (b) is an image in which the color portion of the eighth frequency of the color image of (a) is emphasized). From this maximum frequency distribution image, it can be seen that the eighth frequency component (= 64/8) represents motion. An amplitude image and a phase image for the eighth frequency component are shown in FIGS. 11 and 12, respectively.

  In the amplitude image (FIG. 11), the place where the brightness is high is the place where the black point has moved in the artificial image (64 uterine MR images with black points added). In the phase image (FIG. 12), the pixel value of the pixel having an amplitude less than the predetermined threshold is fixed to “0”, so that it is significant in the same region as the region having high brightness in the amplitude image of FIG. A pixel having a pixel value appears. By using the pixel value (phase value) of this phase image, the motion speed (direction and speed) between the two points can be analyzed. That is, in this case, it can be detected as a movement from the high-luminance position A toward the low-luminance position B.

  Position A and position B in the phase image are the start and end points of the motion when the artificial image is created. The speed of movement is obtained from the distance between these two points and the phase difference. The distance between the position (119, 87) and the position B (91, 108) was 44.65 (mm), and the phases were 3.14 (rad) and -2.36 (rad), respectively. Therefore, the phase difference between the pixels at positions A and B is 5.50 (rad). Therefore, the speed of movement was 3.19 (mm / sec), and a result equal to the set speed was obtained.

The direction in which the phase gradient is obtained is designated as the direction from position A to position B, and the luminance value difference (phase gradient) with the adjacent pixel is obtained for each pixel, and the positive phase gradient (from position A to position B is obtained). Corresponding to the case where the phase increases toward blue), the blue color data is related to the negative phase gradient (corresponding to the case where the phase increases from position B toward position A). When a phase gradient distribution image is created that associates and uses these color data as pixel values, the image shown in FIG. 13 is obtained. That is, an image in which the region between the positions A and B is colored blue is obtained.
[Experiment 1 (image during ovulation)]
The target image shown in FIG. 14 (64 uterine MR images. FIG. 14 shows a representative one of them) is a ovulation-phase uterus of a subject taken by MRI. The image processing of this embodiment is applied to this target image.

FIG. 15A shows a maximum amplitude frequency distribution image (color image). FIG. 15B shows an image in which the color part of the third frequency in the color image of FIG. 15A is emphasized, and FIG. 15C shows the sixth frequency in the color image of FIG. 15A. FIG. 15D shows an image in which the color portion of the seventh frequency in the color image in FIG. 15A is emphasized.
From the image of FIG. 15 (a), it can be seen that the frequency of endometrial motion is not constant, ie, there are several different frequencies of motion in the intima. In this example, an appropriate frequency component (frequency component corresponding to peristalsis of the intima) cannot be specified from the maximum amplitude frequency distribution image. Therefore, amplitude images and phase images were created for several frequency components. As a result, the amplitude image and the phase image of the seventh frequency component were as shown in FIGS. 16 and 17, respectively, and an endometrial image portion was formed. Therefore, it was determined that the seventh frequency component corresponds to uterine peristalsis.

In FIG. 17, since the intima portion is gradually darkened from the cervix of the uterus toward the body, it can be seen that the intima movement direction of the target image uterus is the direction from the cervix toward the body.
Changes in the luminance values of the four points a, b, c, and d in FIG. 17 in the time axis direction are shown in FIGS. FIG. 18 shows the original data (data before Fourier transformation), and FIG. 19 shows the inverse Fourier transformation applied to the DC component and the seventh frequency component. As can be seen from FIG. 19, the luminance value changes at the four points a, b, c, and d are shifted in the time axis direction. This shift in the time axis direction appears as a change in luminance value in the phase image of FIG.

During the ovulation period, it is thought that the intima of the uterus is moving from the cervix to the body due to the transport of sperm, and this was confirmed in this experiment using images of the uterus during the ovulation period. .
The velocity of the upper part of the intima was obtained from the phase of points a and d in FIG. Since the coordinates of the point a are (114, 78) and the coordinates of the point d are (75, 91), the distance between the two points is 50.21 (mm). Since the phases were 1.89 (rad) and -1.57 (rad), respectively, the phase difference was 3.46 (rad), and thus the speed was 5.32 (mm / sec). The velocity of the lower intima was determined from points 6 and 7 in the image of FIG. Since the coordinates of the point 6 are (109, 94) and the coordinates of the point 7 are (89, 109), the distance between the two points is 31.89 (mm). Since the phases were 1.09 (rad) and -1.71 (rad), respectively, the phase difference was 2.80 (rad). Therefore, the speed is 4.17 (mm / sec).

In order to evaluate whether the motion analyzed by this method captures the same motion that can be captured visually when the image sequence is displayed continuously, the motion speed and phase that can be observed in the image are used. The speed was determined.
FIG. 20 shows the 28th uterine MR image of the image sequence of the target image of the experiment, and FIG. 21 shows the 37th uterine MR image of the image sequence. In the target image of this experiment, a black shadow (portion indicated by an arrow in the drawing) between the 28th image shown in FIG. 20 and the 37th image shown in FIG. 73) to point D (57,103), 89.65 (mm), and in the lower part of the intima, it moved 69.67 mm from point E (119,87) to point F (71,115). Since the elapsed time is 18 (sec) (= 2 sec × 9), the motion speed of the upper and lower intima was 4.98 (mm / sec) and 3.87 (mm / sec), respectively. As shown in Table 1 below, the speed obtained by this method and the speed of movement captured in the image are almost close to each other, and the result analyzed by this method and the result detected visually are as follows: It is thought that it captures an equivalent movement.

Therefore, it was confirmed that this technique is an effective technique that can obtain the peristaltic movement of the uterus, which is generally difficult to measure manually, by image analysis.
The direction in which the phase gradient is obtained is designated as the direction from point a to point d in FIG. 17, and a difference in luminance value (phase gradient) from adjacent pixels is obtained for each pixel, and a positive phase gradient (from point a to point d) is obtained. Corresponding to the case where the phase increases toward d), the blue color data is associated, and the red color corresponding to the negative phase gradient (corresponding to the case where the phase increases from point d to point a). FIG. 22 (a) is a color image, and (b) is a diagram in which the blue portion of the color image in (a) is emphasized when data is associated and a phase gradient distribution image having these color data as pixel values is created. In the lower right arrow, an image shown in the lower left is a blue arrow, and an upper right arrow is a red arrow. That is, an image in which the portion corresponding to the endometrium is colored in blue is obtained, and the peristaltic motion from the cervix to the body is clearly visualized.

By overlaying this color image on the original data, the overlay image shown in FIG. 23 is obtained ((a) is a color image, (b) is an image in which the blue area of the color image of (a) is highlighted in white) ).
[Experiment 2 (Menstrual period images)
The target images shown in FIG. 24 (64 uterine MR images. FIG. 24 shows a representative one of them) are images of the same menstrual period as in Experiment 1. The method of the above embodiment was applied to this image, and the difference between the ovulation and menstrual movements was verified.

FIG. 25 (a) shows a maximum amplitude frequency distribution image. For reference, FIG. 25 (b) shows an image in which the color portion of the second frequency in the color image of FIG. 25 (a) is emphasized, and FIG. 25 (c) shows the image in the color image of FIG. 25 (a). The image which emphasized the color part of the 3rd frequency of is shown.
From the image of FIG. 25 (a), it was determined that the frequency of the main motion was the second frequency component. An amplitude image and a phase image created using the amplitude data and phase data of the second frequency component are shown in FIGS. 26 and 27, respectively.

In FIG. 27, it can be observed that the intima portion is gradually darkening from the body part of the uterus toward the neck part. Therefore, it can be determined that the direction of movement is the direction from the body part to the neck part, opposite to the ovulation period. The menstrual uterus is thought to move from the body to the cervix due to the discharge of menstrual blood, which was confirmed in this experiment.
The speed of motion of the upper intima was determined from the phase of the pixels at points 9 and 8 in FIG. 27, and the speed of motion of the lower intima was determined from the phase of points h and e in FIG. Table 2 shows the obtained speed.

  The direction in which the phase gradient is obtained is designated as the direction from point 8 to point 9 in FIG. 27, and the difference in luminance value (phase gradient) from adjacent pixels is obtained for each pixel, and a positive phase gradient (from point 8 to point 9) is obtained. (Corresponding to the case where the phase increases toward 9) and the blue color data are associated with each other, and the red color corresponding to the negative phase gradient (corresponding to the case where the phase increases from point 9 to point 8). When a phase gradient distribution image is created by associating data and using these color data as pixel values, a color image shown in FIG. 28 (a) is obtained ((b) represents a red portion in the color image of (a)). (Highlighted image: The arrow at the lower right is the blue arrow toward the lower left and the red arrow toward the upper right.) That is, an image in which a portion corresponding to the endometrium is colored in red is obtained, and the peristaltic motion from the uterine body part to the neck part is clearly visualized.

By overlaying this color image on the original data, the overlay image shown in FIG. 29 is obtained ((a) is a color image, (b) is an image in which the red area of the color image of (a) is highlighted in white. ).
[Experiment 3 (Luteal phase image)
The target image (64 uterine MR images. FIG. 30 shows only one representative image) shown in FIG. 30 is an image of a luteal phase uterus of a subject. The motion of the intima in this image was analyzed to examine how the endometrium in the luteal phase differs from the ovulatory and menstrual phases.

FIG. 31A shows a maximum amplitude frequency distribution image of the target image. For reference, FIG. 31 (b) shows an image in which the color portion of the first frequency in the color image of FIG. 31 (a) is emphasized, and FIG. 31 (c) shows the color image of FIG. 31 (a). The image which emphasized the color part of the 2nd frequency of is shown.
From the image of FIG. 31A, it is determined that the frequency component of the main motion (the frequency component having the maximum amplitude in each part) is the first frequency component, and based on the amplitude data and the phase data, the amplitude image ( 32) and a phase image (FIG. 33) were created.

In the amplitude image of FIG. 32, a place with high brightness is not clearly present in the intima. FIG. 33 also shows that there is no phase change due to motion because the luminance value of the intima portion is constant.
FIG. 34 is the original data (change in luminance value with time) of the four points q, r, s, and t in FIG. 30, and FIG. 35 is the frequency of the points q, r, s, and t obtained by Fourier transform. It is a spectrum (amplitude data of each frequency component). From FIG. 34, it can be seen that there is no periodic luminance value change in the time axis direction. Further, it can be seen from the amplitude data in FIG. 35 that the frequency component having the maximum amplitude is not clear and the amplitude value of any frequency component is small.

From these results, it can be determined that intimal movements such as ovulation and menstrual periods are not performed in the luteal phase. In the luteal phase (secretory phase), it is thought that the intima does not perform peristaltic movement because of the implantation of the pregnant eggs and the retention of the early pregnancy eggs, which is consistent with the analysis results of this experiment.
Specify the direction for obtaining the phase gradient as the direction from the upper right corner to the lower left corner of the symmetric image, obtain the luminance value difference (phase gradient) with the adjacent pixel for each pixel, and obtain the positive phase gradient (from the upper right corner) Corresponds to the blue color data for the case where the phase increases toward the lower left corner), and the red color for the negative phase gradient (corresponds to the phase increases from the lower left corner toward the upper right corner) When color data is associated and a phase gradient distribution image having these color data as pixel values is created, an image shown in FIG. 36 is obtained. There is no peristaltic motion in this image.
[Experiment 4 (Analysis of muscle layer motion)]
In this experiment, we focused on the movement of the uterine muscle layer that lies outside the uterine intima. The muscular layer is composed of thick muscles and is thought to be periodically exercising like the intima.

The target image is the same as in the case of FIG. 14 (Experiment 1). Therefore, the maximum amplitude frequency distribution image is as shown in FIG.
From the maximum amplitude frequency distribution image of FIG. 15 (a), it is recognized that the frequency of motion is not constant, and therefore the frequency component representing the motion of the muscle layer cannot be uniquely determined. Therefore, amplitude images and phase images were created for several frequency components, and it was determined that the sixth frequency component, which shows the amplitude image and the phase image in FIGS. 37 and 38, corresponds to the muscle layer (described above). As shown, the seventh frequency component corresponds to the peristalsis of the intima).

From the phase image in FIG. 38, it can be determined that the direction of motion is the direction from the neck to the body. It was confirmed that the muscle layer movement is more difficult to capture manually from the image than the intima movement, but can be analyzed by this method.
The movement of the intima during the ovulation period was in the direction from the neck to the body, but this experiment showed that the direction of movement of the muscle layer was from the direction toward the body. That is, the result that the motion direction of the intima and the muscle layer was the same was obtained.

  The speed of movement was determined from point u and point x in FIG. The results are shown in Table 3.

  Next, the speed of the motion was obtained manually from the muscle layer motion captured in the image. From the 18th image (FIG. 39) to the 24th image (FIG. 40) in the image sequence, a black shadow was observed to move from the middle point S to the point T in the muscle layer. Table 4 shows the speed obtained by visual observation from this motion.

Since the speed of motion captured visually from the image and the speed of motion captured by this method are almost the same value, it can be said that this method is also effective for the analysis of muscle peristalsis.
[Other Embodiments]
In the above-described embodiment, in order to select the frequency component corresponding to the uterine peristalsis, the maximum amplitude frequency distribution image is displayed on the display 13, and the frequency component for which the amplitude image and the phase image are to be created is determined by the operator. Although selected, such selection of frequency components may be automated. Specifically, for example, the number of pixels (that is, the area) of individual frequency components in the maximum amplitude frequency distribution image is calculated, and the frequency component having the highest (for example, the largest) number of pixels (area) is automatically calculated. You may make it provide the frequency component selection process part 46 (refer FIG. 6) to select (area rank reference | standard selection means). In addition, the frequency component selection processing unit 46 automatically selects one or a plurality of frequency components corresponding to maximum amplitude data equal to or greater than a predetermined threshold as frequency components having a high probability of corresponding to uterine peristalsis. You may do it.

In the above-described embodiment, the operator specifies the direction of the phase gradient to be calculated from the input device 11, but the direction of the phase gradient is a predetermined direction (from the upper right corner to the lower left corner). Direction, direction from right to left, etc.).
In addition, various design changes can be made within the scope of matters described in the claims.

An example of a uterine MR image (sagittal section) is shown. An example of a uterine MR image is shown. The example which plotted the change of the luminance value of a time-axis direction is shown. An example of a maximum amplitude frequency distribution image (color image) is shown. The image which highlighted the color part of the 1st frequency of the color image of Fig.4 (a) is shown. The image which highlighted the color part of the 2nd frequency of the color image of Fig.4 (a) is shown. The image which highlighted the color part of the 6th frequency of the color image of Fig.4 (a) is shown. It is a block diagram for demonstrating the basic hardware constitutions of the diagnostic imaging support system which concerns on one Embodiment of this invention. It is a block diagram for demonstrating the functional structure of the computer main body which functions as the said image processing apparatus. It is a flowchart which shows collectively the flow of the process for processing uterine MR image data and visualizing uterine peristalsis. It is an example of an image for demonstrating 64 time-sequential uterine MR images created for the simulation experiment. The time change of the luminance value of the pixel of the position A and B of FIG. 8 is shown. 9 shows a maximum amplitude frequency distribution image for the uterine MR image of FIG. An amplitude image is shown. A phase image is shown. A phase gradient distribution image is shown. An example of a uterine MR image in the ovulation period is shown. The maximum amplitude frequency distribution image of the image of FIG. 14 is shown. FIG. 16 shows an image in which the color portion of the third frequency in the color image of FIG. FIG. 16 shows an image in which the color portion of the sixth frequency in the color image of FIG. FIG. 16 shows an image in which the color portion of the seventh frequency of the color image of FIG. An amplitude image corresponding to peristalsis of the endometrium is shown. Shows phase images corresponding to peristalsis of the endometrium FIG. 17 shows changes in luminance values of the four points a, b, c, and d in FIG. 17 in the time axis direction (original data). FIG. 17 shows changes in the luminance values of the four points a, b, c, and d in FIG. 17 in the time axis direction (in which an inverse Fourier transform is applied to a DC component and a seventh frequency component). The 28th uterine MR image of the uterine MR image sequence of FIG. 14 is shown. The 37th uterine MR image of the uterine MR image sequence of FIG. 14 is shown. A phase gradient distribution image is shown. An overlay image is shown. An example of a menstrual uterine MR image is shown. The maximum amplitude frequency distribution image is shown. An amplitude image corresponding to peristalsis of the endometrium is shown. The phase image corresponding to the peristalsis of the endometrium is shown. A phase gradient distribution image is shown. An overlay image is shown. An example of a uterine MR image in the luteal phase is shown. The maximum amplitude frequency distribution image is shown. An amplitude image is shown. A phase image is shown. 30 shows original data (change in luminance value with time) of four points q, r, s, and t in FIG. 30 is a frequency spectrum (amplitude data of each frequency component) at four points q, r, s, and t in FIG. A phase gradient distribution image is shown. An amplitude image corresponding to peristalsis of the myometrium is shown. The phase image corresponding to the peristalsis of the myometrium is shown. The 18th image of the uterine MR image sequence of FIG. 14 is shown. The 24th image of the uterine MR image sequence of FIG. 14 is shown.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Computer main body 11 Input device 11A Keyboard 11B Mouse 13 Display 14 Printer 15 CPU
16 ROM
17 RAM
Reference Signs List 18 hard disk drive 19 CD-ROM drive 20 input control unit 21 display interface 22 printer interface 23 bus 30 image processing operation unit 31 affine transformation processing unit 32 trimming processing unit 33 Fourier transform processing unit 34 maximum amplitude frequency component extraction processing unit 35 maximum amplitude Frequency distribution image data generation processing unit 36 Frequency component selection input reception unit 37 Amplitude image data generation processing unit 38 Phase image data generation processing unit 39 Phase gradient calculation direction input reception unit 40 Phase gradient data generation processing unit 41 Phase gradient distribution image data generation Processing unit 42 Image overlay processing unit 43 Speed measurement point input receiving unit 44 Phase difference calculation unit 45 Peristaltic speed calculation unit 46 Frequency component selection processing unit 50 Data storage unit 51 Atomic-miya MR image data storage unit 52 Affine Replacement uterine MR image data storage unit 53 Target uterine MR image data storage unit 54 Fourier transform data storage unit 55 Maximum amplitude frequency component data storage unit 56 Maximum amplitude frequency distribution image data storage unit 57 Amplitude image data storage unit 58 Phase image data storage Unit 59 phase gradient data storage unit 60 phase gradient distribution image data storage unit 61 overlay image data storage unit 62 phase difference data storage unit 63 peristaltic velocity data storage unit 70 image display control unit

Claims (19)

An image processing device for analyzing and visualizing peristaltic motion of the uterus,
Image data storage means for storing a plurality of uterine MR image data corresponding to a plurality of uterine MR (magnetic resonance) images taken in time series with a cycle shorter than the cycle of uterine peristalsis;
Fourier transform means for performing Fourier transform in the time axis direction on a plurality of uterine MR image data stored in the image data storage means, and generating phase data of a plurality of frequency components for each pixel in the uterine MR image; ,
Using the phase data of at least one frequency component generated by the Fourier transform means, a phase gradient in a predetermined direction of the frequency component in each part of the uterine MR image is calculated, and phase gradient data representing the phase gradient is generated. Gradient data generating means;
Phase that generates phase gradient distribution image data representing a phase gradient distribution image in which each part of the image is colored in accordance with the phase gradient based on the phase gradient data generated by the phase gradient data generation means as image data representing uterine peristalsis An image processing apparatus for visualizing uterine peristalsis, comprising gradient distribution image data generation means. The plurality of uterine MR image data stored in the image data storage means is subjected to geometric transformation for compensating for the variation of the position and shape of the uterine image portion, and the uterine MR image after the geometric transformation processing Further comprising a geometric transformation means for generating data,
2. The image processing apparatus according to claim 1, wherein the Fourier transform means performs Fourier transform on the uterine MR image data after the geometric transformation processing by the geometric transformation means. The Fourier transform means generates amplitude data of a plurality of frequency components for each pixel of the uterine MR image in addition to the phase data,
Frequency component selection means for selecting at least one frequency component based on the amplitude data generated by the Fourier transform means;
The image processing apparatus according to claim 1, wherein the phase gradient data generation unit generates phase gradient data of each part of the uterine MR image with respect to the frequency component selected by the frequency component selection unit. The frequency component selection means includes
Maximum amplitude frequency extracting means for extracting a maximum amplitude frequency component corresponding to maximum amplitude data for each pixel of the uterine MR image;
Maximum amplitude frequency distribution image data generating means for generating maximum amplitude frequency distribution image data representing the distribution in the uterine MR image of the maximum amplitude frequency component extracted for each pixel by the maximum amplitude frequency extracting means. The image processing apparatus according to claim 3, wherein: The frequency component selection means further includes:
5. The image processing apparatus according to claim 4, further comprising frequency component selection input receiving means for receiving a frequency component selection input by an operator. The frequency component selection means further includes:
Based on the maximum amplitude frequency distribution image data generated by the maximum amplitude frequency distribution image data generation means, based on the rank order of the areas of individual frequency components in the image corresponding to the maximum amplitude frequency distribution image data, at least one The image processing apparatus according to claim 4, further comprising an area rank reference selection unit that selects a frequency component.   4. The image processing apparatus according to claim 3, wherein the frequency component selecting means selects a frequency component corresponding to maximum amplitude data not less than a predetermined threshold value. The Fourier transform means generates amplitude data of a plurality of frequency components for each pixel of the uterine MR image in addition to the phase data,
Amplitude image data generating means for generating amplitude image data representing an amplitude image representing an amplitude distribution of the frequency component of each part of the uterine MR image using amplitude data of at least one frequency component generated by the Fourier transform means. The image processing apparatus according to claim 1, further comprising:   Phase image data generating means for generating phase image data representing a phase image representing a phase distribution of the frequency component of each part of the uterine MR image using phase data of at least one frequency component generated by the Fourier transform means The image processing apparatus according to claim 1, further comprising:   It further includes image overlay processing means for overlaying the phase gradient distribution image data generated by the phase gradient distribution image data generation means on the uterine MR image data stored in the image data storage means to generate overlay image data. An image processing apparatus according to claim 1, wherein: A phase difference calculating means for calculating a phase difference between the two pixels based on phase data of at least two spaced apart pixels among phase data of one frequency component generated by the Fourier transform means;
11. The image processing apparatus according to claim 1, further comprising a peristaltic speed calculating means for calculating a peristaltic speed based on the phase difference calculated by the phase difference calculating means. Means for receiving an instruction input of two points in the image by the operator;
12. The phase difference calculating means is configured to calculate a phase difference between the two pixels based on the phase data of the two pixels based on the received instruction input. Image processing device. An image processing apparatus according to any one of claims 1 to 12,
An image diagnosis support system comprising: a display device connected to the image processing device and displaying the phase gradient distribution image.   14. The image according to claim 13, further comprising an input device operable by an operator and connected to the image processing device for inputting an operator's instruction input to the image processing device. Diagnosis support system. An image processing method for analyzing and visualizing peristaltic movement of the uterus,
Preparing a plurality of uterine MR image data corresponding to a plurality of uterine MR (magnetic resonance) images taken in time series in a cycle shorter than the cycle of uterine peristalsis;
Fourier transform step of performing Fourier transform in the time axis direction on the plurality of uterine MR image data and generating phase data of a plurality of frequency components for each pixel in the uterine MR image,
Using the phase data of at least one frequency component generated by the Fourier transform step, the phase gradient for calculating the phase gradient in the predetermined direction of the frequency component in each part of the uterine MR image is generated, and the phase gradient data representing the phase gradient is generated. A gradient data generation step;
Phase that generates phase gradient distribution image data representing a phase gradient distribution image in which each part of the image is colored in accordance with the phase gradient based on the phase gradient data generated by the phase gradient data generation step as image data representing uterine peristalsis An image processing method for visualizing uterine peristalsis comprising a gradient distribution image data generation step. A phase difference calculating step of calculating a phase difference between the two pixels based on phase data of at least two pixels separated from the phase data of one frequency component generated by the Fourier transform step;
16. The image processing method according to claim 15, further comprising a peristaltic speed calculating step for calculating a peristaltic speed based on the phase difference calculated in the phase difference calculating step. To process a plurality of uterine MR image data corresponding to a plurality of uterine MR (magnetic resonance) images taken according to time series in a cycle shorter than the cycle of uterine peristalsis, and to analyze and visualize the peristaltic motion of the uterus A computer program for operating a computer as the image processing apparatus of
The computer,
Fourier transform means for performing Fourier transform in the time axis direction on the plurality of uterine MR image data, and generating phase data of a plurality of frequency components for each pixel in the uterine MR image,
Using the phase data of at least one frequency component generated by the Fourier transform means, a phase gradient in a predetermined direction of the frequency component in each part of the uterine MR image is calculated, and phase gradient data representing the phase gradient is generated. Gradient data generation means, and phase gradient distribution image data representing a phase gradient distribution image in which each part of the image is colored in accordance with the phase gradient based on the phase gradient data generated by the phase gradient data generation means represents uterine peristalsis A computer program for visualizing uterine peristalsis characterized by functioning as phase gradient distribution image data generating means for generating image data. Said computer further
Phase difference calculation means for calculating a phase difference between the two pixels based on phase data of at least two separated pixels among phase data of one frequency component generated by the Fourier transform means; and 18. The computer program according to claim 17, wherein the computer program functions as a peristaltic speed calculating means for calculating a peristaltic speed based on the phase difference calculated by the phase difference calculating means.   A recording medium in which the computer program according to claim 17 or 18 is recorded and readable by a computer.