JP2006000126A - Image processing method, apparatus and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To enhance the speed of processing while retaining the quality of a projected image at a level of meeting practical use without any necessity of specific hardware in the image processing producing the projected image from a three-dimensional medical image such as MIP (Maximum Intensity Projection) processing. <P>SOLUTION: In a case of the MIP processing, when a plurality of search points are set on a visual line passing through projected pixels constituting the projected image, visual points and the three dimensional medical image, whether or not the maximum pixel value in the processed search points on the visual line is updated is evaluated in a search point of a processing object and, when it is updated, a step width between the search point and a next search point is made smaller than a step width when it is not updated. This method, in a figure showing an example of a pixel value change along the visual line, allows a detailed search in a part of a continuous line which has a high possibility of updating the maximum pixel value (retention of the image quality level) and allows a rough search in a part of a dashed line which has a low possibility of the updating (speed enhancement of the processing). <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to medical image processing, and more particularly to processing for generating a projection image from a three-dimensional medical image.

  Conventionally, in the medical field, observation and diagnosis of a projection image obtained by projecting a three-dimensional medical image obtained by a CT apparatus, an MRI apparatus, an ultrasonic diagnostic apparatus (echo) or the like onto a desired projection surface are performed. As a method for obtaining such a projection image, a plurality of search points are set by setting a plurality of search points along a line of sight passing through an arbitrary viewpoint and each projection pixel on the projection plane in a three-dimensional medical image. Image processing for obtaining the pixel value of the projection pixel for each line of sight based on the pixel value of the point is performed. As such image processing, for example, MIP (Maximum Intensity Projection) processing for extracting and projecting the maximum value of the pixel value of the search point for each line of sight or MinIP (Minimum) for extracting and projecting the minimum value. Intensity Projection (minimum value projection) processing is known.

  However, in such image processing, it is necessary to perform size comparison or product-sum operation based on the pixel values for all pixels (search points) along each line of sight, and the amount of data to be processed is three-dimensional. In a medical image, the processing load is considerably high.

Thus, various techniques for speeding up such image processing have been proposed. For example, in MIP processing and MinIP processing, a plurality of CPUs are provided in a computer for calculating the maximum or minimum pixel value for each line of sight, and parallel processing is performed, thereby reducing the time required for the calculation. And a method of searching for a maximum value or a minimum value by thinning out pixels in a pixel row along the line of sight at a certain interval (for example, Patent Document 1).
JP 2001-236492 A

  However, the above-described parallel processing method requires specific hardware provided with a plurality of CPUs, so that a general-purpose computer cannot be used easily, leading to high costs. Further, in the method of thinning out pixels to be searched at a constant interval described in Patent Document 1, the processing speed increases as the number of thinned pixels increases, but the number of thinned pixels (or search points) is increased. There is a pixel that has true maximum and minimum values in it, and there is a high possibility that the true maximum and minimum values will not be searched, and this will cause deterioration of image quality to the extent that it cannot be put to practical use. was there.

  The present invention has been made in view of such circumstances, and a plurality of projection pixels on a projection plane onto which a three-dimensional medical image is projected are connected to an arbitrary viewpoint in the three-dimensional medical image. By sequentially setting a plurality of search points along each line of sight, and determining the pixel value of each projection pixel for each line of sight based on each pixel value of the set plurality of search points In image processing for generating a projection image from an image, a method, an apparatus, and a program that do not require specific hardware and that can speed up the processing while maintaining the image quality of the projection image at a level that can be practically used. Is intended to provide.

  In the image processing method according to the present invention, each of a plurality of lines of sight connecting a plurality of projection pixels on a projection plane on which the three-dimensional medical image is projected and an arbitrary viewpoint in a three-dimensional medical image representing a subject. A plurality of search points are set in order, and the pixel value of each projection pixel is determined for each line of sight based on each pixel value of the set plurality of search points. In the image processing method for generating a projection image composed of: a search point that evaluates the magnitude of the contribution to the determination of the pixel value of the projection pixel for each search point; The next search point is set such that the step width between is smaller than the step width between the search point evaluated to have a smaller contribution and the next search point.

  The image processing apparatus according to the present invention implements this method. That is, in the three-dimensional medical image representing the subject, a plurality of searches along each of a plurality of lines of sight connecting each of a plurality of projection pixels on the projection surface on which the three-dimensional medical image is projected and an arbitrary viewpoint. A search point setting means for setting points in order, and a pixel value of each projection pixel is determined for each line of sight based on each pixel value of a plurality of set search points, and each projection pixel is determined. An image processing apparatus including a projection image generation unit that generates a projection image is further provided with an evaluation unit that evaluates the magnitude of contribution to the determination of the pixel value of the projection pixel for each search point, and the search point setting unit includes: The step width between the search point evaluated to have a larger contribution and the next search point is smaller than the step width between the search point evaluated to have a smaller contribution and the next search point. The feature is that the next search point is set To.

  Furthermore, an image processing program according to the present invention is for causing a computer to execute the above method. That is, in the three-dimensional medical image representing the subject, a plurality of searches along each of a plurality of lines of sight connecting each of a plurality of projection pixels on the projection surface on which the three-dimensional medical image is projected and an arbitrary viewpoint. Projection image composed of projection pixels for which the pixel values are determined by sequentially setting points, determining the pixel value of each projection pixel for each line of sight based on each pixel value of the set plurality of search points In the program for causing the computer to execute the image processing for generating the search point, the computer evaluates the magnitude of the contribution to the determination of the pixel value of the projection pixel for each search point, and the search point evaluated to have a larger contribution and the next The next search point is set such that the step width between the search points is smaller than the step width between the search point evaluated to have a smaller contribution and the next search point. And

  Next, details of the image processing method and apparatus and program according to the present invention will be described.

  The “three-dimensional medical image representing a subject” is a medical image representing a subject, and is defined by giving a pixel value to each of three-dimensionally arranged pixels (voxels). It is an image by. As a specific example, an image obtained by stacking a plurality of two-dimensional slice images obtained by imaging using a CT apparatus or an MRI apparatus in the depth (depth) direction to form a three-dimensional image can be considered.

  A “line of sight” connects a projection pixel on the projection plane and an arbitrary viewpoint. When a three-dimensional medical image is projected in parallel on the projection plane, the viewpoint is located at infinity and the lines of sight are parallel to each other. On the other hand, when perspectively projecting a three-dimensional medical image on a projection surface, the viewpoint is located at a finite distance, and each line of sight is radial. The positional relationship among the “viewpoint”, “three-dimensional medical image”, and “projection plane” does not matter. For example, it may be a positional relationship in which the line of sight passes in the order of “viewpoint”, “3D medical image”, and “projection plane”, or the line of sight of “viewpoint”, “projection plane”, and “3D medical image”. The positional relationship may pass in order.

  The “plurality of search points” along the line of sight may be on the line of sight, or pixels near the line of sight may be substituted. Further, the “pixel value” of the “search point” may be the pixel value of the pixel of the three-dimensional medical image existing at the position of the search point, or based on the pixel value of the pixel of the three-dimensional medical image near the search point. Or a value obtained by interpolation calculation. Furthermore, the pixel values of the pixels of the three-dimensional medical image near the search point may be used as they are.

  When “evaluating the magnitude of the contribution to the determination of the pixel value of the projection pixel for each search point”, the pixel value of the search point may be directly evaluated or based on the pixel value of the search point The evaluation value may be acquired by calculation or table reference. Further, the evaluation may be performed not only based on the pixel value of the search point but also based on the pixel value of a pixel near the search point. As a specific example, the gradient of the pixel value of the search point is calculated based on the pixel value of the pixel in the vicinity of the search point, and the evaluation is performed based on the calculated gradient of the pixel value and the pixel value of the search point. It is possible to do it.

  “Step width” refers to an interval between a search point to be processed along a line of sight and a search point to be processed next along the line of sight. In the present invention, the search point to be processed next is set while changing the step width in accordance with the magnitude of the contribution to the determination of the pixel value of the projection pixel at the search point processed most recently. Therefore, the intervals between a plurality of search points along a certain line of sight are not always constant. Moreover, this invention is not limited to what changes this step width to two steps according to the magnitude | size of this contribution, You may change to three or more steps.

  Next, specific examples of image processing according to the present invention will be given.

  As a specific example of image processing according to the present invention, MIP processing or MinIP processing can be considered. That is, the “pixel value of each projection pixel” is the maximum / minimum value of each pixel value of a plurality of search points for each line of sight (the left side of “/” indicates the case of MIP processing, and the right side indicates the case of MinIP processing. The same applies hereinafter). In this case, for each line of sight, the maximum value (hereinafter referred to as the provisional maximum pixel value) / minimum value (hereinafter referred to as the provisional minimum pixel value) among the pixel values of the already processed search points is updated, and along the line of sight. Then, the next search point is processed to obtain the final maximum pixel value / minimum pixel value for each line of sight.

  In the case of MIP / MinIP processing, “evaluate the magnitude of the contribution to the determination of the pixel value of the projection pixel for each search point” based on whether or not the provisional maximum / minimum pixel value is updated at the search point to be processed, It is conceivable that “step width” is determined based on the evaluation result, and “next search point is set” based on the determined “step width”. That is, when the provisional maximum / minimum pixel value is updated at the search point to be processed, it is evaluated that the contribution to the determination of the pixel value of the projection pixel is large and the provisional maximum / minimum pixel value is not updated. Also, the next search point is set by reducing the step width to the next search point. This is because, in a medical image, the correlation between pixel values between peripheral pixels is high, so that the provisional maximum / minimum pixel value is updated at a certain search point is close to the pixel value at that search point even in the vicinity. This means that a search is made in more detail because there is a high possibility of having a value and there is a high possibility that the provisional maximum / minimum pixel value is further updated in the vicinity thereof. The evaluation of the magnitude of the contribution is not limited to the case where the provisional maximum / minimum pixel value is updated at the search point. For example, the pixel value at the search point is 80% of the provisional maximum pixel value. The relationship between the value obtained by multiplying the provisional maximum / minimum pixel value by a certain ratio and the pixel value at the search point may be used as a reference, such that the larger the value is, the higher the contribution is evaluated. Further, the next search point may be set by changing the step width by classifying into finer stages according to the ratio to the provisional maximum / minimum pixel value or the like.

  Further, in the case of MIP / MinIP processing, when determining the pixel value of the projection pixel for the line of sight near the line of sight for which the pixel value of the projection pixel has already been determined, The first search point may be set at a position close to the position of the search point having the maximum value / minimum value, and the search point may be set in order from the first search point. As a specific example, each line of sight is divided into a plurality of sections, and the maximum value / minimum value of the pixel value is obtained in the line of sight to be processed, and the section to which the search point having the maximum value / minimum value belongs is obtained. When the pixel value of the projection pixel is determined for the line of sight in the vicinity, the first search point may be set in a section corresponding to the obtained section.

  In addition, it is conceivable to apply the image processing of the present invention to volume rendering processing by the ray casting method. This is based on the opacity (Opacity) and luminance value set for each pixel (voxel) constituting the 3D medical image, and sampling and adding these values at each search point along the line of sight In this process, the pixel value of the projection pixel is obtained to generate a translucent projection image.

  In this case, as a specific example of “the magnitude of the contribution to the determination of the pixel value of the projection pixel”, it can be considered that the opacity level at each search point is used as a reference. That is, in this volume rendering process, when the opacity value added for a given line of sight reaches a predetermined value, the process for that line of sight is terminated. Therefore, the search point having a larger opacity value is evaluated as having a larger contribution to the determination of the pixel value of the projection pixel. In addition, since the opacity is obtained based on the signal value (CT value or the like) at each pixel (voxel), it is considered that the correlation between the pixel values (signal values) between neighboring pixels is high in the medical image. And the correlation of the opacity value between the surrounding pixels is also high. Therefore, a large opacity value at a certain search point is highly likely to have a value close to the opacity at the search point in the vicinity, and greatly contributes to the determination of the pixel value of the projection pixel. Since the possibility is high, the search is performed in more detail by reducing the step width. As a specific example of the evaluation of the opacity value, it can be considered that if the opacity at the search point is larger than a predetermined threshold value, the contribution to the determination of the pixel value of the projection pixel is great.

  In the present invention, each projection pixel is set along each of a plurality of lines of sight connecting each projection pixel and an arbitrary viewpoint, which is a basis for determining a pixel value of each projection pixel constituting the projection image based on the three-dimensional medical image. When a plurality of search points are sequentially set, the magnitude of the contribution to the determination of the pixel value of the projection pixel is evaluated for each search point, and the search point evaluated to have a larger contribution and the next search point The next search point is set so that the step width between the search points is smaller than the step width between the search point evaluated to have a smaller contribution and the next search point. Here, since the medical image has a characteristic that the correlation between the pixel values of the surrounding pixels is high, the projection pixel is similarly used in the vicinity of the search point evaluated as having a large contribution to the determination of the pixel value of the projection pixel. There is a high possibility that the contribution to the determination of the pixel value will be large. Therefore, by changing the size of the step width in accordance with the magnitude of the contribution to the determination of the pixel value of the projected pixel at the search point processed immediately as described above, in a portion where this contribution is highly likely to be large , It becomes possible to perform a more detailed search than the part where the possibility that this contribution is small is high, while maintaining the accuracy of the determination of the pixel value of the projection pixel in the part where the contribution to the determination of the pixel value of the projection pixel is large, In a portion where this contribution is small, high-speed processing is prioritized by performing a rough search with a reduced number of search points. As a result, the overall processing speed can be increased while maintaining the image quality of the projected image at a level that can be practically used. In addition, since specific hardware such as parallel processing is not necessary for realizing such image processing, it is also effective in terms of processing cost.

  Further, in the case of MIP / MinIP processing, since the correlation between pixel values between peripheral pixels is high in the medical image as described above, in the processing for the line of sight near the already processed line of sight, the already processed line of sight is used. There is a high possibility that the pixel value becomes maximum / minimum in the vicinity of the position of the search point having the maximum / minimum pixel value. Further, the search point with the maximum / minimum pixel value in the line of sight to be processed has the largest contribution to the determination of the projection pixel. Therefore, when determining the pixel value of the projection pixel for the line of sight near the line of sight for which the pixel value of the projection pixel has already been determined, the search point having the maximum value / minimum value in the line of sight for which the pixel value of the projection pixel has already been determined When the first search point is set at a position close to the position and the search points are set in order from the first search point, the search point having the maximum / minimum pixel value in the line of sight is determined. Set a subsequent search point with a larger step width because it is more likely that the process will be performed early and the contribution to the determination of the projected pixel at the search point processed afterwards is less likely to exceed it. Is possible. As a result, the number of search points to be processed is reduced, so that the processing efficiency is further improved.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a hardware configuration diagram showing an overview of a three-dimensional medical image processing system. As shown in the figure, in this system, the modality 1, the image storage server 2, and the image processing workstation 3 can communicate with each other via a network 9 based on a protocol such as DICOM (Digital Imaging and Communication in Medicine). Connected in a normal state.

  The modality 1 acquires a three-dimensional medical image V representing the subject. Specifically, a CT apparatus, an MRI apparatus, an ultrasonic diagnostic apparatus, or the like.

  The image storage server 2 is a computer that stores and manages the three-dimensional medical image V acquired and generated by the modality 1 and the image processing workstation 3 in an image database, and is a large-capacity external storage device or database management software (for example, ORDB (Object Relational DataBase) management software).

  The image processing workstation 3 is a computer that performs image processing on the three-dimensional medical image V acquired from the modality 1 or the image storage server 2 in response to a request from the interpreter and displays the generated image. An input device such as a keyboard and a mouse for inputting a request from an image interpreter, a main storage device (hereinafter referred to as an image memory) having a capacity capable of storing the acquired three-dimensional medical image V, and a high-level display for displaying a generated image It is equipped with a fine liquid crystal display.

  FIG. 2 is a block diagram showing functions of the image processing workstation 3. As shown in the figure, the image processing workstation 3 includes an image acquisition unit 31 that acquires a three-dimensional medical image V to be processed from the modality 1 and the image storage server 2, and an image processing desired by a user such as an image interpreter. The image processing setting unit 32 for setting the type and detailed content, and the type and content of the image processing set by the image processing setting unit 32 for the three-dimensional medical image V acquired by the image acquisition unit 31 The image processing unit 33 performs image processing in accordance with the image processing unit 33 and the screen display unit 34 displays the image P processed by the various image processing unit 33 on the screen. The image processing setting unit 32 includes one in which an image interpreter operates interactively by operating an input device such as a keyboard and a mouse, and one that is set in advance in a setting file. The various image processing units 33 include an MPR (Multi-Planar Reconstruction) processing unit 33A, an MIP processing unit 33B, a MinIP processing unit 33C, a volume rendering processing unit 33D, and the like. Here, the MPR (Multi-Planar Reconstruction) process is not only the axial displacement (Axial) of a predetermined area in a three-dimensional medical image, but also coronal and sagittal (Sagittal). ), And a process of generating an arbitrary cross-sectional image by oblique displacement (Oblique). These functions are realized by executing an application program on the image processing workstation 3 with an input / selection operation by an interpreter.

  Next, the operation of the image interpreter and the flow of processing in this medical image processing system, particularly the image processing workstation 3, will be described.

  First, the image interpreter selects a three-dimensional medical image to be processed while viewing the screen shown in FIG. 3 displayed on the high-definition liquid crystal display of the image processing workstation 3. The screen shown in FIG. 3 is generated by the image acquisition unit 31, and the search conditions for the image to be inspected are classified into hierarchized categories. The image acquisition unit 31 sets the category selected by the interpreter by operating the mouse, the keyboard, or the like, in the search request message to the image storage server 2 as the search condition for the image to be processed. For example, as shown in FIG. 3, when the current date and time is 16:45 on January 10, 2004, and the radiographer selects “examination for the last 24 hours” or “CT”, the image acquisition unit 31 A search condition in which the date and time is “January 9, 2004 16:46 to January 10, 2004 16:45” and the modality is “CT” is set in the search request message. When the image interpreter clicks the search execution button by operating the mouse, the image acquisition unit 31 transmits a search request message to the image storage server 2 via the network 9. The image storage server 2 searches the image database based on the received search request message, and uses image processing work as a response message with list information such as patient name, patient ID, examination date, modality, and number of images of the corresponding image data. Transmit to station 3. In the image processing workstation 3, the list information received by the image acquisition unit 31 is displayed as shown in FIG. When the image interpreter selects a line representing the image data of the patient to be interpreted from the information on the listed image data by operating the mouse 34, the keyboard 35, or the like, the image acquisition unit 31 selects the list information of the selected line. The image acquisition request message is set and transmitted to the image storage server 2. The image storage server 2 searches the image database based on the received list information, and transmits the corresponding image data to the image processing workstation 3. In the image processing workstation 3, the image acquisition unit 31 temporarily stores the received image data in the external auxiliary storage device 33 and writes this image data in the image memory. In this image data, the position of each pixel is defined in a three-dimensional coordinate system in which the left-right direction of the subject is the x-axis, the front-rear direction is the y-axis, and the vertical direction is the z-axis. Associated with position coordinates.

  Next, the image acquisition unit 31 reads the setting file, acquires information related to image processing for generating an image to be displayed first, and requests image processing from various image processing units 33 to appropriate ones. A request message is created and transmitted to the image processing unit. Upon receiving the request message, the image processing unit processes the three-dimensional medical image V read into the image memory based on the image processing parameters included in the request message from the image acquisition unit 31, and the image processing unit The generated image is displayed on the high-definition liquid crystal display 38 of the image display unit 34. For example, when the request message is set so that the MPR process is performed on the entire area of the subject and three cross-sectional images are generated by the axial cut, the coronal cut, and the sagittal cut, the MPR process The unit 33A generates three cross-sectional images based on the request message, and the image display unit 34 displays the generated three cross-sectional images on the high-definition liquid crystal display 38 (see FIG. 5).

  Next, the image interpreter confirms the displayed image, selects a menu on the screen, inputs various parameters, and the like by operating the mouse 34, the keyboard 35, and the like, and gives an instruction to perform desired image processing. For example, FIG. 5 shows an example of a screen on which a menu of image processing to be performed next is displayed by right-clicking on a cross-sectional image due to coronal cut out of three cross-sectional images generated by the MPR processing unit 33A. Yes. When the image interpreter selects a desired process from the displayed menu and performs a left click with the mouse 34, the selected image process is performed.

  Here, when the radiogram interpreter selects “MIP Maximum”, the MIP processing unit 33B performs MIP processing for the entire subject, which is one of the image processing embodiments of the present invention. This will be described in detail below with reference to the flowchart of FIG.

  First, as shown in FIG. 7, the size of the projection surface S on which the three-dimensional medical image V to be processed is projected, the number of pixels to be projected (projection pixels), the direction of the normal vector N perpendicular to the projection surface S, The position of the viewpoint Q is determined (# 1). In the example of FIG. 5, the size of the projection surface S and the number of projection pixels are the same size and pixel interval (for example, 512 pixels × 512 pixels) as the slice surface when the three-dimensional medical image V is acquired based on the setting contents of the setting file. The pixel interval is determined to be about 0.7 mm. The direction of the normal vector N is the same as that of the normal vector N of the cross-sectional image by the coronal section because the interpreter right-clicks on the cross-sectional image by the coronal section in the previous operation (the front-rear direction of the subject). Determined). The position of the viewpoint Q is determined at infinity based on the setting contents of the setting file. Accordingly, parallel projection is performed on the projection surface S, and the lines of sight L connecting the viewpoint Q and the projection pixels are parallel to each other.

  Next, the line of sight L to be processed first is determined (# 2). Specifically, the line of sight L passing through the upper left projection pixel of the projection plane S is determined.

  After the line of sight L to be processed is determined, the first search point in the line of sight L is determined (# 3). Specifically, the point closest to the viewpoint Q within the range of the line of sight L and within the three-dimensional medical image V is determined as the first search point.

  Then, the pixel value f at the determined search point is acquired (# 4). Here, since the direction of the normal vector N on the projection plane S, that is, the direction of the line of sight L is the front-rear direction (y-axis direction) of the subject, the pixel values of the pixels constituting the three-dimensional medical image V are used as they are. It is possible to obtain. On the other hand, when the normal vector N of the projection plane S is not in the diagonal direction, that is, parallel to any of the x, y, and z axes, the pixel values in the eight pixels near the determined search point are set. The pixel value of the search point is calculated by the interpolation processing based on it.

  Here, the maximum value (provisional maximum pixel value) fmax of the search points processed so far in the line of sight L is compared with the pixel value f of the search point currently being processed (# Five). However, since the search point to be processed is the first one, the pixel value f of this search point is set to the provisional maximum pixel value fmax (# 6).

  Next, the step width from this search point to the next search point is determined as S1 (# 7). The step width S1 is set to the same minimum value as the substantial minimum step width, that is, the spatial resolution (x, y-axis direction resolution, sampling pitch) of the three-dimensional medical image V.

  Then, a point advanced from the search point along the line of sight L toward the projection plane S by the step width S1 is determined as the next search point (# 9).

  Next, it is checked whether or not the determined search point is within the existence range of the three-dimensional medical image V (# 10), and if it is within this existence range (# 10; N), The pixel value f is acquired (# 4).

  If this pixel value f is larger than the provisional maximum pixel value fmax (# 5; Y), the provisional maximum pixel value fmax is updated to this pixel value f (# 6), and the steps up to the next search point are performed. The width is determined as S1 (# 7). On the other hand, if the pixel value f is less than or equal to the provisional maximum pixel value fmax (# 5; N), the provisional maximum pixel value fmax is not updated, and the step width to the next search point is determined as S2. (# 8). Here, the step width S2 is set to a value larger than the step width S1. Specifically, about 3 times the value of S1 is appropriate.

  Then, the next search point is determined according to the determined step width (# 9). Hereinafter, until the search point is out of the existing range of the three-dimensional medical image V (# 10; Y), Acquisition of pixel value f (# 4), comparison with provisional maximum pixel value fmax (# 5), update of provisional maximum pixel value fmax (# 6; only when f> fmax), step width setting (# 7, # 8) The determination of the next search point (# 9) is repeated.

  When the search point is outside the range in which the three-dimensional medical image V exists (# 10; Y), the processing for the line of sight L is finished, and the provisional maximum pixel value fmax at that time is projected through the line of sight L. This is the pixel value of the pixel. Then, the line of sight L to be processed next is determined (# 11). Specifically, the line of sight L passing through the projection pixel to the right of the projection pixel whose pixel value has been determined becomes the line of sight L to be processed next. When the projection pixel for which the pixel value is determined is the rightmost on the projection plane S, the line of sight L passing through the leftmost projection pixel in the lower row becomes the line of sight L to be processed next.

  Then, the same processing as described above is performed for the determined line of sight L (# 3 to 10), and the final provisional maximum pixel value fmax becomes the pixel value of the projection pixel through which the line of sight L passes. Similarly, the process is repeated for the line of sight L passing through all the projection pixels (# 12; Y), and the final provisional maximum pixel value fmax is obtained to determine the pixel value of the projection pixel. Since the pixel values of all the projected pixels are determined when the unprocessed line of sight L disappears, that is, when the process of the line of sight L passing through the projection pixel at the lower right of the projection surface S is completed, the processing is completed. (# 12; N).

  Through the above MIP processing, a projection image (maximum value projection image) P is generated by extracting the maximum value of the pixel values in the front-rear direction of the subject from the three-dimensional medical image V and projecting it.

  The image display unit 34 displays the maximum value projection image P generated by the MIP processing unit 33B on a high-definition liquid crystal display.

  As described above, in the MIP processing that is an embodiment of the image processing according to the present invention, when the provisional maximum pixel value fmax is updated at the search point, the contribution of the search point to the determination of the pixel value of the projection pixel is large. The step width to the next search point is made smaller than when the provisional maximum pixel value fmax is not updated. FIG. 8 visualizes the effect of this, and shows the change in the pixel value in the search direction of a certain line of sight L (direction from the viewpoint Q toward the projection plane S). The provisional maximum pixel value fmax is updated at the search point corresponding to the solid line portion of the curve in the figure. In this case, it is presumed that the provisional maximum pixel value fmax is sequentially updated at the subsequent search points, and the search points are set with the minimum step width, and the search is performed in more detail. In contrast, the provisional maximum pixel value fmax is not updated at the search point corresponding to the one-dot chain line portion of the curve in the figure. In this case, it is presumed that the provisional maximum pixel value fmax is not updated even in the subsequent search points, the search points are set by setting the step width larger than the minimum step width, and the number of search points is reduced. Search roughly. As a result, the accuracy of searching for the provisional maximum pixel value fmax is maintained in a portion where the provisional maximum pixel value fmax is likely to be updated, and the processing speed is high in a portion where the provisional maximum pixel value fmax is unlikely to be updated. By giving priority to the performance, the processing speed can be increased while maintaining the image quality of the projected image at a level that can be practically used.

  Actually, when the step width S1 is set to the minimum pixel interval of the three-dimensional medical image V and the step width S2 is set to three times S1, the generated projected image is hardly deteriorated, and the search point The number of is about half that when all parts of the line of sight are searched at the minimum pixel interval.

  By the way, in such a search for the maximum pixel value, when the “mountain” portion where the pixel value is maximum has an impulse shape, the provisional maximum pixel value fmax is updated at the search point immediately before that. Otherwise, since the step width is not the smallest, the next search point is determined across this “mountain” portion, and the provisional maximum pixel value fmax may not be updated correctly. However, in a medical image such as CT, although the pixel value (signal value) changes to some extent at an edge portion or the like, the “mountain” portion does not have an impulse shape, and the surrounding pixels The correlation between the pixel values between the pixels is high, and it is unlikely that the provisional maximum pixel value fmax is erroneously updated as described above. Therefore, the provisional maximum pixel value fmax is updated as described above as a reference. It becomes effective to change the step width.

  In the image processing apparatus according to the present invention, the search point setting means is # 3, # 7, # 8, # 9 in FIG. 6, the evaluation means is # 5, and the projection image generation means is # 1, # 2, # 4. , # 5, # 6, # 10, # 11, and # 12, respectively.

  In the above embodiment, the step widths S1 and S2 are determined based only on whether or not the provisional maximum pixel value fmax is updated at the latest search point, but the direction of the line of sight may also be considered. This is effective when the resolution in each of the x-, y-, and z-axis directions of the three-dimensional medical image V is not isotropic. For example, if the resolution in the direction (z-axis direction) perpendicular to the slice plane of the three-dimensional medical image V (CT image) is lower than the resolution in the x-axis and y-axis directions, and the line-of-sight direction is parallel to the z-axis The step widths S1 and S2 are made larger than when the line-of-sight direction is parallel to the x-axis and y-axis. Thereby, an efficient search according to the high resolution of the original image becomes possible.

  In determining the step widths S1 and S2, the resolution of the original three-dimensional medical image V may be taken into consideration. That is, the step widths S1 and S2 are reduced as the resolution of the original image is higher. As a result, an efficient search corresponding to the resolution of the original image can be performed.

  In the above-described embodiment, the step width is set to two stages of S1 and S2, but it can be set to three or more stages. For example, if the pixel value f at the search point is larger than the provisional maximum pixel value fmax, the step width is set to S1, and if it is larger than 70% of the provisional maximum pixel value fmax and less than the provisional maximum pixel value fmax, the step width is set to S2. If the pixel value fmax is 70% or less, the step width is set to S3. Thus, it is conceivable that the step width is set in three steps (S1 <S2 <S3, hereinafter, the magnitude relationship is the same). Thereby, since the step width can be set more flexibly, the maintenance of the processing accuracy and the high-speed processing can be realized more finely.

  In the above embodiment, when the pixel value of the search point is acquired, the pixel values of the pixels constituting the three-dimensional medical image V are acquired as they are or by interpolation processing based on the pixel values of the eight pixels near the search point. However, the pixel value of the pixel having the maximum pixel value among the eight pixels in the vicinity of the search point may be used as the pixel value of the search point. As a result, it is not necessary to perform an interpolation process by a product-sum operation with a high processing load, which contributes to speeding up of the process.

  Further, it is conceivable to combine the stepwise setting of the step width and the acquisition of the pixel value of the search point by substituting the pixel values of the neighboring eight pixels. That is, first, the pixel value of the pixel having the maximum pixel value among the eight pixels in the vicinity of the search point is substituted as the pixel value of the search point (hereinafter, the substituted value is referred to as a substitute value). Comparison with the provisional maximum pixel value fmax is performed. If this substitute value is less than or equal to the provisional maximum pixel value fmax, the step width is set to S3. If this substitute value is larger than the provisional maximum pixel value fmax, it is determined that there is a possibility that the pixel value of the search point itself is larger than the provisional maximum pixel value fmax, and the pixel value of eight pixels in the vicinity of the search point is set. Based on this, an interpolation process is performed, a pixel value of the search point itself is calculated (hereinafter, the calculated value is referred to as an interpolation value), and the interpolation value is compared with the provisional maximum pixel value fmax. If this interpolation value is less than or equal to the provisional maximum pixel value fmax, the provisional maximum pixel value fmax is not updated, the step width is set to S2, and if this interpolation value is greater than the provisional maximum pixel value fmax, the provisional maximum pixel value fmax is not updated. The pixel value fmax is updated to this interpolation value, and the step width is set to S1. With the above processing, it becomes possible to achieve both the improvement of processing efficiency and speedup and the maintenance of processing accuracy more finely.

  Further, each line of sight L is divided into a plurality of sections, the maximum value of the pixel value is obtained in the line of sight L to be processed, the section to which the search point having the maximum value belongs is stored in the memory as position information, and the line of sight of the line of sight When the pixel value of the projection pixel is determined for the nearby line of sight L, the first search point may be set in a section corresponding to the calculated section. For example, when each line of sight L is divided into three sections k1, k2, and k3, and a search point having the maximum pixel value in a certain line of sight L belongs to the section k2, in the process for the line of sight near the section k2, k3 , K1 are set in the order of k1. Here, in the three-dimensional medical image V, since the correlation between the pixel values between the surrounding pixels is high, if the section to which the search point having the maximum pixel value belongs in the already processed line of sight L is the section k2, this There is a high possibility that the pixel value is maximized in the section k2 even in the processing for the nearby line of sight. In addition, in the process for the line of sight in the vicinity, if the pixel value becomes the maximum in the section k2, the provisional maximum pixel value fmax is not updated thereafter, so the search point is set with a larger step width S2. Done. Therefore, in this case, since the number of search points to be processed is reduced, the processing efficiency is further improved.

  As described above, the case where the image processing of the present invention is applied to the MIP processing has been described. However, the image processing can also be applied to the MinIP processing for the entire subject. In the three-dimensional medical image processing system according to the embodiment of the present invention, “MinIP Maximum” is selected in FIG. 5, and processing is performed by the MinIP processing unit 33C. The details of the specific process may be read by reversing the magnitude relationship in the description of the MIP process. That is, in the process of each line of sight L, the pixel value f of the search point is compared with the provisional minimum pixel value fmin. If f <fmin, the provisional minimum pixel value fmin is updated, and the process until the next search point is performed. When the step width is set to S1 and f ≧ fmin, the temporary minimum pixel value fmin is not updated, and the step width to the next search point is set to S2 (S1 <S2). Thereby, it is possible to obtain the same effect as the MIP process.

  Further, the same effect can be obtained when the image processing of the present invention is applied to so-called slab MIP / MinIP processing that targets only a part of the subject (region of interest).

  The image processing of the present invention can also be applied to volume rendering processing. In the three-dimensional medical image processing system according to the embodiment of the present invention, “VR Default” or “VR...” Is selected in FIG. 5, and processing is performed by the volume rendering processing unit 33D. First, an overview of volume rendering processing will be described.

  The processing performed by the volume rendering processing unit 33D includes preprocessing, luminance value setting, opacity setting, and ray casting processing.

  First, in the preprocessing, noise removal, image enhancement, and the like are performed on the three-dimensional medical image V by filtering calculation or the like.

  Next, a luminance value is set for each pixel (voxel) constituting the three-dimensional medical image V. Here, the gradient value obtained from the pixel values of the pixels in the vicinity of the pixel to be set is used as a normal vector, and a von shading model or the like is applied in consideration of the light source to obtain the luminance value of each pixel.

  Further, an opacity is set for each pixel constituting the three-dimensional medical image V. Here, the opacity is obtained from the pixel value and the gradient value for each pixel based on a two-dimensional mapping table or a mathematical expression defined by the radiogram interpreter. Note that the opacity may be obtained based only on the pixel value of each pixel.

  Then, ray casting processing is performed based on the set luminance value and opacity, and the pixel values of the projection pixels constituting the projection image are obtained. First, similarly to the MIP processing described above, the size of the projection surface S on which the three-dimensional medical image V to be processed is projected, the number of pixels to be projected (projection pixels), and the direction of the normal vector N perpendicular to the projection surface S By determining the position of the viewpoint Q, the line of sight L connecting the viewpoint Q and each of the projection pixels is defined. Next, a plurality of search points are sequentially set along the line of sight L, and the luminance value and opacity at the search point are obtained by interpolation processing based on the luminance values and opacity of neighboring pixels. Next, based on the amount of light incident on the search point (incident luminance value) and the luminance value and opacity at the search point, the amount of light incident on the search point that passes through the search point (luminance) Value) and the amount of light (brightness value) reflected at the search point among the light from the light source, and the obtained amount of light (brightness value) is set as the incident luminance value for the next search point. Further, when the incident luminance value is obtained in this way, the opacity at the search point is sequentially added. When the line of sight L exits the 3D medical image V, that is, when the search point is outside the range of the 3D medical image V, or when the added opacity becomes 1, the line of sight The process for L is terminated, and the incident luminance value at that time is set as the pixel value of the projection pixel on the line of sight L. The above processing is performed for all lines of sight, and pixel values of all projection pixels are obtained.

  Next, application of image processing according to the present invention to volume rendering processing will be described. In this embodiment, when setting a plurality of search points along the line of sight L, pay attention to the opacity at the set search points, and if the opacity is greater than a predetermined threshold, the pixel value of the projection pixel If the search point is determined to have a large contribution to the determination of S, the step width to the next search point is set to S1, and if the opacity is equal to or less than a predetermined threshold, the search point for the determination of the pixel value of the projection pixel The step width to the next search point is set to S2. Here, the size of the set step width is S1 <S2. By doing so, as in the case of the MIP processing described above, a more detailed search is performed in a portion where the contribution of the search point to the determination of the pixel value of the projection pixel is large, and the pixel information in that portion is the projection pixel. By performing a coarser search in a portion where the contribution of the search point to the determination of the pixel value of the projection pixel is small, while the image quality level of the projection image is maintained and the image value level of the projection image is maintained. The search points to be processed are reduced, and high speed processing is realized.

  In the present embodiment, the image processing workstation 3 performs both image processing and image display. However, an image processing server is separately provided and connected to the network 9, and the image processing is performed on the image processing server. You may make it perform. Thus, distributed processing is achieved. For example, when images are displayed on a plurality of terminals, it is not necessary to install a plurality of high-performance image processing workstations, which contributes to a reduction in the cost of the entire system.

Hardware configuration diagram showing an overview of a 3D medical image processing system Block diagram showing functions of image processing workstation The figure which shows an example of the screen which sets the search condition of the process target image in an image processing workstation The figure which shows an example of the screen where the search result list is displayed with the image processing workstation The figure which shows an example of the screen which selects the image processing which an image interpreter desires in an image processing workstation The flowchart showing the flow of MIP processing to which image processing of the present invention is applied. Schematic diagram showing the relationship between the 3D medical image and the viewpoint, projection plane, and line of sight The figure which visualized the search process for every eyes | visual_axis by the MIP process to which the image processing of this invention was applied

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Modality 2 Image storage server 3 Image processing workstation 9 Network 31 Image acquisition part 32 Image processing setting part 33 Various image processing parts 33A MPR processing part 33B MIP processing part 33C MinIP processing part 33D Volume rendering processing part 34 Image display part

Claims (6)

In the three-dimensional medical image representing the subject, a plurality of search points are provided along each of a plurality of lines of sight connecting each of a plurality of projection pixels on a projection plane on which the three-dimensional medical image is projected and an arbitrary viewpoint. A projection image that is set in order, determines a pixel value of each projection pixel for each line of sight based on each pixel value of a plurality of set search points, and is composed of each projection pixel for which the pixel value is determined In an image processing method for generating
Evaluate the magnitude of the contribution to the determination of the pixel value of the projection pixel for each search point;
The step width between the search point evaluated to be greater than the contribution and the next search point is larger than the step width between the search point evaluated to be less than the contribution and the next search point. An image processing method characterized in that the next search point is set to be small. The pixel value of each projection pixel is determined as the maximum value or the minimum value of each pixel value of a plurality of search points for each line of sight
When setting the search point along the line of sight near the line of sight in which the pixel value of the projection pixel has already been determined, the pixel value of the projection pixel has the maximum value or the minimum value in the line of sight that has already been determined 2. The image processing method according to claim 1, wherein a first search point is set at a position close to the position of the search point, and the search point is set in order from the first search point. In the three-dimensional medical image representing the subject, a plurality of search points are provided along each of a plurality of lines of sight that connect each of a plurality of projection pixels on a projection plane on which the three-dimensional medical image is projected and an arbitrary viewpoint. Search point setting means for setting in order;
Projection image generation for determining a pixel value of each projection pixel for each line of sight based on each pixel value of a plurality of set search points, and generating a projection image composed of each projection pixel for which the pixel value is determined And an image processing apparatus comprising:
An evaluation unit that evaluates the magnitude of the contribution to the determination of the pixel value of the projection pixel for each search point;
The search point setting means includes a step width between the search point evaluated to have a larger contribution and a next search point, and the search point and the next search point evaluated to have a smaller contribution. An image processing apparatus characterized in that the next search point is set so as to be smaller than a step width between. The projection image generating means determines a maximum value or a minimum value of each pixel value of a plurality of search points for each line of sight as a pixel value of each projection pixel;
A position information storage unit that stores the position of the search point having the maximum value or the minimum value in the line of sight in which the pixel value of the projection pixel has already been determined;
When the search point setting means sets the search point along the line of sight near the line of sight where the pixel value of the projection pixel has already been determined, the position close to the position stored in the position information storage unit 4. The image processing apparatus according to claim 3, wherein a first search point is set to the first search point, and the search point is set in order from the first search point. In the three-dimensional medical image representing the subject, a plurality of search points are provided along each of a plurality of lines of sight that connect each of a plurality of projection pixels on a projection plane on which the three-dimensional medical image is projected and an arbitrary viewpoint. A projection image that is set in order, determines a pixel value of each projection pixel for each line of sight based on each pixel value of a plurality of set search points, and is composed of each projection pixel for which the pixel value is determined In a program for causing a computer to execute image processing for generating
In the computer,
For each search point to evaluate the magnitude of the contribution to the determination of the pixel value of the projection pixel;
The step width between the search point evaluated to be greater than the contribution and the next search point is larger than the step width between the search point evaluated to be less than the contribution and the next search point. A program characterized in that the next search point is set to be smaller. The pixel value of each projection pixel is determined as the maximum value or the minimum value of each pixel value of a plurality of search points for each line of sight,
When setting the search point along the line of sight in the vicinity of the line of sight for which the pixel value of the projection pixel has already been determined, the pixel value of the projection pixel has the maximum value or the minimum value in the line of sight for which the pixel value has already been determined. 6. The program according to claim 5, wherein a first search point is set at a position close to the position of the search point, and the computer is caused to perform processing so as to set the search point in order from the first search point. .

Images