JP2011160824A - Spinal curvature determination support apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To easily determine spinal curvature according to the shape of a body surface. <P>SOLUTION: Cross section data forming means 5i forms data including at least a part of the cross section of a plurality of human bodies from the three-dimensional data of a human body back surface. Single elliptic approximation means 5j identifies a single ellipse by performing the approximation operation processing of the data including at least a part of the cross section. Identification means 5k identifies a spinal curvature portion where an error between the cross section data and the single ellipse is equal to or larger than a preset threshold, and generates identification output. Spinal curvature candidate acquisition means acquires the candidate of the spinal curvature on the basis of the generation of the identification output. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a spondylosis determination support device for simplifying determination of spondylosis.

  As in Patent Document 1, spondylolisthesis is diagnosed by confirming the three-dimensional arrangement of the spine using X-ray photographs in the front-rear direction and the left-right direction of the body.

  However, because X-ray imaging involves the risk of exposure, in order to extract patients with spondylosis with a low incidence by mass screening, it is important to perform multiple X-ray imaging for all patients in terms of cost and safety. This is not a desirable method.

  Therefore, group screening has to rely on the doctor's visual judgment to find patients with possible spondylosis, but visual diagnosis often overlooks patients with mild spondylosis. There is also a problem that early detection is impossible while light.

  Therefore, as described in Patent Document 2, the symptoms of the undulation of the back of a patient with spondylosis are asymmetric, and the possibility of spondylosis is determined by the doctor from the moire pattern projected on the back of the human body. There has been proposed a diagnosis assisting device that facilitates diagnosis.

JP 2003-290217 A JP 2005-137462 A

  However, in the apparatus of Patent Document 2, the determination criterion is not constant because the determination person visually recognizes the moire and determines the asymmetry of the body surface, and it may be difficult to make an objective determination.

  In addition, because the device is expensive, it is difficult to spread it to elementary and junior high schools, which is essential, and because it is not possible to receive a medical checkup during a regular health checkup, the symptoms are delayed and the symptoms improve when they are discovered. There were many cases where became extremely difficult.

  The present invention has been made in view of such circumstances, and an object thereof is to provide an apparatus for objectively and automatically determining body asymmetry.

  One embodiment of the present invention is a spinal curvature certificate diagnosis support apparatus. The apparatus includes a cross section data forming means for forming a plurality of cross section data including at least a part of a cross section of the human body from the three-dimensional data on the back of the human body, and approximate calculation processing of the cross section data to obtain a single unit. Single ellipse approximation means for specifying an ellipse, identification means for identifying a spinal curve portion where an error between the cross-sectional data and the single ellipse is equal to or greater than a preset threshold, and generating an identification output, and generation of an identification output And a spondylosis candidate acquisition means for acquiring a spondylosis candidate.

  The present invention also provides a double ellipse approximation means for identifying two ellipses by dividing the cross-sectional data and performing approximate calculation processing, respectively, and a spine on the short axis of the single ellipse at a normal site. A spine position estimating means for estimating and specifying the center position and estimating the center position of the spine from a straight line connecting the intersections of the two ellipses at the spinal curve portion; and an estimated spine based on the center position of the spine Estimated spine line data forming means for specifying a line and display data generating means for supplying the estimated spine line and display means may be additionally arranged.

  In addition, the spine position estimating means includes at least a center position of the spine obtained from a subject known to be healthy and a center position of the spine obtained from a subject known to have spondylosis. The central position of the spine may be acquired by referring to a database including the database.

  Furthermore, the present invention provides a projection unit that projects a dot pattern on the back of a human body, an imaging unit that outputs an image of the back of the human body in a state where the dot pattern is projected from two different directions, and the image that is simultaneously captured. Further, a back three-dimensionalization means for outputting the spatial positions of the dots constituting the dot pattern as three-dimensional data may be additionally arranged.

  ADVANTAGE OF THE INVENTION According to this invention, the apparatus which determines the asymmetry of a body objectively and automatically without relying on a doctor's intuition can be provided.

It is arrangement | positioning explanatory drawing of the still image imaging device of embodiment. It is a figure which shows the relationship between arrangement | positioning of the still image imaging device of FIG. 1, and the cross section of a human body. It is a figure which shows the relationship between the cross section of the human body of a normal site | part, and a measurement line. It is a figure which shows the relationship between the cross section of the human body of a spinal curve part, and a measurement line. It is the figure which made the ellipse approximate to the measurement line of the cross section of a human body in a normal part. It is the figure which approximated one ellipse to the measurement line of the human body cross section of a spinal curve part. It is the figure which approximated two ellipses to the measurement line of the human body cross section of a spinal curve part. It is a figure which shows the relationship between the sampling point and ellipse for deriving | leading-out the formula for specifying the ellipse to approximate. It is a functional block diagram of PC concerning an embodiment. It is a flowchart which shows operation | movement of PC concerning embodiment. An example of the approximate ellipse of the closed curve of the body surface part is illustrated. It is the figure which showed an example of the structure of the database which stored the normal person's spine center position. The example of the result of having approximated the closed curve of a body surface part with a double ellipse about the spondylosis part of a spondylosis patient is shown in figure. It is the figure which showed an example of the structure of the database which stored the spine center of the spine in the spondylosis part of the patient who is known to have the spondylosis part. It is the figure which showed typically the internal structure of the spine position estimation means concerning embodiment.

  Hereinafter, embodiments for carrying out the present invention will be described with reference to the drawings.

  In patients with scoliosis, a portion of the spine turns left or right and becomes twisted. As a result, the ribs are also distorted at the curved portion of the spine that is bent and twisted, and the back of the human body becomes asymmetrical.

  The inventor has a small error when approximating the cross section of the human body at the normal site with a single ellipse, and the error is large when approximating the human cross section at the spinal curve portion with a single ellipse. We found that the error would be smaller if the surface was approximated by separate ellipses.

  It was also found that the normal part has the center of the spine on the short axis radius of a single ellipse, and the spine curved part has the center position of the spine almost on the straight line connecting the intersections of the two ellipses.

  Furthermore, by approximating with an ellipse, it can be said that there is almost no trouble in determining whether or not the spinal curvature is proved if the ellipse is rotated and approximated even if the human body cross-section is imaged with a certain degree of inclination. It was also confirmed that there is a merit.

  In addition, when the inventor obtains the three-dimensional data of the entire circumference of the human body, the measuring device becomes large, so even if only the three-dimensional data of the back of the human body obtained by measuring only the back side of the human body is used, It was confirmed that spondylosis could be judged without any difficulty.

<< Embodiment >>
An outline of the embodiment will be described. In the embodiment, a single ellipse is approximated to the measurement line of each cross section obtained by dividing the back surface of the human body in multiple stages, and the mean square error between the single ellipse and the measurement line of the cross section is large over a plurality of continuous cross sections. The vertebrae with a large mean square error are estimated by estimating the center position of the spine on the short-axis radius of a single ellipse for a normal section with a small mean square error. For the cross section of the curved part, approximate the two ellipses on the left and right, estimate the center position of the spine in a straight line connecting the intersections of the two ellipses, form the spine line by connecting the estimated center positions, and with the simple judgment result The curved state of the spine is displayed on the monitor.

  As shown in FIG. 1, in the present embodiment, three-dimensional data on the back of a human body is measured using a still image capturing device with a simple configuration.

  In the still image pickup device, the projector 2 is fixed to the center of the fixed plate 1 and the first camera 3 and the second camera 4 are fixed to the left and right sides of the fixed plate 1.

  The projector 2 projects the dot pattern from directly behind the back of the human body, and the first camera 3 and the second camera 4 face each other inward so that the projected dot pattern is imaged from left and right positions. Is fixed.

  The projector 2, the first camera 3 and the second camera 4 are connected to a PC (Personal Computer) 5, and a monitor 6 is connected to the PC 5. Note that an input device from a user (not shown) such as a keyboard and a mouse is also connected to the PC 5.

  In the determination, first, a dot pattern is instantaneously projected from the projector 2 to the back of the human body in synchronization with the command of the PC 5.

  This dot pattern is composed of a large number of dots arranged in a lattice pattern at regular intervals in the vertical direction, and dots of different colors are arranged in order to correctly identify each dot without confusion.

  As a result, in the present embodiment, each dot can be identified without being confused in later processing.

  In order to identify the position, in addition to changing the color arrangement, dot confusion can be prevented by changing the light intensity of each dot or changing the shape of each dot. .

  The PC 5 captures still images of the first camera 3 and the second camera 4 at the moment when the dot pattern is projected on the back of the human body.

  Each dot imaged by the first camera 3 and the second camera 4 is identified based on the color arrangement, and the imaging position for each dot is correctly stored in the PC 5 as two-dimensional imaging position data.

  The three-dimensional spatial position data of the dots on the undulating human body surface can be specified by the principle of triangulation by calculating the imaging position data in the first camera 3 and the imaging position data in the second camera 4. .

  This specific principle will not be described in further detail with respect to a known principle disclosed and used in, for example, Japanese Patent Application Laid-Open No. 2007-101276.

  In this embodiment, using this principle, the spatial position data of the dots is calculated from the imaging position data in the first camera 3 and the imaging position data in the second camera 4 for each dot.

  In the case of the present embodiment, the three-dimensional data specified by this calculation can keep the undulation error within 1 mm when the dot interval is 1 cm.

  Therefore, even if an image is taken with the underwear worn, a slight error occurs, but this does not cause a significant hindrance to the determination. In this embodiment, the three-dimensional data on the human body surface (back surface) This includes not only the surface 3D data but also the underwear surface 3D data.

  As shown in FIG. 2, since the imaging ranges of the first camera 3 and the second camera 4 are different, the range (measurement line) that can be measured on the outer periphery of the cross section is limited to the common imaging range.

  FIG. 3 shows the relationship between the cross section of the human body at the normal site and the measurement range (measurement line), the broken line indicates the human body surface, and the solid line indicates the measurement line.

  FIG. 4 shows the relationship between the cross section of the human body at the spinal curve site and the measurement range (measurement line), the broken line indicates the human body surface, and the solid line indicates the measurement line.

  FIG. 5 shows a diagram in which an approximate ellipse is approximated to a measurement line of a cross section of a human body at a normal site. As can be seen from FIG. 5, in the case of a normal part, the difference from the approximate ellipse line approximated to the measurement line is small as shown by hatching.

  FIG. 6 is a diagram in which a single approximate ellipse is approximated to the measurement line of the human body cross section of the spinal curve region. As can be seen from FIG. 6, in the case of the spinal curve portion, the difference between the measurement line and the approximate ellipse line approximated to the measurement line increases as shown by hatching.

  FIG. 7 shows a diagram in which two approximate ellipses are approximated to the measurement line of the human body cross section of the spinal curve portion.

  As can be seen from FIG. 7, in the case of a curved portion of the spine, if two approximate ellipses are approximated to the measurement line, the difference from the measurement line is smaller than that in the case of one approximation as shown by hatching. .

As shown in FIG. 8, when specifying an approximate ellipse as shown in the figure from two-dimensional data of a cross section obtained from three-dimensional data, the center coordinates of the approximate ellipse are (X0, Y0), and the major axis and the X axis are formed. If the angle is θ, the radius of the major axis is a, the radius of the minor axis is b (where a ≠ b), and the coordinates of the points on the approximate ellipse are (Xi, Yi), the ellipse equation is as follows.

When this formula is expanded, the following formula is obtained.

In order to simplify this mathematical expression, if the square coefficient of Xi is 1, and A, B, C, D, and E are applied to the coefficients and constants of the remaining terms, the following expression is obtained.

この数式を前述の５つの定数Ａ、Ｂ、Ｃ、Ｄ、Ｅでそれぞれ偏微分した式の値を０とし、定数Ａ、Ｂ、Ｃ、Ｄ、Ｅに依存しない項を右辺に移項すると以下の数式を得る。

Substituting n for the number of data in the cross section and the coordinates (Xi, Yi) (i = 0, 1,..., N−1) of the cross section data and substituting the square value into e, It becomes the following formula.

When this equation is partially differentiated by the above-mentioned five constants A, B, C, D, E, the value of the equation is set to 0, and a term that does not depend on the constants A, B, C, D, E is transferred to the right side, Get the formula.

When this equation is converted to obtain constants A, B, C, D, and E, the following equation is obtained.

In order to obtain (X0, Y0) for the constants A, B, C, D, and E obtained by this equation, the following equation is used.

Furthermore, in order to obtain θ, the following mathematical formula is obtained.

Similarly, the major axis radius a is obtained by the following mathematical formula.

Further, the short axis radius b is obtained by the following formula.

The elliptical equation fitted with the multipliers obtained by these mathematical formulas is integrated by substituting n cross-sectional data into the squared formula as shown in the following mathematical formula, and then divided by the number n of the cross-sectional data. And the mean square error ε is obtained. Although the integrated value varies depending on the number n of cross-sectional data of the cross-sectional data, the number of data is not always constant for all cross-sectional data, so the integrated value is the number of cross-sectional data for error evaluation. It is necessary to obtain a mean square error (error per data) ε divided by n.

  From the above formula, the ellipse closest to the cross-sectional data can be obtained by the least square method, that is, in the sense that the mean square error is minimized, and the mean square error can also be obtained by the above formula 11.

  FIG. 9 is a functional block diagram of the PC 5 according to the embodiment. A projector 2, a first camera 3, a second camera 4, and a monitor 6 are connected to the PC 5.

  The display data generating means 5b inside the PC 5 is connected to the projector 2 via an external output terminal, in addition to the monitor 6 attached to the PC 5. This projector functions as both a projection unit and a display unit.

  The first memory 5c and the second memory 5d in the PC 5 are connected to the first camera 3 and the second camera 4 via the external input terminal of the PC 5, respectively.

  The display data generating means 5b supplies the display data to the projector 2 so that the projector 2 instantaneously projects a grid-like dot pattern with a 1 cm interval on the back of the human body on the back of the human body.

  The first memory 5c and the second memory 5d receive image data from the first camera 3 and the second camera 4, respectively, and store the image data at the dot pattern projection timing.

  The timing pulse generator 5a supplies timing pulses to the display data generator 5b, the first memory 5c, and the second memory 5d in order to synchronize the display timing of the dot pattern and the storage timing of the image data.

  The two-dimensionalization means 5e converts the corresponding dots into two-dimensional coordinates from the image data stored in the first memory 5c and the second memory 5d as a clue, and converts them into two-dimensional coordinates. Form two-dimensional data.

  From the pair of two-dimensional data, the three-dimensionalization means 5f converts the data into three-dimensional coordinates in order to specify the spatial position of each dot, and stores the three-dimensional data in the third memory 5g.

  The Delaunay plotting means 5h gives the human body by coloring the surface formed by the triangle according to the inclination state of the triangle specified by the data of the three adjacent points in the three-dimensional coordinates converted into the data by the three-dimensionalizing means 5f. In order to display the rear surface as a three-dimensional surface that is easy to see, Delaunay mapping data is formed.

  Specifically, the Delaunay mapping means 5h specifies a small triangle by combining three adjacent points from the three-dimensional data read from the third memory 5g, calculates the inclination of the small triangle, and colors the slope according to the inclination. The Delaunay mapping data subjected to is supplied to the display data generating means 5b.

  The display data generating means 5b inputs the Delaunay mapping data supplied from the Delaunay mapping means 5h and supplies Delaunay diagram image information to the monitor 6 attached to the PC 5.

  Note that the location information of the Delaunay plotting data is also stored in the third memory 5g and used for accurate cross-sectional data calculation.

  The cross section data forming means 5i uses the Delaunay diagram image and the three-dimensional data stored in the third memory 5g to identify each cross section when the back of the human body is cut from the head side of the human body at 1 cm intervals. By performing arithmetic processing, two-dimensional cross-sectional data is sequentially output.

  The single ellipse approximating means 5j performs an operation of approximating the cross section data of each stage with a single ellipse, and specifies a short-axis radius line segment of the single ellipse.

  The identification means 5k integrates the square value of the error between the approximated single ellipse and each cross-sectional data, and then sets an average square error obtained by dividing the integrated value by the number n of the cross-sectional data. An identification output is issued when the threshold value is exceeded. This threshold value may be determined experimentally based on data for creating a database, which will be described later.

  The double ellipse approximating means 51 receives the cross section data as input and approximates the cross section with two ellipses, and specifies a line segment connecting the intersections of the two approximate ellipses (hereinafter referred to as “double ellipse”).

  The spine position estimating means 5m obtains the first estimated position as the spine center at the normal site from the short axis radius specified by the single ellipse approximating means 5j, and stores it in the fourth memory 5n, and from the identifying means 5k to the single ellipse. When an identification output generated when approximation cannot be performed with the input is input, a second estimated position is obtained as a spine center in the curved portion of the spine from a line connecting the intersection points of the double ellipses, and the second estimated position is substituted for the first estimated position. The position is stored in the fourth memory 5n. Although details will be described later, the spine position estimating means 5m identifies the spine center at the normal site and the spine center at the spinal curve site by referring to a database created in advance.

  As a result, the fourth memory 5n stores the center of the spine estimated according to the normal part and the spine curved part as the estimated position.

  The estimated spine line data forming means 5o temporarily connects the two-dimensional estimated positions stored in the fourth memory 5n to form three-dimensional line data.

  Next, the estimated spine line data forming means 5o gently corrects the discontinuous part of the line generated at the boundary between the normal part and the spine curved part to form continuous three-dimensional line data.

  Further, the estimated spine line data forming unit 5o supplies the display data generating unit 5b with the estimated spine line data to which information for identifying and displaying the spinal curve portion in the form of color, luminance, or line is displayed.

  The ideal spine line data forming means 5p selects only the first estimated position of the normal part from the two-dimensional estimated positions stored in the fourth memory 5n, and temporarily forms the three-dimensional line data.

  Next, the ideal spine line data forming unit 5p regards the spinal curve portion lacking the line data as a straight line, and interpolates the line data to form spine line data.

  Furthermore, the ideal spine line data forming means 5p gently corrects the unnatural discontinuous part of the line generated at the boundary between the normal part and the spine curved part, and forms continuous three-dimensional ideal spine line data. And supplied to the display data generating means 5b.

  Here, when storing the first estimated value and the second estimated position, the fourth memory 5n adds and stores identification information.

  Therefore, when displaying the estimated spine line, the second estimated position can be identified and displayed as a spinal curve portion.

  Further, when displaying an ideal spine line, only the first estimated position determined as a normal part can be selected.

  The display data generating means 5b inputs the estimated spine line data formed by the estimated spine line data forming means 5o and the ideal spine line data formed by the ideal spine line data forming means 5p, and the estimated spine line is displayed on the screen of the monitor 6. And the ideal spine line are displayed three-dimensionally.

  The display data generating means 5b also supplies the projector 2 with image information for projecting back two lines of estimated spine line data and ideal spine line data.

  As a result, immediately after the projection of the dot pattern, an estimated spine line tracing the spine is displayed on the back of the human body, and the spinal curve portion is identified and displayed with different colors, brightness, and line display forms.

  Furthermore, by the display switching operation, the ideal spine line is also displayed on the back of the human body at the same time as the estimated spine line, the position of the spinal curve and the state of the curve are displayed in an easy-to-understand manner, and the undulating state of the curve can be confirmed by palpation.

  Further, the display data generating means 5b supplies image information to display on the monitor 6 a cross section based on the cross section data or an approximated single ellipse or double ellipse as necessary.

  The comparison calculation means 5q inputs the estimated spine line data and the ideal spine line data, calculates the difference between them, a range where the deviation is greater than a predetermined amount (length in the estimated spine line), the direction of deviation, the amount of deviation, etc. Calculate useful numbers to determine the symptoms of spondylosis.

  The determination unit 5r determines whether or not the candidate is a spondylosis candidate based on whether or not the identification output has continuously occurred, and estimates the degree of the symptom based on the numerical value obtained by the comparison calculation unit 5q. To do.

  The determination result and the estimation result are input to the display data generating means 5b and displayed on the monitor 6.

  Although the above has been described as a functional block of the PC 5, the above functional block can be realized by an arbitrary processor, memory, or other LSI (Large Scale Integration) in terms of hardware, and loaded into the memory in terms of software. Realized by programs. Therefore, those skilled in the art will understand that these functional blocks can be realized in various forms by hardware only, software only, or a combination thereof.

  FIG. 10 is a flowchart of the determination operation executed by the PC 5 according to this embodiment.

  The projector 2 instantaneously projects the dot pattern on the back of the human body (S10).

  The 1st camera 3 and the 2nd camera 4 image the dot pattern projected on the human body back, and memorize | store an imaging screen in the 1st memory 5c and the 2nd memory 5d, respectively (S12).

  The two-dimensionalization means 5e corresponds to the position of each dot based on the color arrangement from the dot pattern of the imaging screen of the first camera 3 and the second camera 4 stored in the first memory 5c and the second memory 5d. Then, the position in each imaging screen is calculated for each dot, and a pair of two-dimensional data is formed (S14).

  The three-dimensionalization means 5f calculates and calculates three-dimensional data for specifying the spatial position from each pair of two-dimensional data associated with the two-dimensionalization means 5e for all dots within the imaging range. 3 is stored in the memory 5g (S16).

  The Delaunay mapping means 5h specifies a small triangle having a set of three adjacent points in the 3D data specified by the 3D conversion means 5f in order to form a Delaunay chart from the 3D data of each dot. The direction of the tilt is calculated and colored according to the tilt to form a three-dimensional image of the back side of the human body, and the captured back side of the human body is three-dimensionally displayed on the monitor 6 via the display data generating means 5b (S18). .

  The cross section data forming means 5i specifies the position at the specified height on the surface of the small triangle specified by the Delaunay mapping means 5h in order to specify the cross section when the back of the human body is cut in 1 cm increments. Then, two-dimensional cross-sectional data (corresponding to position data of the measurement line shown in FIG. 2 and subsequent figures) is calculated by calculation (S20).

  Note that the above-described cross-section data forming step S20 is a step for improving the accuracy of the data. When a slight error is allowed, the three-dimensional data of the dots projected at the same height in the dot pattern is used. It can be used as it is to obtain two-dimensional cross-sectional data, which is not an essential step in the present invention.

  The single ellipse approximating means 5j reads the uppermost cross section data when the head side of the human body is the upper side, specifies the approximate ellipse by the least square method described above, and then determines the short axis radius of the specified approximate ellipse , The predetermined position of the line segment is regarded as the center position of the spine, the center position obtained by the calculation is stored in the memory as the first estimated position, and the square of the distance between the specified approximate ellipse and the cross-sectional data is calculated The values are integrated and the result is divided by the number of data to calculate the mean square error associated with the approximation of a single ellipse (S22).

  The discriminating means 5k compares the mean square error obtained by the single ellipse approximating means 5j with a predetermined threshold value, and determines that a part where the mean square error is larger than the threshold value is a spinal curve part where twist (curvature) occurs. If it is small (S24Y), it is determined as a normal part (S24N).

  If the double ellipse approximating means 5l is determined to be a vertebral curved part, the double ellipse approximating means 5l identifies a position that is recessed near the center of the cross-sectional data as a discontinuous point (S26). This can be realized, for example, by calculating the difference between the coordinates of adjacent cross-sectional data and specifying the difference as a discontinuous point when the difference exceeds a predetermined threshold value.

  The double ellipse approximation means 5l further specifies two approximate ellipses using the above-mentioned least square method using the left and right cross-sectional data divided by the specified discontinuous points, and the spine position estimation means 5m , The predetermined position of the line segment connecting the intersections of the two approximate ellipses is regarded as the center position of the spine in the curved portion of the spine, and the approximate position calculated and stored in step S12 is set as the second estimated position. The first estimated position of the spine based on the ellipse is replaced and stored in the fourth memory 5n (S28).

  By the above-described step S28, the estimated position of the spine is specified in the uppermost cross section data.

  That is, if the back of the human body can be approximated by one approximate ellipse, the first estimated position can be approximated. If it cannot be approximated by one approximate ellipse, the second estimated position can be replaced by the estimated position at the center of the spine instead of the first estimated ellipse. Is stored in the fourth memory 5n.

  As a result of determining whether or not the above-described estimation work is an estimation based on the cross-sectional data at the lowest stage, if the estimation work has not reached the lowest stage (S30N), the cross-section data forming means 5i sets the cross section to 1 cm. Update to the lower level (S32).

  After the update, the cross section data forming unit 5i proceeds to the above-described S22, reads the next (lower) cross section data from the third memory 5g, and repeats the above-described steps S22 to S30.

  In the present embodiment, the estimation calculation is repeated by the number n of the cross sections, and the estimated position of the center of the spine of each cross section is specified and stored, and the estimated spine position is determined at the bottom of the cross section by the above-described step S30. If it is determined that it has been identified (S30Y), the estimation work is terminated and the process proceeds to the next step S34.

  The estimated spine line data forming unit 5o forms an estimated spine line that connects the estimated positions of the obtained spine centers up and down, and the ideal spine line data forming unit 5p is an ideal spine assumed from the estimated position of the normal part. The display data generating means 5b displays two spine lines on the monitor 6 and the projector 2 (S34).

  The comparison calculation means 5q compares both lines, calculates the difference, and quantifies the range where the deviation is greater than a predetermined amount (estimated spine line length), the direction of deviation, the amount of deviation, etc. (S36).

  The determination means 5r obtains data for determining the scoliosis and the degree of symptom on the basis of the digitized range, amount, and direction of the shift, and displays the result on the monitor 6. (S38).

  In the data used for the determination of spondylosis, if at least the mean square error continuously exceeds the threshold value for a predetermined number of times, it is determined as a spondylosis candidate. It is useful to consider other factors.

  In addition, when taking an image with underwear, it is expected that the dent in the center of the back cannot be detected as a discontinuity, but in that case, the center of the measurement line is regarded as a discontinuous point and an approximation of a double ellipse is performed. Do. Also, since it is feared that wrinkles in the underwear are detected and misidentified as discontinuous points, even if the 3D data is processed in advance using a smoothing filter or the like to suppress fine undulations as necessary Good.

  In the present embodiment, the image information supplied to the monitor 6 and the image information obtained by imaging the estimated spine line and the ideal spine line projected by the projector 2 on the back of the human body with the first camera are stored in the solid-state memory for the PC 5. It is also possible to display the image displayed on the monitor 6 and the images of both spine lines projected on the back of the human body on another PC.

  Further, the image information recorded in the recording means 5s can be reproduced not only by being transmitted to the hospital via the communication network but also by bringing the above-mentioned recording means 5s directly to the hospital.

  Hereinafter, a method for specifying the spine center at the normal site and the spine center at the spine curve site by the spine position estimating means 5m will be described. The outline is that a spine center in a normal site and a spinal curve site is specified by referring to a database prepared in advance.

  Cross-sectional images in the axial direction of a range including at least the lumbar region and the cervical region for a plurality of patients whose spondylosis is known to exist and a plurality of healthy individuals whose spondylosis is not known Prepare. The cross-sectional image is created using, for example, X-ray CT (Computed Tomography) or MRI (Magnetic Resonance Imaging). Here, the characteristics of each of a patient and a healthy person (hereinafter collectively referred to as “subject” when both are included) are assumed to be known. The characteristics include, for example, sex, age, residential area, height, weight, waist circumference, and BMI.

  Here, it is assumed that the number of the spine captured in each cross-sectional image is also known. The spine number is a number in which 24 spines are sequentially numbered from the top. This can be realized, for example, by storing the spine number in the header portion of each cross-sectional image file or by storing it in the database 5v as a list (not shown) linked to the name of each cross-sectional image file. As will be described later, the database 5v exists inside the spine position estimating means 5m. Or you may arrange | position independently on the outer side of 5 m of spine position estimation means.

  Based on a cross-sectional image of a healthy person, an approximate ellipse is obtained by the above least square method for the closed curve of the body surface portion. Alternatively, the user may visually fit the ellipse. FIG. 11 shows an example of the approximate ellipse of the closed curve of the body surface portion thus obtained. Here, the end point on the abdomen side of the short axis of the ellipse is point A, and the end point on the back side is point B. The user observes the cross-sectional image and specifies a point C that is the center of the spine on the end axis. The user specifies the approximate ellipse, the point A, the point B, and the point C for all the images of the healthy persons prepared.

  The obtained ellipse is normalized using the short axis. Specifically, the approximate ellipse is subjected to similarity transformation so that the lengths of the short axes are all equal. Here, the length of the short axis is assumed to be 1. The position of the point C that is the center of the spine when the length of the short axis is 1, for example, is a relative distance r (0 <r <1) that is a relative distance from the point B that is the end point on the back side of the short axis. ).

  FIG. 12 is a diagram showing an example of a structure in the database 5v storing the spine center position of a healthy person. The feature 5 and the relative position of the spine center are stored in the database 5v together with the image ID, the database serial number, and the relative length a (a> 1) of the long axis.

  FIG. 13 illustrates an example of a result obtained by approximating a closed curve of a body surface part with a double ellipse for a spondylosis part of a spondylosis patient. The spine center in the normal part of the patient with spondylosis is obtained by the same method as that of the healthy person, and the structure of the database is the same. Therefore, the following description will focus on the spinal curve portion.

  Regarding the spinal curve portion, the double ellipse may be specified using the above-mentioned least square method, or the user may specify it visually. Of the two intersections of the double ellipse, the intersection on the abdomen side is point D, and the intersection on the back side is point E. The user observes the cross-sectional image and specifies a point F that is the center of the spine in the spinal curve portion on the line segment connecting the two intersections. The user identifies a double ellipse and point D, point E, and point F for all prepared patient images that are known to have spondylosis.

  The obtained double ellipse is normalized using the length of a line segment connecting two intersection points of the double ellipse. Specifically, the double ellipse is subjected to similarity transformation so that the lengths of the line segments are all equal. Here, the length of the line segment is set to 1 as in the case of a healthy person. The point F, which is the center of the spine at the curved portion of the spine when the length of the line segment is 1, is a relative distance s (0 <s <1) from the point E that is the intersection of the two line segments on the back side. Can be specified. The double ellipse, the point D, the point E, the point F, and the relative distance s are stored in the same database 5v as the database that stores the patient's characteristics and the like and the spine center position of the healthy person.

  FIG. 14 is a diagram showing an example of the structure of the database 5v that stores the spine center in the spondylosis part of a patient whose spondylosis part is known to exist. The database 5v stores the above-described features and the relative position of the spine center together with the image ID, the serial number of the database, and the relative length a of the long axis. The difference from the database of healthy people is that the coordinates of the center points of the two ellipses, the lengths of the major and minor axes, and the tilt angle θ of the second ellipse are stored so that a double ellipse can be reproduced. It is a point. Here, an arbitrary coordinate system may be used when measuring coordinates and angles. For example, for a standardized double ellipse, the center of the first ellipse is the origin, the straight line including the major axis includes the X axis, and the minor axis. A coordinate system having a straight line as the Y axis may be used.

  FIG. 15 is a diagram schematically showing the internal configuration of the spine position estimating means 5m according to the embodiment. The spine position estimation means 5m includes a database 5v, a parameter acquisition means 5t, and a database search means 5u. The parameter acquisition unit 5t acquires a feature used for a search from a user as a parameter via an input / output device (not shown). The database search means 5u searches the database 5v for the spine center in the normal part or the spinal curve part based on the identification output generated by the identification means 5k, using the parameter acquired by the parameter acquisition means 5t as a search key. .

  For example, it is assumed that the parameter acquisition unit 5t acquires age, sex, and waist circumference as a search key. When there is no identification output, that is, in the case of a normal part, the database search means 5u matches the age, sex, and abdominal circumference that are input as search keys among the normal part data stored in the database 5v. Data is searched and the average value r ′ of the relative distances r is calculated and obtained. Subsequently, the database search means 5u obtains the short axis length l of the ellipse input from the single ellipse approximation means 5j, and sets a point separated by l × r ′ from the short point on the back side of the short axis to the spine center. To the fourth memory 5n.

  When there is an identification output, that is, in the case of a spinal curve portion, the database search means 5u calculates an average value s ′ of the relative distance s of the patient that matches the search key from the data of the spinal curve portion from the database 5v. Get. Subsequently, the database search means 5u obtains the length l ′ of the line segment connecting the intersection points of the two ellipses inputted from the double ellipse approximation means 5l, and l ′ × s ′ from the intersection point on the back side of the short axis. The distant point is output to the fourth memory 5n as the spine center at the spinal curve portion.

  The database search means 5u may acquire not only a value set as a search key but also data having a value close to that value. For example, when 145 cm is set with height as a key, a range from 140 cm to 150 cm may be retrieved and acquired. In this case, matching data increases, which is advantageous in that the accuracy of the average value can be improved. Further, as a search key, for example, a search can be performed by setting a condition in a negative format, such as a condition that the height does not include 140 cm to 145 cm. The database search means 5u may use a predetermined condition as a search key instead of the search key acquired by the parameter acquisition means 5t.

  The database search means 5u may also search for data having the same spine number when searching. For example, when acquiring the spine center of the cross-sectional image showing the 18th spine, the database search means 5u may search only the 18th spine data. This makes it possible to reflect, in the search target data, a specific symptom or tendency at a site where the center of the spine to be actually obtained exists, which is advantageous in that the estimation accuracy can be improved. Not only the data with the same spine number but also the data near the spine number (for example, data of Nos. 17, 18, and 19) may be searched.

  As described above, according to the embodiment, the ideal spine line data forming unit 5p for forming the estimated spine line data is provided, the ideal spine line data is input to the display data generating unit 5b, and the projector 2 (display unit) ) To project the estimated spine line that identifies and displays the spinal curve site on the back of the human body immediately after imaging, so that the state of the spine curve site can be confirmed on the back of the human body. Since highly versatile equipment can be used, the body surface can automatically determine the degree of asymmetry of the circle at a low cost. Since the degree of curvature is obtained quantitatively, a candidate for spondylosis can be objectively and accurately determined.

  The present invention has been described based on the embodiments. Those skilled in the art will understand that these embodiments are exemplifications, and that various modifications can be made to combinations of the respective constituent elements and processing processes, and such modifications are also within the scope of the present invention. By the way.

  1) In this embodiment, when underwear is worn, a large number of small protrusions due to wrinkles of the underwear may appear on the cross section, but data based on such small protrusions can be eliminated by smoothing filter processing or the like. In addition, a dent on the spine does not appear due to the tension of the underwear, but a double ellipse can be formed by bisecting it at the center of the measurement line of the cross section. It is not limited.

  2) In this embodiment, the spine curve part is defined as a part where the mean square error of the single ellipse and the cross section exceeds the threshold value, and the spine estimation line near the boundary between the normal part and the spine curve part is not correct. Although it changes naturally (discontinuity), a significant part of the change can be corrected to a smooth estimated spine line by a filtering (flattening) process, and such a configuration is included in this embodiment.

  3) In this embodiment, since the back of the human body is expressed as a Delaunay diagram, the accuracy of the cross-sectional data can be improved using this Delaunay diagram data. However, when the Delaunay diagram is not used, It is also possible to specify the cross section data from the positions of the dots projected side by side, and in the present invention, Delaunay mapping is not essential for the formation of the cross section data.

  4) In this embodiment, the cross-sectional data is formed at intervals of 1 cm. However, if the accuracy is not questioned, it is possible to increase the interval and reduce the number of stages. When the identification output is generated even once without being based on the generation of the output, it is necessary to determine spondylosis, and such a configuration is also included in the present embodiment.

  5) In the present embodiment, the case where the dot pattern projected on the back of the human body is simultaneously imaged from a position separated by two cameras to form three-dimensional data has been described. The range of the original data does not matter. Therefore, if accuracy is not required, it is possible to obtain two pieces of image data by imaging with one stereo camera. Further, if accuracy is not an issue, even if it is imaging data of one camera, it is possible to form cross-sectional data from the projection position and imaging position of the dot pattern.

  6) In the present embodiment, the three-dimensional data on the back of the human body is obtained from the imaging screen obtained by imaging the dot pattern on the back of the human body from different positions at the same timing. In this embodiment, the “rear three-dimensionalization means” includes a first memory 5c, a second memory 5d, a two-dimensionalization means 5e, and a three-dimensionalization means 5f.

  7) In the present embodiment, the case where the single ellipse approximating means 5j and the double ellipse approximating means 5l specify the single ellipse and the double ellipse using the least square method has been described. You may specify with reference. In this case, the database search means 5u or a dedicated ellipse search means (not shown) is prepared and searched. Specifically, the characteristics of the subject may be acquired from the user, and the major axis and minor axis of the ellipse in the database 5v may be acquired using the characteristics as a key.

  8) In the present embodiment, the case where the database 5v is created in advance has been described, but the data stored in the database 5v is acquired on the assumption that the consent of the subject has been obtained. You may make it add and update whenever it does, or at arbitrary timings. This is performed using database update means (not shown). Since the data stored in the database 5v can be increased, it is advantageous in that the reliability of the database can be improved. Further, in order to erase the data stored in the database 5v, data erasing means (not shown) may be prepared. This is advantageous in that the quality of the database can be maintained, for example, by preventing data in a specific area from being biased.

  5a Timing pulse generation means, 5b Display data generation means, 5e Two-dimensionalization means, 5f Three-dimensionalization means, 5h Delaunay mapping means, 5i Cross section data formation means, 5j Single ellipse approximation means, 5k identification means, 5l 2 Double ellipse approximation means, 5m spine position estimation means, 5o estimation spine line data formation means, 5p ideal spine line data formation means, 5q comparison operation means, 5r determination means, 5s record means, 5t parameter acquisition means, 5u database search means, 5v database.

Claims (4)

Cross section data forming means for forming a plurality of cross section data including at least part of the cross section of the human body from the three-dimensional data on the back of the human body;
A single ellipse approximation means for specifying a single ellipse by approximating the cross section data;
An identification means for identifying a spinal curve portion where an error between the cross-sectional data and the single ellipse is equal to or greater than a preset threshold value and generating an identification output;
A spondylosis determination support apparatus, comprising: a spondylosis candidate acquisition unit that determines and acquires a spondylosis candidate based on the generation of an identification output. A double ellipse approximation means for specifying two ellipses by dividing the cross-sectional data and performing approximate calculation processing, respectively;
A spine position that estimates and specifies the center position of the spine on the short axis of the single ellipse in a normal part and estimates the center position of the spine from a straight line connecting the intersection points of the two ellipses in the spine curved part An estimation means;
Estimated spine line data forming means for specifying an estimated spine line based on the center position of the spine;
The spondylosis determination support apparatus according to claim 1, further comprising: the estimated spine line and display data generation means for supplying to the display means.   The spine position estimating means includes a database including at least a central position of the spine acquired from a subject known to be healthy and a central position of the spine acquired from a subject known to have spondylosis. The spinal curvature determination support apparatus according to claim 2, wherein the central position of the spine is acquired by referring to the spine. Projection means for projecting a dot pattern on the back of the human body;
Imaging means for outputting an image obtained by photographing the back of the human body in a state where the dot pattern is projected from two different directions;
The spine according to any one of claims 1 to 3, further comprising a back three-dimensionalization unit that outputs a spatial position of each dot constituting a dot pattern from the simultaneously imaged images as three-dimensional data. A curvature determination support device.

Images