Medical image processing apparatus, medical image processing method, and medical image processing program

Abstract

To provide a medical image processing device capable of generating a contour line in which a user's intention is reflected while taking into account a pixel value of a position through which the contour line passes.SOLUTION: A medical image processing device includes a processing unit. The processing unit acquires volume data on a subject; generates a first image representing part of a tissue in the subject on the basis of the volume data; generates a contour line for connecting a first point and a second point on the contour of the tissue on the first image on the basis of at least one of Laplacian and a gradient of a pixel value of each pixel in the first image, a pixel value of the second point, and a pixel value of each pixel on a candidate route connecting the first point and the second point; and causes a display unit to display the first image and the contour line.SELECTED DRAWING: Figure 11

Description

Technical Field

[0001] The present disclosure relates to a medical image processing apparatus, a medical image processing method, and a medical image processing program.

Background Art

[0002] Conventionally, a tool of the Intelligent Scissors method has been known (see Non-Patent Documents 1 and 2). The Intelligent Scissors method, also called the Livewire method, is an algorithm that automatically generates a contour from a manually placed node via an operation unit in a two-dimensional image.

Prior Art Documents

Non-Patent Documents

[0003]

Non-Patent Document 1

Non-Patent Document 2

Summary of the Invention

[0004] In conventional Intelligent Scissors methods, even if a user specifies a predetermined image position on a 2D image, the pixel value at that user-specified position is not taken into consideration when generating contours. As a result, contours are generated without reflecting the user's intentions, and the contours the user wants to observe may not be obtained. Furthermore, contour lines may be generated in areas with strong edges within the 2D image, preventing the user from obtaining the contours they want to observe. In addition, if there are areas with strong edges in areas where the display is saturated due to WW (Window Width) / WL (Window Level) processing (i.e., areas with pixel values ​​larger than the displayable pixel values), contour lines may be generated in areas that the user cannot perceive, preventing the user from obtaining the contours they want to observe.

[0005] This disclosure is made in view of the above circumstances and provides a medical image processing apparatus, a medical image processing method, and a medical image processing program that can generate contour lines that reflect the user's intentions by taking into account the pixel values ​​at the positions through which the contour lines pass. [Means for solving the problem]

[0006] One aspect of the present disclosure is a medical image processing device comprising a processing unit, the processing unit acquiring volume data of a subject, generating a first image representing a portion of the tissue within the subject based on the volume data, and connecting a first point and a second point on the contour of the tissue in the first image. A local cost is calculated for each pixel on the path candidate, and the local cost includes a first cost based on at least one of the Laplacian and gradient of the first image at that pixel, and a second cost based on the difference between the pixel value of that pixel and the pixel value of the second point, and based on the sum of the local costs calculated for each pixel on the path candidate, a contour line connecting the first point and the second point is drawn. This is a medical image processing device that generates and displays the first image and the contour line on a display unit, and the second point is a point specified by the user via an operation unit.

[0007] One aspect of the present disclosure is a method comprising the steps of: acquiring volume data of a subject; generating a first image representing a portion of the tissue within the subject based on the volume data; and connecting a first point and a second point on the contour of the tissue in the first image. A local cost is calculated for each pixel on the path candidate, and the local cost includes a first cost based on at least one of the Laplacian and gradient of the first image at that pixel, and a second cost based on the difference between the pixel value of that pixel and the pixel value of the second point, and based on the sum of the local costs calculated for each pixel on the path candidate, a contour line connecting the first point and the second point is drawn. The steps of generating and the FirstA medical image processing method comprising the steps of displaying an image and the contour line on a display unit, wherein the second point is a point specified by the user via an operation unit.

[0008] One aspect of this disclosure is a medical image processing program for causing a computer to perform the above-described medical image processing method. [Effects of the Invention]

[0009] According to this disclosure, by taking into account the pixel values ​​at the positions through which the contour lines pass, it is possible to generate contour lines that reflect the user's intentions. [Brief explanation of the drawing]

[0010] [Figure 1] Block diagram showing an example of the hardware configuration of a medical image processing device in the first embodiment. [Figure 2] Block diagram showing an example of the functional configuration of a medical image processing device. [Figure 3] The first explanatory diagram shows the difference between the first example of contour lines generated in this embodiment and the first example of contour lines created in the comparative example. [Figure 4A] Figure showing a second example of the contour lines generated in this embodiment. [Figure 4B] Figure showing a second example of the contour lines generated in the comparative example. [Figure 5] The first explanatory diagram shows that the candidate contour lines differ depending on the candidate position of the endpoint node. [Figure 6] The second explanatory diagram shows that the candidate contour lines differ depending on the candidate position of the endpoint node. [Figure 7] A diagram showing the first example of adding a node. [Figure 8] A diagram showing a second example of adding a node. [Figure 9] A diagram showing the first example of moving an existing node. [Figure 10] A diagram showing a second example of moving an existing node. [Figure 11] Flowchart showing an example of operation of a medical image processing device. [Figure 12] A diagram showing an example of the shape of an outline. [Modes for carrying out the invention]

[0011] The embodiments of this disclosure will be described below with reference to the drawings.

[0012] (First embodiment) Figure 1 is a block diagram showing an example configuration of a medical image processing device 100 in the first embodiment. The medical image processing device 100 includes a port 110, a UI 120, a display 130, a processor 140, and memory 150.

[0013] A CT scanner 200 is connected to the medical image processing device 100. The medical image processing device 100 acquires volume data from the CT scanner 200 and processes the acquired volume data. The medical image processing device 100 may also consist of a PC and software installed on the PC.

[0014] The CT scanner 200 irradiates the subject with X-rays and acquires an image (CT image) by utilizing the differences in X-ray absorption by the tissues within the body. The subject may include living organisms, humans, or animals. The CT scanner 200 acquires a synogram from the X-ray detector and generates a tomographic image (also called a slice image or slice data) of the subject by image reconstruction based on the synogram. Based on the slice data, the CT scanner 200 generates volume data, for example, by stacking the slice data. The slice data and volume data include information about any location inside the subject. The CT scanner 200 transmits the volume data as a CT image to the medical image processing device 100 via a wired or wireless line. When acquiring a CT image, imaging conditions related to CT imaging and contrast conditions related to the administration of contrast agents may be considered. Contrast enhancement may be performed on blood vessels, digestive organs, bile ducts, etc. Contrast enhancement may be performed multiple times at different timings depending on the characteristics of the organ.

[0015] Port 110 within the medical image processing device 100 includes a communication port, an external device connection port, or a connection port to an embedded device, and acquires volume data obtained from CT images. The acquired volume data may be immediately sent to the processor 140 for various processing, or it may be stored in the memory 150 and then sent to the processor 140 for various processing when needed. The volume data may also be acquired via a recording medium or recording media. Furthermore, the volume data may be acquired in the form of intermediate data, compressed data, synograms, or slice data. In addition, the volume data may be acquired from information from a sensor device attached to the medical image processing device 100. Port 110 functions as an acquisition unit that acquires various data such as volume data.

[0016] The UI120 may include a touch panel, a pointing device, a keyboard, or a microphone. The UI120 accepts arbitrary input operations from the user of the medical image processing device 100. The user may include a physician, a radiologist, a student, or other medical professional (paramedic staff).

[0017] UI120 accepts various operations. For example, it accepts operations such as specifying a region of interest (ROI) and setting brightness conditions (e.g., window information) for volume data and images based on volume data (e.g., 3D images and 2D images described later). The region of interest may include areas of various tissues (e.g., blood vessels, bronchi, organs, tissues, bones, brain). Tissues may include diseased tissue, normal tissue, or tumor tissue. Window information includes at least one of the window width (WW) and window level (WL), and is also written as "WW / WL". Window information is information that adjusts the brightness of the displayed image.

[0018] The display 130 may include, for example, an LCD, and displays various information. The various information may include three-dimensional images and two-dimensional images obtained from volume data. The three-dimensional images may include volume rendering images, surface rendering images, virtual endoscopic images, virtual ultrasound images, or CPR images, etc. The volume rendering images may include RaySum images, MIP images, MinIP images, average value images, or raycast images, etc. The two-dimensional images may include axial images, sagittal images, coronal images, or MPR images, etc.

[0019] Memory 150 includes primary storage devices such as various ROMs and RAMs. Memory 150 may also include secondary storage devices such as HDDs and SSDs. Memory 150 may also include tertiary storage devices such as USB memory or SD cards. Memory 150 stores various information and programs. The various information may include volume data acquired by port 110, images generated by processor 140, configuration information set by processor 140, and various programs. Memory 150 is an example of a non-transient recording medium on which programs are recorded.

[0020] The processor 140 may include a CPU, DSP, or GPU. The processor 140 functions as a processing unit 160 that performs various processing and control by executing a medical image processing program stored in memory 150.

[0021] Figure 2 is a block diagram showing an example of the functional configuration of the processing unit 160.

[0022] The processing unit 160 comprises a region processing unit 161, an image generation unit 162, a contour line processing unit 163, and a display control unit 165. The processing unit 160 coordinates all parts of the medical image processing device 100. The processing unit 160 performs processing related to contour line generation and visualization, for example. Note that each part included in the processing unit 160 may be implemented as a different function by a single piece of hardware, or as different functions by multiple pieces of hardware. Furthermore, each part included in the processing unit 160 may be implemented by dedicated hardware components.

[0023] The region processing unit 161 acquires volume data of the subject, for example, via port 110. The region processing unit 161 extracts any region included in the volume data. The region processing unit 161 may automatically specify a region of interest and extract it, for example, based on the voxel values ​​of the volume data. The region processing unit 161 may manually specify a region of interest and extract it, for example, via UI 120.

[0024] The image generation unit 162 generates various images. Based on at least a portion of the acquired volume data (for example, volume data of an extracted region), the image generation unit 162 generates 3D images, 2D images, and tomographic images. The image generation unit 162 may generate images by performing various renderings (for example, volume rendering or surface rendering).

[0025] The contour processing unit 163 performs contour processing to visualize the outer perimeter of various tissues (e.g., kidneys) within the image. Details of the contour processing will be described later.

[0026] The display control unit 165 displays various data, information, or images on the display 130. The images represent a part of the tissue within the subject and may include, for example, images generated by the image generation unit 162, cross-sectional images of a predetermined cross-section (MPR images), and tomographic images of a predetermined torso. The display control unit 165 displays contour lines along with these images.

[0027] Next, we will explain the processing related to contour lines.

[0028] The contour processing unit 163 generates contour lines that visualize the outer perimeter of any tissue on a 3D image, 2D image, or tomographic image, for example, by modifying the IntelligentScissors method. The 3D image, 2D image, or tomographic image on which contour lines are generated is also referred to here as the target image. The pixel values ​​of the target image may be scalar values ​​or may consist of multiple values ​​(multimodality data, dual-energy CT, spectral CT, etc.).

[0029] The contour processing unit 163 may set (place) multiple nodes at user-desired locations on the target image via the UI 120. These nodes include at least two: a start node indicating the start point of the contour line and an end node indicating the end point of the contour line. In addition to the start node and end node, the nodes may also include intermediate nodes. The contour processing unit 163 may sequentially specify arbitrary locations on the target image via the UI 120 and set the nodes sequentially. In this case, the first specified location becomes the start node, and the last specified location becomes the end node. For example, the contour processing unit 163 may set nodes by accepting a click operation at a desired location on the target image via the UI 120. The pixel location on the target image specified by the click operation is also called the click point. The contour processing unit 163 generates a contour line that passes through the set nodes.

[0030] Furthermore, the contour processing unit 163 may sequentially change the candidate positions of the endpoint nodes of the contour by accepting a drag operation at a desired position on the target image via the UI 120. In other words, a node may be specified with a click operation, while a candidate endpoint node that will be the endpoint of the contour may be specified without specifying a node with a drag operation. The contour processing unit 163 may, for example, set (confirm) the position of the last candidate endpoint node as the endpoint node position by receiving an instruction to confirm the endpoint node via the UI 120. Specifying a candidate endpoint node may, for example, be done by ending the drag operation or performing a double-click operation.

[0031] In addition, the contour line processing unit 163 derives (for example, calculates) a contour line connecting any two points (a starting node and an ending node) according to, for example, the shortest path search method. This shortest path search method may include, for example, Dijkstra's method or the A* method (A star). The contour line processing unit 163 defines a cost for each pixel of the target image and calculates a path with the minimum cost between the above-mentioned arbitrary starting node and ending node. Further, the contour line processing unit 163 may derive a candidate for the contour line based on the cost between the ending node and the candidate position of the ending node before the ending node is determined, in the same manner as the derivation of the contour line based on the cost between any two points (the starting node and the ending node).

[0032] The contour line processing unit 163 calculates the cost in consideration of, for example, the following cost function. The cost function is, for example, f Z function and f G function and f D function and at least one of f X function, and includes Z This f G function, f D function, and f Z function may be the same as the f G function, f D [[ID=?]]function described in Non-Patent Document 1.

[0033] The f Z function is Laplacian Zero-crossing, that is, a binary function indicating an edge extracted by a Laplacian filter. The f Z function incurs a cost at a location with a peak (maximum pixel value) or valley (minimum pixel value) formed by the pixel values of each pixel on the target image, and adjusts so that the contour line does not cross this peak or valley.

[0034] The f G function is Gradient Magnitude, that is, it indicates the strength of the gradient. The smaller the value (cost) derived by the f G function, the easier it is for the contour line to be formed toward pixels with a strong shading difference (that is, a large difference in pixel values between adjacent pixels) in the target image. On the other hand, f G It seems there is an error in the original text at line . It should be something like "This f function, f function, and f function may be the same as the f function, f function, and f function described in Non-Patent Document 1." which I have tried to correct and translate as best as possible in the above translation. Please check and let me know if there are any further issues or if the translation needs to be adjusted based on the correct original text.The larger the value derived by the function, the easier it is for contour lines to be formed so that they do not converge towards pixels with strong contrast in the target image. The gradient can be calculated using a first-order differential filter.

[0035] f D The function is Gradient Direction, that is, it indicates the direction of the gradient. D The smaller the value (cost) derived by the function, the easier it is for contour lines to be formed in the target image, moving towards pixels with strong contrast (i.e., large differences in pixel values ​​between adjacent pixels) while avoiding crossing edges. On the other hand, f D The larger the value derived by the function, the easier it becomes for contour lines to form, extending over edges towards pixels with strong contrast differences in the target image.

[0036] Also, PH X The function is a function that shows the cost of the pixel value of each pixel on the target image. The smaller the difference in pixel values ​​between each point on the contour line and the click point (end node), the lower the cost of f. X Since the value (cost) derived by the function becomes smaller, contour lines are more easily formed for pixel values ​​close to the click point (endpoint node). In other words, the contour processing unit 163 specifies an arbitrary point on the target image (for example, a click point) via the UI 120 and uses the pixel value of the specified point as the cost.

[0037] (First example of contour line derivation) First, let's explain the first example of contour line derivation.

[0038] The contour processing unit 163 calculates, for example, the local cost of moving from point p on the target image to point q (adjacent point) adjacent to point p, by assigning weights to each cost. In other words, the local cost is the cost required to move one pixel on the target image. The contour processing unit 163 may calculate the local cost according to, for example, (Equation 1). l(p,q)=ω Z ×f Z (q)+ω D ×f D (p,q)+ω G ×fG (q)+ω X ×f X (q,d)...(Formula 1)

[0039] l(p,q) represents the cost (local cost) from point p to point q. Z (q) represents the edge obtained by the Laplacian filter at point q. D (p,q) is the direction of the gradient between point p and point q. G (q) represents the intensity of the gradient at point q. Z (q), f D (p,q), and f G (q) is f described in Non-Patent Document 1 Z (q), f D (p,q), and f G (q) is the same as (q). Also, f X (q,d) is the difference in pixel values ​​between point q and point d. Point d is a candidate position for the endpoint node on the target image. In other words, if X(a) is the pixel value at point a, then f X (q,d) can be expressed as, for example, (Equation 2). f X (q,d)=|X(q)-X(d)| (Formula 2)

[0040] Also, ω Z , ω D , ω G , and ω X These are, respectively, f Z function, f G function, f D Functions, and f X This indicates the weights for a function. For example, ω Z =0.43, ω D =0.43, ω G =0.14, ω X = 0.3, but not limited to this value.

[0041] The contour processing unit 163 specifies the starting node of the contour line, for example via the UI 120, and first sets this starting node to point p, and sets the point adjacent to point p to point q, and calculates the local cost l(p,q). Point q can be any point adjacent to point p in any direction on the target image.

[0042] Next, the previous point q is set to point p, and the point q adjacent to point p is set to this point q, and the local cost l(p,q) is recalculated. The contour processing unit 163 repeats this setting of points p and q and calculation of local costs until the position of point q reaches point d, which is a candidate position for the endpoint node. The contour processing unit 163 then calculates the sum of the local costs (also called the total cost) from the starting node to the candidate position for the endpoint node. The contour processing unit 163 generates a contour line by deriving the contour line that minimizes the total cost. In this case, if there is a node (intermediate node) between the starting node and the endpoint node, the contour line that minimizes the total cost among the contour lines that pass through this intermediate node is derived. Note that one of the candidate positions for the endpoint node is determined to be the endpoint node, so the cost calculation based on the candidate position for the endpoint node is equivalent to the cost calculation based on the endpoint node.

[0043] Furthermore, the candidate positions of the endpoint nodes can be changed, for example, by operating via UI120. In this case, each time the candidate position of the endpoint node changes, points p and q are set, the local cost is calculated, and the total cost is calculated. As a result, for example, when the candidate position of the endpoint node changes via UI120, the total cost changes, and the shape of the candidate contour line changes. When the contour line processing unit 163 receives an instruction to confirm the endpoint node via UI120, it confirms the position of the confirmed endpoint node with the contour line that would have been used if the candidate position of the endpoint node had been used, and generates the contour line. The display control unit 165 may display the appearance of the candidate contour line whose shape is changing.

[0044] Furthermore, the contour processing unit 163 may generate contour lines whose total cost satisfies a predetermined criterion (for example, the total cost being less than a threshold), rather than deriving the contour line with the minimum total cost. Also, the contour processing unit 163 calculates the cost function (f) of each cost included in (Equation 1). Z function, f G function, f D Functions, and f X In addition to the function, the local cost may also be calculated by taking into account the cost function that takes into account other points p and q (i.e., denoted by (p,q)).

[0045] Furthermore, if the target image is a color image, the contour processing unit 163 may calculate the total cost for each color (e.g., RGB) and then calculate a total cost (also called the total color cost) by summing the total costs for each color. Alternatively, the contour processing unit 163 may convert the color image to HSV, calculate the total cost for each HSV component, and use the total cost for H as the main cost, and the total costs for S and V as sub-costs (used auxiliaryly) to calculate the total color cost. The contour processing unit 163 may generate contours that minimize the total color cost.

[0046] Here, the starting node and ending node are exemplified as the start and end points of the contour line. However, if there are three or more nodes, the contour line processing unit 163 may designate the starting node of two adjacent nodes and the ending node of the other two nodes. The contour line processing unit 163 may then derive a partial contour line (a part of the contour line) connecting these two adjacent nodes. In this case as well, the local cost and total cost described above may be derived, and the partial contour line may be determined based on the total cost. The contour line processing unit 163 may then connect multiple partial contour lines connecting adjacent nodes to generate an overall contour line OL connecting the start and end points of three or more nodes.

[0047] (Second example of contour line derivation) Next, we will explain the second example of contour line derivation. Note that explanations of matters similar to those explained in the first example of contour line derivation will be omitted or simplified.

[0048] The contour processing unit 163 may calculate the local cost from point p to point q adjacent to point p on the target image, similar to the first example of contour derivation, by assigning weights to each cost. The contour processing unit 163 may calculate the local cost according to, for example, (Equation 3). In other words, (Equation 3) is a simplified equation compared to (Equation 1). l(p,q)=1+ω D ×f D (p,q)+ω X ×f X (q,d)...(Formula 3)

[0049] Regarding weights, for example, ω D =1.0, ω X = 10.0. Also, in the second example, f X (q,d) has a different definition than in the first example. That is, f X (q,d) is not expressed by (Equation 2), but rather by, for example, (Equation 4) below. f X (q,d)=|fs(X(q))-fs(X(d))| (Equation 4)

[0050] Here, fs(i) is a function that normalizes (feature scales) a given pixel value i to [0,1] while clipping it within the WW / WL range (set window information). Specifically, fs(i) can be expressed, for example, by (Equation 5) below. (Equation 5) is shown with case distinctions.

[0051] fs(i) = 0; if i <WL-WW / 2 fs(i)=((i-WL) / WW+1 / 2); if WL-WW / 2≦i≦WL+WW / 2 fs(i)=1; if WL+WW / 2

[0052] Note f D ​(p,q) is f as described in Non-Patent Document 1 D (p,q) may be the same as or different from (p,q). For example, it may be expressed in a simplified form as in (Equation 6) below, or as in (Equation 7). f D (p,q)=|fs(X(p))-fs(X(q))| (Equation 6) f D (p,q)=exp(-20.0*(||grad(q)×uv(p,q))||^2) ···(Equation 7)

[0053] Here, "^" represents the exponentiation operator. "grad(q)" is the gradient vector of the target image at point q. "×" is the cross product operator. "*" is the multiplication operator. "uv(p,q)" is the unit vector in the direction from point p to point q. norm(v) indicates the magnitude of the vector, and "||" represents the norm symbol.

[0054] In the second example, as in the first example of contour line derivation, the contour line processing unit 163 calculates the total cost, which is the sum of the local costs from the starting node to the candidate position of the ending node. Then, the contour line processing unit 163 generates the contour line by deriving the contour line that minimizes the total cost.

[0055] The contour line processing unit 163 may generate a linear contour line (linear contour line) by connecting each node between the starting node and the ending node in series, or it may generate a ring contour line (ring contour line) by connecting each node in a ring. Both linear and ring contour lines may be derived and generated in such a way that the total cost described above is minimized. The contour line processing unit 163 may have a linear contour line mode and a ring contour line mode as operating modes for generating contour lines. When the operating mode is the linear contour line mode, the contour line processing unit 163 may calculate the cost for generating a linear contour line and generate a linear contour line. When the operating mode is the ring contour line mode, the contour line processing unit 163 may calculate the cost for generating a ring contour line and generate a ring contour line.

[0056] Next, we will provide additional information regarding the medical tomographic images obtained by the CT scanner 200 and the window information.

[0057] Unlike general photographic images, medical tomographic images do not produce shadows from the light source within the image. As a result, no gradient occurs along the contour of the object being observed, and the contour of tissue often has the same pixel value. It can also be inferred that the pixel value of the point specified by the user contains information about the contour the user wants to observe. Furthermore, tissues with the same composition within the same image often have the same pixel value. In addition, the user can clip the range of pixel values ​​of the displayed image by specifying WW / WL via UI120. In this case, the contour line processing unit 163 may estimate the WW / WL to be generated as a contour line based on the specified WW / WL information and generate the contour line.

[0058] Furthermore, in CT images, the composition of the corresponding tissue can be generally determined by the CT value (pixel value obtained by imaging with the CT device 200). For example, for the contours of organs, it is generally sufficient for the CT values ​​to be concentrated in the range of 50 to 100. Also, bone, metal, and air have larger absolute CT values ​​compared to other substances. For example, metal has a CT value of 1000 or more, bone has a CT value of 100 to 1000, and air has a CT value of -1000. Therefore, extreme gradients can occur at the boundaries of bone, metal, and air. Conventional Intelligent Scissors methods are easily affected by these large gradients, but the medical image processing device 100 of this embodiment can reduce the influence of these large gradients and generate contour lines of organs.

[0059] Furthermore, the contour processing unit 163 can generate contour lines based on WW / WL. WW / WL is often specified by the user and reflects the range of pixel values ​​that the user wants to observe. Therefore, the medical image processing device 100 can prioritize tissues, etc., within that range of pixel values ​​and generate contour lines that pass through pixels having pixel values ​​within that range, for example. This allows the medical image processing device 100 to suppress the generation of contour lines at the boundaries of bone, metal, or air, for example, due to the influence of pixel values ​​such as bone, metal, or air, where the absolute value of the CT value is extremely large.

[0060] Figure 3 is the first explanatory diagram showing the difference between the first example of contour lines generated in this embodiment and the first example of contour lines created in the comparative example. In Figure 3, we want to create the contour of the inner wall of the angiographic aorta where it leaves the heart, and the calcifications, fatty tissue, and angiographic atria on the aorta are visible around the contour of the inner wall of the aorta.

[0061] Region R1 is a calcified region in the blood vessel wall, and a decrease in brightness is observed outside the blood vessel. As shown in region R1, the contour line OLX of the comparative example tends to be drawn towards (sucked into) the outside of the blood vessel with a steeper gradient in the target image G1. In contrast, the contour line OL of this embodiment can be made less affected by the gradient. This is because of the f in (Equation 1) and (Equation 3). X function (f X This is due to the action of (g,d). In other words, the contour line processing unit 163 takes into account the pixel value of point d, which is the endpoint node EN, and generates the contour line OL such that the cumulative difference between the pixel value of each point on the contour line OL and the endpoint node EN (point d) is as small as possible. As a result, behavior such as the contour line OLX in the comparative example, where the pixel values ​​are drawn towards the edge side where they are significantly different, is suppressed.

[0062] Furthermore, in region R1, a large drop in brightness, presumably due to fat, is observed outside the blood vessel wall. As shown in region R2, the contour line OLX of the comparative example tends not to maintain pixel values ​​(CT values) at each point through which the contour line OLX passes. That is, it passes near the edge, but passes through pixels with low pixel values ​​(black) or pixels with high pixel values ​​(white). In contrast, the contour line OL of this embodiment tends not to fluctuate much at each point through which the contour line OL passes, depending on the pixel value of the endpoint node EN, and the pixel value tends to be maintained at each point through which the contour line OL passes. This is due to the f in (Equation 1) and (Equation 3). X function (f X This is due to the action of (g,d). In other words, the contour line processing unit 163 takes into account the pixel value of the endpoint node EN and generates the contour line OL such that the cumulative difference between the pixel value of each point on the contour line OL and the endpoint node EN is as small as possible. Therefore, it avoids as much fluctuation in pixel values ​​as possible, as seen in the contour line OLX of the comparative example.

[0063] As shown in the comparative example, in the conventional shortest path search method, the cost function for generating the contour line OLX is not affected by the start and end points of the contour line. This is because, for example, in the cost function of Patent Document 1, the f shown in (Equation 1), etc. X This is due to the absence of a function. In contrast, the medical image processing device 100 of this embodiment has f X Due to the action of (q,d), the pixel values ​​of each point (point q) adjacent to the contour line OL (or candidate contour line) and the pixel values ​​of the endpoint node EN (or its candidate position) (point d) are influenced, making it easy to generate a desired contour line OL (or candidate contour line OL) that reflects the user's intentions. Furthermore, in medical images, it is significant to generate a contour line OL that connects pixels having the same or equivalent pixel values ​​as a predetermined threshold, and such a contour line OL is obtained in this embodiment.

[0064] Figure 4A shows a second example of the contour line generated in this embodiment. Figure 4B shows a second example of the contour line generated in the comparative example.

[0065] In this embodiment, contour line OL1 is generated based on the pixel value of the endpoint node EN (point d). Therefore, if the user wants to place contour line OL1 larger than the tissue (slightly outside the tissue), the medical image processing device 100 can generate contour line OL1 as intended by the user. On the other hand, in the comparative example, contour line OLX1 is generated along the area with the greatest gradient, regardless of the pixel value of the endpoint node EN. Therefore, in the comparative example, the contour line extends from the low-pixel-value area (black) into the high-pixel-value area (white), but in this embodiment, contour line OL1 can be generated so that it does not extend from the low-pixel-value area (black) into the high-pixel-value area (white).

[0066] Figure 5 is the first explanatory diagram showing that the candidate contour lines (OLC) differ depending on the candidate endpoint node location (ENC).

[0067] Figure 5 shows candidate contour line OLC1 when candidate endpoint node position ENC1 is clicked in target image G11. It also shows candidate contour line OLC2 when candidate endpoint node position ENC2 is clicked in target image G12. Furthermore, it shows candidate contour line OLC3 when candidate endpoint node position ENC3 is clicked in target image G13. In other words, by sequentially changing the candidate endpoint node positions ENC(ENC1, ENC2, ENC3) in target images G11-G13, the candidate contour line OLC(OLC1, OLC2, OLC3) changes and is displayed sequentially. Note that target images G11-G13 are the same image.

[0068] In each of the target images G11, G12, and G13 in Figure 5A, the shape of the candidate contour line OLC changes slightly as the candidate endpoint node EDC is finely adjusted, for example, by dragging via UI120. Such fine adjustments are useful, for example, when the contour line processing unit 163 enlarges the target images G11, G12, and G13 to adjust the contours of small tissues.

[0069] Figure 6 is a second explanatory diagram showing that the candidate contour lines (OLC) differ depending on the candidate endpoint node location (ENC).

[0070] In target image G22, compared to target image G21, the candidate contour line OLC attached to target image G22 has moved upward by approximately 0.5 pixels. In other words, the position of candidate contour line OLC21 in target image G21 moves upward in the image due to drag operations via UI120, causing it to move to the position of candidate contour line OLC22 in target image G22. That is, a slight difference in the position of the click point in the target image results in a slight difference in the pixel value, which causes the position of the candidate contour line OLC to move slightly upward. This is because, in this embodiment, the pixel value within the target image is used in the cost function. On the other hand, in the conventional Intelligent Scissors method, even if the candidate position ENC of the endpoint node was moved, most of the contour line remained in the same place, making fine adjustment of the contour line impossible. Furthermore, when the endpoint node was moved, the contour line sometimes moved significantly against the user's intention. This is because the cost function is prone to falling into a local minimum. Note that target images G21 and G22 represent the same image. The medical image processing device 100 of this embodiment allows for easy fine-tuning of the position of the candidate contour line OLC by fine-tuning the user's click position, that is, the position of the contour line OL can be fine-tuned.

[0071] Next, we will explain how to add a new node.

[0072] The contour processing unit 163 generates a contour line OL based on at least two nodes. If two nodes are specified, the node specified earlier in the time series becomes the first starting node SN, which is the starting point of the contour line OL, and the node specified later in the time series becomes the ending node EN, which is the ending point of the contour line OL. The positions of the adjacent pixels between the starting node SN and the ending node EN become points p and q, which appear in (Equation 1) or (Equation 3), etc. If it is desired to extend this contour line OL, the contour processing unit 163 generates a new contour line by using the previous ending node as the new starting node SN and the newly specified node as the new ending node EN, and extends the contour line OL by connecting the new contour line to the previous contour line OL. Also, if the newly specified node is the first starting node SN of the previous contour line OL, the contour processing unit 163 can connect the ending point of the contour line OL to the starting point to generate a closed curve, making the area inside it a region (closed region).

[0073] The contour processing unit 163 may modify the contour line OL (e.g., OL2, OL3, ...) after a contour line OL has been generated based on at least two nodes, a starting node SN and an ending node EN, by adding one or more intermediate nodes IN. When intermediate nodes IN are added, partial contour lines OLP (e.g., OLP1, OLP2, ...) are generated between the intermediate node IN and each of the two nodes on either side of the intermediate node IN (e.g., the starting node SN, the ending node EN, or another intermediate node IN), and the total cost is calculated using a cost function.

[0074] Figure 7 shows a first example of adding a node. In Figure 7, it is assumed that one intermediate node IN is added between the starting node SN and the ending node EN. In this case, partial contour line OLP1 is determined based on the total cost of using the starting node SN as the starting node and the intermediate node IN as the ending node, so as to minimize this total cost. Similarly, partial contour line OLP2 is determined based on the total cost of using the ending node EN as the starting node and the intermediate node IN as the ending node, so as to minimize this total cost. Based on partial contour lines OLP1 and OLP2, the contour line processing unit 163 modifies the previously generated contour line OL2 into contour line OL3, which connects partial contour lines OLP1 and OLP2. In this case, the entire section of contour line OL2 between the starting node SN and the ending node EN is modified. By doing so, the medical image processing unit 100 can prioritize the pixel values ​​of the last specified intermediate node IN.

[0075] Figure 8 shows a second example of adding nodes. In Figure 8, partial contour line OLP3 passes through three nodes, and it is assumed that a second intermediate node IN2 is added between the starting node SN and the first intermediate node IN1. In this case, partial contour line OLP4 is determined based on the total cost of using the starting node SN as the starting node and the second intermediate node IN2 as the ending node, so as to minimize this total cost. Similarly, partial contour line OLP5 is determined based on the total cost of using the first intermediate node IN1 as the starting node and the second intermediate node IN2 as the ending node, so as to minimize this total cost. Based on partial contour lines OLP4 and OLP5, the contour line processing unit 163 modifies the previously generated partial contour line OLP3 into contour line OLP6, which connects partial contour lines OLP4 and OLP5. In this case, a portion of the contour line OL between the starting node SN and the ending node EN (the portion between the starting node SN and the first intermediate node IN1) is modified. By doing so, the medical image processing device 100 can prioritize the pixel value of the last specified intermediate node IN1.

[0076] Next, we will explain how to migrate existing nodes.

[0077] The contour processing unit 163 accepts user input via the UI 120 and may move nodes (existing nodes) through which the generated contour line OL passes based on this input. In this case, the contour processing unit 163 determines partial contour lines OLP (e.g., OLP11, OLP12,…) based on the moved node (moved node) and the two nodes that were adjacent on both sides of the contour line OL before the moved node was positioned, and modifies the generated contour lines OL (OL4, OL5,…).

[0078] Figure 9 shows a first example of moving an existing node. In Figure 9, three nodes are already placed and a contour line OL4 has been generated. It is assumed that an intermediate node IN11, located between the starting node SN and the ending node EN on contour line OL4, will move. The three nodes are, for example, designated and placed in the order of starting node SN, intermediate node IN11, and ending node EN. In this case, contour line OL4 may be modified based on the intermediate node IN11 as the moving node, and the starting node SN and ending node EN adjacent to the position of the intermediate node IN11 before it moved. Specifically, based on the total cost of using the starting node SN as the starting node and the intermediate node IN11 after the move as the ending node, a partial contour line OLP11 connecting the starting node SN and the intermediate node IN11 after the move is determined so as to minimize this total cost. Furthermore, based on the total cost of using the endpoint node EN as the starting node and the intermediate node IN11 as the endpoint node, the partial contour line OLP12 connecting the endpoint node EN and the intermediate node IN11 is determined so as to minimize this total cost. Based on the partial contour lines OLP11 and OLP12, the contour line processing unit 163 modifies the previously generated contour line OL4 into contour line OL5, which connects the partial contour lines OLP11 and OLP12. In this case, the entire section of contour line OL4 between the starting node SN and the endpoint node EN is modified. By doing so, the medical image processing device 100 can prioritize the pixel values ​​of the last specified intermediate node IN11.

[0079] Figure 10 shows a second example of moving an existing node. In Figure 10, four or more nodes have been placed and a partial contour line OLP13 has been generated, and it is assumed that the second intermediate node IN22, which is located between the starting node SN and the first intermediate node IN21 on the partial contour line 13, will move. The four nodes are, for example, designated and placed in the order of starting node SN, intermediate node IN21, intermediate node IN22, and ending node EN. In this case, the partial contour line OLP13 may be modified based on the second intermediate node IN22 as the moving node, and the starting node SN and the first intermediate node IN21 which are adjacent to the position of the second intermediate node IN22 before it moves. Specifically, based on the total cost of using the starting node SN as the starting node and the second intermediate node IN22 after the move as the ending node, the partial contour line OLP14 connecting the starting node SN and the second intermediate node IN22 after the move is determined so as to minimize this total cost. Furthermore, based on the total cost of using the first intermediate node IN21 as the starting node and the second intermediate node IN22 as the ending node, the partial contour line OLP15 connecting the first intermediate node IN21 and the second intermediate node IN22 is determined so as to minimize this total cost. Based on the partial contour lines OLP14 and OLP15, the contour line processing unit 163 modifies the previously generated partial contour line OLP13 into a partial contour line OLP16 that connects the partial contour lines OLP14 and OLP15. In this case, a portion of the contour line OL between the starting node SN and the ending node EN (the portion between the starting node SN and the first intermediate node IN21) is modified. By doing so, the medical image processing device 100 can prioritize the pixel value of the last specified intermediate node IN22.

[0080] Next, an example of the operation of the medical image processing device 100 will be described. Figure 11 is a flowchart illustrating an example of the operation of the medical image processing device 100. Figure 11 illustrates the generation of an annular contour line as the contour line.

[0081] First, port 110 acquires volume data of the subject (e.g., patient) (S11). The region processing unit 161 extracts a region of interest (e.g., kidney) specified via, for example, UI 120, and the image generation unit 162 generates an MPR image of a predetermined cross-section including the region of interest (S12). The display control unit 165 displays the generated MPR image (S12). This allows the user to confirm the region of interest. The region of interest is the area that the user observes.

[0082] The contour processing unit 163 accepts user input, for example via the UI 120, and sequentially places (positions) nodes on the contour of the region of interest (e.g., the kidney) (S13). For example, nodes may be sequentially placed by sequentially clicking on arbitrary positions on the MPR image via the UI 120.

[0083] The contour processing unit 163 calculates the local cost and total cost for each of the specified adjacent nodes, and generates a partial contour line based on the total cost. The contour processing unit 163 connects each of the partial contour lines that complement each adjacent node to generate a contour line from the starting node SN to the ending node EN (S14).

[0084] The contour processing unit 163 generates a closed contour line (i.e., an annular contour) passing through each node if three or more nodes are specified and the starting node, which was initially specified as the endpoint node, is specified again (S15). The display control unit 165 displays the annular contour line on the MPR image (S15).

[0085] The contour processing unit 163 displays the closed region on the MPR image based on the closed annular contour line (S16). For example, the closed region is displayed in a way that distinguishes the area outside and inside of the closed region from each other. In addition, statistical values ​​such as the area, diameter, maximum, minimum, and average values ​​of the pixel values ​​of the closed region may be displayed.

[0086] The contour processing unit 163 moves any node on the contour line generated in step S15 on the MPR image via the UI 120 (S17). The contour processing unit 163 corrects the contour line (or partial contour line) between two nodes that were adjacent to the moved node before the moved node was moved (S18). The display control unit 165 displays the corrected, closed, annular contour line on the MPR image (S19).

[0087] Next, variations of this embodiment will be described.

[0088] The contour processing unit 163 calculates the value of the local cost function based on the pixel value of the latest click point (i.e., the candidate position of the endpoint node or the confirmed position of the endpoint node at the time of the click), which is included in the local cost function. X The calculation of the value of (q,d) is illustrated as an example. Note that the contour processing unit 163 may calculate the local cost function not only based on the pixel value of the candidate position of the endpoint node, but also based on the pixel value of the node placed immediately before the candidate position of the endpoint node on the contour candidate (also referred to as the previous node). In other words, if the previous node of the candidate position of the endpoint node is the starting node, the value of the local cost function is determined based on the pixel value of the starting node and the pixel value of the candidate position of the endpoint node. If the previous node of the candidate position of the endpoint node is an intermediate node, the value of the local cost function is determined based on the pixel value of this intermediate node and the pixel value of the candidate position of the endpoint node.

[0089] Furthermore, interpolation processing is sometimes performed between adjacent pixels to adjust the pixel values ​​so that they change smoothly between adjacent pixels. For example, by setting the interpolation values ​​so that the pixel values ​​differ in a range of less than one pixel, the pixel values ​​can be made to change smoothly between adjacent pixels (e.g., curvilinearly) rather than changing abruptly (e.g., resulting in a jagged shape) on a pixel-by-pixel basis. As a result, even when the target image is enlarged, the change in the pixel values ​​of the displayed image becomes smoother. In such an enlarged image, the contour processing unit 163 specifies the endpoint node or candidate position of the endpoint node by clicking or dragging an arbitrary point via the UI 120, and uses the interpolation value of this endpoint node or candidate position of the endpoint node to f X The value of (q,d) may be calculated. This allows the medical image processing device 100 to derive contour lines that suitably follow the contours of the tissue, even in magnified images.

[0090] Furthermore, while this embodiment primarily illustrates the generation of contour lines that take into account the planar shape of tissues, etc., on a two-dimensional image, it is not limited to this. The contour line processing unit 163 may also generate contour lines that take into account the three-dimensional shape of tissues, etc., in three-dimensional space. In this case, the contour line processing unit 163 may generate three-dimensional contour lines that take into account the three-dimensional shape of tissues, etc., in three-dimensional space by applying the cost function derivation method that takes into account the pixel values ​​in the target image of this embodiment described above to the technology in Reference Non-Patent Literature 1 below.

[0091] (Reference Non-Patent Document 1: Miranda Poon, Ghassan Hamarneh, Rafeef Abugharbieh, “Efficient interactive 3D Livewire segmentation of complex objects with arbitrary topology”, ScienceDirect, Computerized Medical Imaging and Graphics, 32 (2008) 639-650)

[0092] Furthermore, while the example mainly uses an MPR image as the target cross-sectional image, it is not limited to this. For example, the target cross-sectional image may be an axial image, coronal image, sagittal image, CPR image, or other image. Also, the target cross-sectional image may be a cross-sectional image with thickness.

[0093] Furthermore, the contour line processing unit 163 is not limited to deriving, for example, a polylinear contour line as a contour line or partial contour line connecting each node. Figure 12 shows an example of the shape of a contour line. The contour line may be the aforementioned rectangular polylinear contour line OLa. Alternatively, the contour line may be a polyline contour line OLb connecting the centers of pixels. Alternatively, the polyline contour line may be a contour line OLc in which points on the polyline are moved to a position where the interpolated value obtained by interpolation (for example, linear interpolation) between adjacent pixels becomes the pixel value at the click position. Alternatively, the contour line may be a spline curve contour line OLd.

[0094] Thus, in this embodiment, the medical image processing device 100 derives the cost function value based on the pixel values ​​of the node positions that influence the shape of the contour line, so it can generate a contour line that reflects the user's intention to specify the node positions. For example, it is possible to derive a contour line that passes through pixels having pixel values ​​close to the pixel values ​​specified by the user. Furthermore, the medical image processing device 100 takes into account not only the Laplacian and gradient values, but also the proximity of pixel values ​​along the searched path, so it can reduce fluctuations in the pixel values ​​at the positions through which the contour line passes.

[0095] Furthermore, the processing unit 160 may assume the candidate position of the endpoint node is the endpoint node and calculate the total cost in the same way as for the endpoint node. In this case, the total cost will be the total cost between the starting node and the candidate endpoint node. The display control unit 165 may display the candidate contour lines based on this total cost as a preview on the display 130. The processing unit 160 may move the candidate position of the endpoint node by dragging via the UI 120, and sequentially display candidate paths (contour line candidates) until the endpoint node is confirmed. As a result, the medical image processing device 100 can visualize the state of the contour line candidates while the candidate position of the endpoint node changes sequentially, rather than the contour line shape changing discretely for each node specified. Therefore, the user can easily fine-tune the contour line while checking this display.

[0096] Although various embodiments have been described above with reference to the drawings, it goes without saying that this disclosure is not limited to such examples. It is clear to those skilled in the art that various modifications or alterations can be conceived within the scope of the claims, and these will naturally also fall within the technical scope of this disclosure.

[0097] Furthermore, the medical image processing device 100 may include at least a processor 140 and memory 150. The port 110, UI 120, and display 130 may be external to the medical image processing device 100.

[0098] Furthermore, it was illustrated that the volume data, as captured CT images, is transmitted from the CT scanner 200 to the medical image processing device 100. Alternatively, the volume data may be transmitted to a server on the network (e.g., an image data server (PACS) (not shown)) for storage. In this case, the port 110 of the medical image processing device 100 may acquire the volume data from the server via a wired or wireless line when necessary, or via any storage medium (not shown).

[0099] Furthermore, it was illustrated that the volume data of the captured CT image is transmitted from the CT scanner 200 to the medical image processing device 100 via port 110. This includes cases where the CT scanner 200 and the medical image processing device 100 are essentially combined into a single product. It also includes cases where the medical image processing device 100 is treated as the console for the CT scanner 200.

[0100] Furthermore, while the example illustrates the acquisition of images using a CT scanner 200 and the generation of volume data containing information about the inside of the subject, images may be acquired and volume data generated using other devices. Other devices include MRI (Magnetic Resonance Imaging) scanners, PET (Positron Emission Tomography) scanners, angiography scanners, or other modality devices. In addition, PET scanners may be used in combination with other modality devices.

[0101] Furthermore, the operation of the medical image processing device 100 can be expressed as a defined medical image processing method. It can also be expressed as a program that causes a computer to execute each step of the medical image processing method.

[0102] (Summary of the above embodiment) As described above, the medical image processing apparatus 100 of the above embodiment comprises a processing unit 160 and a display 130. The processing unit 160 acquires volume data of a subject and may generate a target image (an example of a first image) that represents at least a part of the tissue within the subject based on the volume data. The processing unit 160 generates a contour line OL connecting a first point (e.g., a starting node SN) and a second point (e.g., an ending node EN or an intermediate node IN) on the contour of the tissue on the target image, based on at least one of the Laplacian and gradient of each pixel of the target image, the pixel value of the second point, and the pixel value of each pixel on the candidate path connecting the first point and the second point. The processing unit 160 may specify the first point and the second point via a UI 120 (an example of an operation unit). The processing unit 160 displays the target image and the contour line OL on the display 130 (an example of a display unit). This gradient may include, for example, the magnitude of the gradient, the direction of the gradient, the gradient vector, or simply the difference in pixel values ​​of two pixels whose relative positions are known. The first and second points, as well as the third and fourth points described later, are examples of control points (nodes).

[0103] As a result, the medical image processing device 100 derives contour lines based on the pixel values ​​of pixels on the candidate path and the pixel values ​​of endpoint nodes and intermediate nodes specified by the user, thereby generating contour lines OL that reflect the user's intention in specifying the node positions. Therefore, for example, it is possible to derive contour lines that pass through pixels with pixel values ​​close to the pixel values ​​specified by the user. Furthermore, the medical image processing device 100 can generate contour lines OL by taking into account not only the Laplacian and gradient values, but also the pixel values ​​of each pixel on the candidate path.

[0104] Furthermore, the volume data may be data obtained by imaging the subject with the CT scanner 200. The target image may also be an MPR image. This allows the medical image processing device 100 to suitably draw contour lines in a predetermined cross-section of the volume data obtained by the CT scanner 200.

[0105] Furthermore, the contour line OL may pass through three or more nodes, including the first and second points specified via UI120, to form a closed region containing tissue within the subject. This allows the medical image processing device 100 to generate a closed region surrounding the outer periphery of the tissue while suppressing the influence of high-luminance and low-luminance points in the vicinity of the tissue.

[0106] Furthermore, the generated contour line OL may pass through a second point (e.g., an intermediate node IN11) located between the first point and the third point (e.g., the endpoint node EN). The processing unit 160 may move the second point on the target image. Based on the pixel value of the second point after the move, the processing unit 160 may generate a partial contour line OLP11 (an example of a first partial contour line), which is part of the contour line OL connecting the first point and the second point, and a partial contour line OLP12 (an example of a second partial contour line), which is another part of the contour line OL connecting the second point and the third point. The processing unit 160 may modify the contour line OL based on the partial contour line OLP11 and the partial contour line OLP12.

[0107] As a result, if the position of the nodes through which the generated contour line OL passes is changed, the medical image processing device 100 can follow this change and regenerate the contour line OL passing through each node. Therefore, the medical image processing device 100 can flexibly change the shape of the generated contour line OL to reflect, for example, the user's intention.

[0108] Furthermore, the processing unit 160 may add a fourth point (e.g., an intermediate node IN) to the generated contour line OL. Based on the pixel value of the fourth point, the processing unit 160 may generate a partial contour line OLP1 (an example of a third partial contour line) which is part of the contour line OL connecting the first point and the fourth point, and a partial contour line OLP2 (an example of a fourth partial contour line) which is another part of the contour line OL connecting the fourth point and the second point (e.g., an endpoint node EN). The processing unit 160 may modify the contour line OL based on the partial contour lines OLP1 and OLP2.

[0109] This allows the medical image processing device 100 to add nodes through which the generated contour line OL passes after the contour line OL has been generated, and to regenerate the contour line OL passing through each node as a result of this addition. Therefore, the medical image processing device 100 can flexibly change the shape of the generated contour line OL to reflect, for example, the user's intention.

[0110] Furthermore, the processing unit 160 may display candidate contour lines (OLCs) on the display 130 based on candidate positions of the second point (e.g., candidate position ENC) before determining the position of the second point (e.g., endpoint node EN, intermediate node IN) on the target image. In this case, the processing unit 160 may, for example, when setting the endpoint node EN by dragging via the UI 120, use the position being traversed by the dragging operation as the candidate position ENC of the endpoint node, and calculate a cost function assuming the candidate position ENC of the endpoint node is the endpoint node EN, thereby generating candidate contour lines (OLCs) based on the candidate position ENC of the endpoint node. Therefore, until the candidate position ENC of the endpoint node is determined, the candidate position ENC of the endpoint node moves, and the calculation result of the cost function changes with this movement, causing the candidate contour lines to undergo OLC deformation. The processing unit 160 then sequentially displays candidate contour lines (OLCs) that change in accordance with the movement of the candidate position ENC of the endpoint node. Therefore, the medical image processing device 100 can determine the position of the endpoint node EN by checking candidate OLCs of the changing contour lines, while searching for a suitable position for the endpoint node EN.

[0111] Furthermore, the processing unit 160 may generate contour lines OL based on at least one of the window width and window level. This allows the medical image processing device 100 to generate contour lines OL while taking into account the window information that the user wishes to observe.

[0112] Furthermore, the second point (for example, the endpoint node EN, the intermediate node IN) can be a point between pixels in the target image. The pixel value of the second point can be the pixel value obtained by interpolating the pixels surrounding the second point on the target image. This allows the medical image processing device 100 to use interpolated values ​​instead of just pixel values, so that even when the target image is enlarged, the pixel value between two pixels does not change abruptly, but changes smoothly using the interpolated values. By using such interpolated values ​​as the pixel values ​​of endpoint nodes and intermediate nodes, the medical image processing device 100 can generate a contour line OL with a different path than the contour line OL based on either of the pixel values ​​of two adjacent pixels flanking the endpoint node or intermediate node. [Industrial applicability]

[0113] This disclosure is useful for medical image processing apparatus, medical image processing method, and medical image processing program, etc., which can reduce fluctuations in pixel values ​​at the location through which the contour line passes and generate contour lines that reflect the user's intent. [Explanation of Symbols]

[0114] 100 Medical Image Processing Equipment 110 ports 120 User Interface (UI) 130 displays 140 processors 150 memory 160 Processing Unit 161 Area Processing Unit 162 Image generation unit 163 Contour Line Processing Unit 165 Display Control Unit 200 CT equipment

Claims

1. A medical image processing device, Equipped with a processing unit, The aforementioned processing unit, We obtain volume data of the subjects, Based on the volume data, a first image is generated that represents at least a portion of the tissue within the subject. For each pixel on a candidate path connecting a first point and a second point on the contour of the tissue in the first image, a local cost is calculated, the local cost including a first cost based on at least one of the Laplacian and gradient of the first image at the pixel, and a second cost based on the difference between the pixel value of the pixel and the pixel value of the second point. Based on the sum of the local costs calculated for each pixel on the candidate path, a contour line connecting the first point and the second point is generated. The first image and the outline are displayed on the display unit. The second point mentioned above is a point specified by the user via the control unit. Medical image processing equipment.

2. The first point is the starting point of the contour line, The second point is the endpoint of the contour line. The medical image processing apparatus according to claim 1.

3. The processing unit, via the operation unit, specifies the first point and the second point. The medical image processing apparatus according to claim 1 or 2.

4. The volume data mentioned above is data obtained by imaging the subject using a CT scanner. The medical image processing apparatus according to claim 1 or 2.

5. The first image is a cross-sectional image of a predetermined cross-section of the subject in the volume data. The medical image processing apparatus according to claim 1 or 2.

6. The contour line passes through three or more points, including the first point and the second point, as specified via the operating unit, and forms a closed region including the tissue within the subject. The medical image processing apparatus according to claim 3.

7. The generated contour line passes through the second point located between the first point and the third point, The aforementioned processing unit, Move the second point on the first image, Based on the pixel value of the second point after movement, a first partial contour line is generated, which is part of the contour line connecting the first point and the second point, and a second partial contour line is generated, which is another part of the contour line connecting the second point and the third point. The contour line is modified based on the first partial contour line and the second partial contour line. The medical image processing apparatus according to claim 1 or 2.

8. The aforementioned processing unit, A fourth point is added to the generated contour line, Based on the pixel value of the fourth point, a third partial contour line is generated, which is part of the contour line connecting the first point and the fourth point, and a fourth partial contour line is generated, which is another part of the contour line connecting the fourth point and the second point. Based on the third partial contour line and the fourth partial contour line, the contour line is modified. The medical image processing apparatus according to claim 1 or 2.

9. Before determining the position of the second point on the first image, the processing unit causes the display unit to display candidate contour lines based on the candidate positions of the second point. The medical image processing apparatus according to claim 1 or 2.

10. The processing unit generates the contour line based on at least one of the window width and the window level. The medical image processing apparatus according to claim 1 or 2.

11. The second point is a point between pixels in the first image, The pixel value of the second point is the pixel value obtained by interpolating the pixels around the second point on the first image. The medical image processing apparatus according to claim 1 or 2.

12. Steps to obtain volume data of the subject, The steps include generating a first image representing a portion of the tissue within the subject based on the volume data, A step of generating a contour line connecting the first point and the second point, based on the sum of the local costs calculated for each pixel on a candidate path connecting a first point and a second point on the contour of the tissue on the first image, wherein the local cost includes a first cost based on at least one of the Laplacian and gradient of the first image at the pixel, and a second cost based on the difference between the pixel value of the pixel and the pixel value of the second point, and a contour line connecting the first point and the second point is generated based on the sum of the local costs calculated for each pixel on the candidate path. The steps include displaying the first image and the outline on the display unit, It has, The second point mentioned above is a point specified by the user via the control unit. Medical image processing methods.

13. A medical image processing program for causing a computer to execute the medical image processing method described in claim 12.

Images