CN114693907A - Region detection method, device, equipment and storage medium - Google Patents
Region detection method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN114693907A CN114693907A CN202011630482.0A CN202011630482A CN114693907A CN 114693907 A CN114693907 A CN 114693907A CN 202011630482 A CN202011630482 A CN 202011630482A CN 114693907 A CN114693907 A CN 114693907A
- Authority
- CN
- China
- Prior art keywords
- region
- gray
- gradient
- pixel
- pixel point
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
Abstract
The embodiment of the invention discloses a region detection method, a region detection device, region detection equipment and a storage medium. The method comprises the following steps: clustering each gray pixel point according to the pixel gray of each gray pixel point in the medical image, and taking the area where each gray pixel point belonging to the same category is as a candidate area; generating a gradient image corresponding to the medical image according to the gray level of each pixel, and determining a gradient edge in the gradient image according to the pixel gradient of each gradient pixel point in the gradient image; and detecting a target area from each candidate area according to the coupling degree between the area boundary and the gradient edge of each candidate area. The technical scheme of the embodiment of the invention solves the problem of low detection precision of the target region caused by various reasons based on clustering, and realizes the effect of accurate detection of the target region by comparing the complete region boundary of the obtained candidate region with the gradient edge of the region boundary with higher probability as the target region.
Description
Technical Field
The embodiment of the invention relates to the technical field of medical image processing, in particular to a region detection method, a region detection device, region detection equipment and a storage medium.
Background
Low gray scale regions with too low gray scale values and/or high gray scale regions with too high gray scale values often occur in X-ray images; wherein, the low gray scale region may be from an implant (such as an artificial joint, a stent, a pacemaker, a steel plate, a screw, etc.) in the body of the subject, a fixing device (such as an external fixing device) for fixing the subject during the operation, a positioning device (such as a positioning needle, a clamp, etc.) for positioning the lesion, an imaging device (such as a needle holding device in the mammary gland puncture image, etc.) in some medical devices, and the like, and the low gray scale region may also be called as a high attenuation region; the high gray areas may come from areas of the X-ray image that do not pass through the body, which may also be referred to as directly exposed areas.
Normally, the highly attenuating regions and the directly exposed regions in the X-ray image are not the regions that the physician needs to be concerned with, and they may interfere with the imaging of normal human tissue. In particular, problems and phenomena fed back from the clinic indicate that if there are high attenuation regions and/or directly exposed regions in the X-ray image, this will directly affect the display effect of normal human tissue under the default window width level, and therefore it is necessary to detect high attenuation regions and/or directly exposed regions from the X-ray image.
It should be noted that, because the gray scale difference between the high attenuation region and the direct exposure region in the X-ray image and the normal human tissue is not constant, and there is a gray scale change in the high attenuation region and the direct exposure region, and the influence of factors such as image noise is added, it is very difficult to accurately detect the high attenuation region and/or the direct exposure region from the X-ray image.
Disclosure of Invention
The embodiment of the invention provides a region detection method, a region detection device and a storage medium, and aims to achieve the effect of accurately detecting a target region in a medical image.
In a first aspect, an embodiment of the present invention provides a region detection method, which may include:
clustering each gray pixel point according to the pixel gray of each gray pixel point in the medical image, and taking the area where each gray pixel point belonging to the same category is as a candidate area; generating a gradient image corresponding to the medical image according to the gray level of each pixel, and determining a gradient edge in the gradient image according to the pixel gradient of each gradient pixel point in the gradient image; and detecting a target region from each candidate region according to the coupling degree between the region boundary and the gradient edge of each candidate region.
Optionally, clustering the gray pixel points according to the pixel gray levels of the gray pixel points in the medical image may include: the method comprises the steps of sequencing pixel gray levels of gray level pixel points in a medical image, determining gray level category splitting points in a gray level sequencing result based on preset category numbers, and taking the gray level category splitting points as initial clustering centers; and clustering the gray pixel points based on the clustering center and the pixel gray of each gray pixel point.
On this basis, optionally, clustering each gray pixel based on the pixel gray of the clustering center and each gray pixel may include: aiming at each gray pixel point, determining the pixel gray of the gray pixel point and the gray distance between the clustering centers, and clustering the gray pixel points to the category of the clustering center corresponding to the minimum gray distance; determining gray level distortion according to the minimum gray level distance corresponding to each gray level pixel point, and judging whether the gray level distortion meets a preset clustering end condition; if not, determining the clustering center of each category again according to the pixel gray levels of the clustered gray level pixel points belonging to the categories; and repeating the step of determining the pixel gray of the gray pixel points and the gray distance between the clustering centers until the gray distortion meets the clustering ending condition, and ending the clustering.
Optionally, determining a gradient edge in the gradient image according to the pixel gradient of each gradient pixel point in the gradient image includes: screening out a gray class region corresponding to the region attribute of the target region to be detected from each candidate region according to the region gray of each candidate region; and determining a gradient edge in the gradient image according to the pixel gradients of the gradient pixel points in the gradient image, which correspond to the pixel points in each region in the gray class region.
Optionally, determining a gradient edge in the gradient image according to the pixel gradient of each gradient pixel point in the gradient image may include: sorting the gradient pixel points according to the pixel gradients of the gradient pixel points in the gradient image, screening the gradient pixel points according to the sorting positions of the gradient pixel points in the sorting result, and generating a binary image corresponding to the gradient image; and taking a binary edge formed by each binary pixel point in the binary image as a gradient edge of the gradient image.
Optionally, for a current region in each candidate region, detecting a target region from each candidate region according to a coupling degree between a region boundary and a gradient edge of each candidate region, which may include: acquiring first similarity of each current pixel point in the current boundary of the current region and each edge pixel point in the gradient edge at the pixel position, and second similarity of each current pixel point and each target pixel point in the target boundary of the candidate region detected as the target region at the pixel position; and judging whether the current area is the target area or not according to the first similarity or the first similarity and the second similarity so as to realize the detection of the target area.
Optionally, detecting a target region from each candidate region according to the degree of coupling between the region boundary and the gradient edge of each candidate region may include: determining the region detection sequence of each candidate region according to the region attribute of the target region to be detected, and screening out the current region from each candidate region according to the region detection sequence; judging whether the current area is used as a target area according to the coupling degree between the area boundary and the gradient edge of the current area, and realizing the detection of the target area according to the judgment result; and taking the next region positioned in the current region in the region detection sequence as the current region, and repeatedly executing the step of judging whether the current region is taken as the target region according to the coupling degree between the region boundary and the gradient edge of the current region until the detected candidate region and/or the gradient edge meet the preset judgment end condition.
In a second aspect, an embodiment of the present invention further provides an area detection apparatus, which may include:
the candidate region determining module is used for clustering all gray pixel points according to the pixel gray of all the gray pixel points in the medical image and taking the region where all the gray pixel points belonging to the same category are located as a candidate region;
the gradient edge determining module is used for generating a gradient image corresponding to the medical image according to the gray level of each pixel and determining a gradient edge in the gradient image according to the pixel gradient of each gradient pixel point in the gradient image;
and the target area detection module is used for detecting the target area from each candidate area according to the coupling degree between the area boundary and the gradient edge of each candidate area.
In a third aspect, an embodiment of the present invention further provides an area detection device, which may include:
one or more processors;
a memory for storing one or more programs;
when the one or more programs are executed by the one or more processors, the one or more processors implement the region detection method provided by any embodiment of the present invention.
In a fourth aspect, the embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the area detection method provided in any embodiment of the present invention.
According to the technical scheme of the embodiment of the invention, the gray pixel points are clustered according to the pixel gray of the gray pixel points in the medical image, and the region where the gray pixel points belonging to the same category are located is taken as a candidate region; generating a gradient image corresponding to the medical image according to the gray level of each pixel, and determining a gradient edge in the gradient image according to the pixel gradient of each gradient pixel point in the gradient image; and detecting a target region from each candidate region according to the coupling degree between the region boundary and the gradient edge of each candidate region. According to the technical scheme, the problem of low detection precision of the target region caused by uncertain gray level difference between the target region and human tissues, image noise, gray level transition inside the target region and the like is solved aiming at the clustering of the regions, and the effect of accurately detecting the target region is achieved by comparing the complete region boundary of the candidate region obtained by the method with the gradient edge of the region boundary with high probability as the target region.
Drawings
Fig. 1 is a flowchart of a region detection method according to a first embodiment of the present invention;
FIG. 2 is a flowchart of a region detection method according to a second embodiment of the present invention;
fig. 3 is a flowchart of a region detection method according to a third embodiment of the present invention;
fig. 4a is a schematic diagram of an alternative example of a region detection method in the third embodiment of the present invention;
fig. 4b is a schematic diagram of an alternative example of a region detection method in the third embodiment of the present invention;
fig. 5 is a block diagram of a region detecting apparatus according to a fourth embodiment of the present invention;
fig. 6 is a schematic structural diagram of an area detection apparatus in the fifth embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.
In the prior art, the detection scheme of the target region is usually implemented based on algorithms such as gray threshold, region growing and the like, but since the dose setting is likely to be performed according to the posture of the subject during the acquisition of the X-ray image and the characteristics of different high attenuation regions and/or direct exposure regions are unknown, it is difficult to detect the high attenuation regions and/or direct exposure regions by setting an accurate gray threshold in the detection algorithm based on the gray threshold; in addition, due to the influence of image noise and other factors, it is difficult to accurately set the growth criterion in the detection algorithm based on region growth, and at this time, over-segmentation or under-segmentation is easy to occur, and these schemes are difficult to meet the actual requirements in clinical application.
Example one
Fig. 1 is a flowchart of a region detection method according to an embodiment of the present invention. The present embodiment is applicable to a case where a target region is detected from a medical image, and is particularly applicable to a case where a target region is detected from a medical image based on a degree of coupling between a region boundary and a gradient edge. The method may be performed by an area detection apparatus provided in an embodiment of the present invention, where the apparatus may be implemented by software and/or hardware, and the apparatus may be integrated on an area detection device, where the area detection device may be a terminal or a server.
Referring to fig. 1, the method of the embodiment of the present invention specifically includes the following steps:
s110, clustering the gray pixel points according to the pixel gray of each gray pixel point in the medical image, and taking the area where each gray pixel point belonging to the same category is as a candidate area.
The medical image may be an image obtained by acquiring an image of a human tissue based on a medical imaging technology, such as an X-ray image, a Computed Tomography (CT) image, a Digital Radiography (DR) image, a B-mode ultrasound image, and the like.
For an imaging object in a medical imaging process, the imaging object may include a human tissue and other imaging objects except the human tissue, in general, the attenuation degrees of rays involved in medical imaging by the human tissue and the other imaging objects are different, and the higher the attenuation degree is, the lower the pixel gray level in the medical image is; the lower the attenuation degree is, the higher the pixel gray scale in the medical image is, so that the imaging region of the rest of the imaging object in the medical image can be used as the target region to be detected, and the target region can be a low gray scale region (i.e. a high attenuation region) with the pixel gray scale being significantly lower than that of the human tissue, a high gray scale region (i.e. a direct exposure region) with the pixel gray scale being significantly higher than that of the human tissue, or a set of the two, and so on.
The medical image is a gray scale image, the gray scale pixel points are pixel points in the medical image, and the pixel gray scale is a pixel value (i.e., gray scale value) of the gray scale pixel points. The gray pixel points are clustered according to the pixel gray levels of the gray pixel points, that is, the gray pixel points which are relatively similar in pixel gray level are classified into the same category, and the gray pixel points which are relatively different in pixel gray level are classified into different categories. After the gray pixel points are clustered, each gray pixel point can be respectively clustered into a corresponding category, and at this time, the region where each gray pixel point belonging to the same category is located can be used as a candidate region, in other words, each gray pixel point belonging to the same candidate region is a gray pixel point belonging to the same category, and the gray pixel points have strong similarity in the aspect of pixel gray. It should be noted that the number of candidate regions formed by gray-scale pixels belonging to the same category may be one, two, or more, and is not specifically limited herein.
On this basis, optionally, if the gray pixels are directly clustered, it may happen that the gray pixels belonging to the same category cannot form a candidate region because of being scattered, there is a hole in the formed candidate region (that is, most of the gray pixels in a certain region belong to the same category and very few gray pixels do not belong to the category), and so on. On this basis, in order to ensure that the candidate region obtained after clustering is a complete region, the following optional processing modes can be performed: before clustering, performing noise reduction, smoothing and other processing on the medical image to remove noise in the medical image, or performing region hole filling and other processing on the candidate region obtained after clustering, which is not specifically limited herein.
And S120, generating a gradient image corresponding to the medical image according to the gray level of each pixel, and determining a gradient edge in the gradient image according to the pixel gradient of each gradient pixel point in the gradient image.
Wherein, the gradient image is an image corresponding to the medical image generated by performing gradient operation on the pixel gray of each gray pixel point in the medical image, the gradient operation process can be understood as a calculation process of gradient information involved in an edge extraction process of the medical image, such as a calculation process of gradient information in the edge extraction process based on local variance, sobel, prewitt, canny, and the like, exemplarily, taking calculating the local variance in the medical image to generate the gradient image as an example, a preset sliding window with a window size of W × H is obtained, the sliding window is continuously moved, the variance of the pixel gray of all gray pixel points in the sliding window after each movement is sequentially calculated, and the average value of all variances is taken as the pixel gradient (i.e. the pixel value) of a gradient pixel point corresponding to a gray pixel point in the sliding window, the gradient pixel point is a pixel point in the generated gradient image, that is, each gradient pixel point in the gradient image has a unique gray pixel point corresponding to the gradient pixel point at the pixel position, thereby generating the gradient image corresponding to the medical image.
The gradient edge is an edge formed by a plurality of gradient pixel points in the gradient image, the gradient pixel points on the gradient edge can be called edge pixel points, the edge pixel points can be gradient pixel points with larger pixel gradients, the gradient edge is set to have the significance that the gray levels of the gray pixel points on two sides of the region boundary of the target region of the medical image have strong gray level difference, the region boundary is the boundary of the target region, which means that the gradient of the gradient pixel points corresponding to the gray pixel points on the region boundary is larger, namely the gradient edge is likely to be the region boundary of the target region, and then the candidate region where the region boundary similar to the gradient edge is located can be taken as the target region subsequently.
On the basis, various implementation modes are available for determining the gradient edge in the gradient image according to the pixel gradient of each gradient pixel point in the gradient image, for example, the gradient image can be subjected to binarization processing to obtain a binary image, and the edge formed by each binary pixel point with the pixel value of 1 in the binary image is used as the gradient edge; for another example, taking a high attenuation region as an example, edge pixel points on a gradient edge corresponding to the high attenuation region mostly correspond to gray pixel points in a candidate region with smaller regional gray, and the regional gray can show the overall level of the pixel gray of each gray pixel point in the candidate region, so that a gray category region can be screened out from each candidate region according to the regional attribute of a target region to be detected, the regional attribute can show the overall level of the pixel gray of each gray pixel point in the target region, which can reflect whether the target region is a direct exposure region or a high attenuation region, then edge pixel points are screened out from gradient pixel points respectively corresponding to each regional pixel point in the gray category region, and then the edge formed by each edge pixel point is taken as a gradient edge; etc., and are not specifically limited herein.
And S130, detecting a target region from each candidate region according to the coupling degree between the region boundary and the gradient edge of each candidate region.
For each candidate region, the coupling degree may be a similarity between a region boundary and a gradient edge of the candidate region, and specifically, may be a similarity of each boundary pixel point on the region boundary and each edge pixel point on the gradient edge at a pixel position. The coupling degree can be calculated in various ways, such as obtaining a first number of boundary pixels on the boundary of the region and a second number of boundary pixels which can be the same as or similar to a certain edge pixel in pixel position, and determining the coupling degree according to the number ratio between the second number and the first number. Further, the candidate region corresponding to the region boundary with the higher coupling degree may be used as the target region. Of course, in addition to considering the coupling degree between the region boundary and the gradient edge, considering that if a certain coupling degree exists between a certain region boundary and the gradient edge and the target boundary of the candidate region detected as the target region, such a candidate region may also be the target region, so that it is possible to jointly determine whether the candidate region corresponding to the region boundary is the target region according to the coupling degree between the region boundary and the gradient edge and the coupling degree between the region boundary and the target boundary; etc., and are not specifically limited herein.
It should be noted that the reason why the target region is selected according to the gradient edge and the region boundary together, rather than directly according to the gradient edge, is that due to the fact that there is likely to be overlap between the high attenuation imaging object and the human tissue, the high attenuation imaging object has a breakpoint on the gradient edge corresponding to the high attenuation region in the medical image, which means that the detection of the target region cannot be realized by region filling on the gradient edge where the breakpoint exists. Correspondingly, the region detection method can obtain a complete candidate region based on a clustering algorithm, and the region boundary and the gradient edge do not need to be completely overlapped, so that the effect of detecting the complete target region under the condition that a breakpoint exists in the gradient edge is achieved.
According to the technical scheme of the embodiment of the invention, the gray pixel points are clustered according to the pixel gray of the gray pixel points in the medical image, and the region where the gray pixel points belonging to the same category are located is taken as a candidate region; generating a gradient image corresponding to the medical image according to the gray level of each pixel, and determining a gradient edge in the gradient image according to the pixel gradient of each gradient pixel point in the gradient image; and detecting a target region from each candidate region according to the coupling degree between the region boundary and the gradient edge of each candidate region. According to the technical scheme, the problem of low detection precision of the target region caused by uncertain gray level difference between the target region and human tissues, image noise, gray level transition inside the target region and the like is solved aiming at the clustering of the regions, and the effect of accurately detecting the target region is achieved by comparing the complete region boundary of the candidate region obtained by the method with the gradient edge of the region boundary with high probability as the target region.
An optional technical solution is that a gradient edge in a gradient image is determined according to a pixel gradient of each gradient pixel point in the gradient image, and specifically, the determining may include: screening out a gray class region corresponding to the region attribute of the target region to be detected from each candidate region according to the region gray of each candidate region; and determining a gradient edge in the gradient image according to the pixel gradients of the gradient pixel points in the gradient image, which correspond to the pixel points in each region in the gray class region. Considering that the target area can be a high attenuation area or a direct exposure area, the high attenuation area is mostly detected from the candidate areas with smaller area gray scale, and the direct exposure area is mostly detected from the candidate areas with larger area gray scale, so that a gray class area corresponding to the area attribute of the target area to be detected can be screened from each candidate area, namely the gray class area is a candidate area which is similar to the target area on the whole level of the pixel gray scale; furthermore, the gradient edge in the gradient image is determined according to the pixel gradients of the gradient pixel points in the gradient image, which correspond to the pixel points in each region in the gray category region, for example, the pixel gradients of the gradient pixel points corresponding to the gray pixel points on the region boundary of the target region are relatively large, so that the edge formed by the gradient pixel points with relatively large pixel gradients in the gray category region can be used as the gradient edge, and according to the technical scheme, the gradient edge is determined on the pixel gradients corresponding to the gray category regions with similar region attributes to the target region, so that the determination speed and the determination accuracy of the gradient edge are improved; moreover, the determination mode of the gradient edge means that the detection of a direct exposure area can be realized on a high-gray candidate area and the detection of a high attenuation area can be realized on a low-gray candidate area, and since the high-gray candidate area and the low-gray candidate area are self-adaptive clustering results which are not affected by dose, the technical scheme has better adaptability to medical images with different doses and target areas with different types, thereby realizing the effect of accurately detecting the target areas.
An optional technical solution, considering that a pixel gradient of a gradient pixel point corresponding to a gray pixel point on a region boundary of a target region is generally large, determining a gradient edge in a gradient image according to a pixel gradient of each gradient pixel point in the gradient image, may specifically include: sorting gradient pixel points according to the pixel gradients of the gradient pixel points in the gradient image, screening the gradient pixel points according to the sorting positions of the gradient pixel points in the sorting result, generating a binary image corresponding to the gradient image, namely, reserving gradient pixel points with larger pixel gradients in the gradient image, and abandoning gradient pixel points with smaller pixel gradients in the gradient image, wherein the sorting positions can embody the relative sizes of the pixel gradients of certain gradient pixel points in all the gradient pixel points; and taking the binary edge formed by each binary pixel point in the binary image as the gradient edge of the gradient image, wherein the binary pixel point is the pixel point of the binary image, and the pixel value of the binary pixel point is 1 or 0, so that the binary edge formed by each binary pixel point with the pixel value of 1 can be taken as the gradient edge. According to the technical scheme, the gradient pixel points are sorted according to the pixel gradient, the gradient pixel points belonging to the gradient edge can be quickly and accurately screened out from the gradient pixel points, the gradient edge formed based on the gradient pixel points is further obtained, and the effect of accurately determining the gradient edge is achieved.
In order to better understand the specific implementation process of the gradient edge determination, the following description is made in an exemplary manner with reference to specific examples. For example, in a general case, after clustering each gray-scale pixel point, the regional gray scale of each candidate region can be obtained, the regional gray scales of the candidate regions belonging to the same category are similar, and the difference of the regional gray scales of the candidate regions belonging to different categories is large. Assuming that the number of preset classes is N, 1/2 of them is taken as a low-gray class, and 1/2 is taken as a high-gray class, the class in which each candidate region is located can be determined according to the numerical value of the region gray of each candidate region, thereby obtaining a low-gray candidate region belonging to the low-gray class and a high-gray candidate region belonging to the high-gray class. When the target area to be detected is a high attenuation area, the low gray level candidate area is the gray level category area, the gradient pixels are sorted according to the pixel gradients of the gradient pixels respectively corresponding to the area pixels in the gray level category area in the gradient image, a certain proportion of edge pixels are extracted according to the sorting result to generate a binary image corresponding to a gradient subimage, and the gradient subimage is the image corresponding to the gray level category area in the gradient image, so that the target area is detected from the low gray level candidate area subsequently, and the detection speed and the detection efficiency of the target area are improved. Of course, when the target region to be detected includes the direct exposure region, the high gray candidate region may be used as the gray category region to perform the corresponding step, which is similar to the performing process of the high attenuation region and is not described herein again.
Example two
Fig. 2 is a flowchart of a region detection method according to a second embodiment of the present invention. The present embodiment is optimized based on the above technical solutions. In this embodiment, optionally, clustering each gray pixel according to the pixel gray of each gray pixel in the medical image may specifically include: the method comprises the steps of sequencing pixel gray levels of gray level pixel points in a medical image, and determining gray level category splitting points in a gray level sequencing result based on preset category numbers; and taking the gray class splitting point as an initial clustering center, and clustering each gray pixel point based on the clustering center and the pixel gray of each gray pixel point. The explanations of the same or corresponding terms as those in the above embodiments are omitted.
Referring to fig. 2, the method of the present embodiment may specifically include the following steps:
s210, pixel gray levels of all gray level pixel points in the medical image are sequenced, and gray level category splitting points are determined in a gray level sequencing result based on the preset category number.
In the process of clustering each gray pixel point, a gray category split point which can be used as an initial clustering center has certain influence on clustering speed and clustering precision, so that in order to improve the clustering effect, the pixel gray levels of each gray pixel point in the medical image can be sorted first, and then the gray category split point is determined in a gray sorting result based on the preset category number. For example, assuming that 900 gray-scale pixel points are included in a medical image and the number of categories is 90, the gray-scale of the pixels ranked at 10 th, 20 th, … th, 900 th may be used as the gray-scale category splitting point. In practical applications, optionally, since the region boundary obtained by clustering is subsequently defined by the gradient edge to detect the target region, in the case that the number of classes is very small, the region boundary is likely to be larger than the gradient edge, which means that the region boundary cannot be defined based on the gradient edge, that is, the candidate region corresponding to the target region obtained by clustering may be an under-segmentation result of the target region, which facilitates subsequent traversal of the region boundary of each candidate region to gradually approach the gradient edge, and therefore the number of classes may be a larger value.
S220, taking the gray class splitting point as an initial clustering center, clustering the gray pixel points based on the clustering center and the pixel gray of each gray pixel point, and taking the area where each gray pixel point belonging to the same class is as a candidate area.
The gray level category splitting point can be used as an initial clustering center, each gray level pixel point is clustered based on the clustering center and the pixel gray level of each gray level pixel point, for example, for each gray level pixel point, the gray level distance between the gray level pixel point and each clustering center is compared, the category of the clustering center corresponding to the minimum gray level distance is used as the clustering result of the gray level pixel point, that is, the gray level pixel point is classified into the category of the clustering center corresponding to the minimum gray level distance, and therefore the accurate clustering effect of each gray level pixel point is achieved.
On this basis, in order to further improve the clustering accuracy, the following scheme can be specifically adopted for clustering: determining the pixel gray of each gray pixel point and the gray distance between each clustering center, and clustering the gray pixel points to the category of the clustering center corresponding to the minimum gray distance; determining gray level distortion according to the minimum gray level distance corresponding to each gray level pixel point, and judging whether the gray level distortion meets a preset clustering end condition, wherein the clustering end condition can be whether the gray level distortion is smaller than a preset threshold value, whether the absolute value of the difference value between the gray level distortion of the iteration and the gray level distortion of the previous iteration is smaller than a relative error threshold value, and the like; if not, re-determining the clustering center of each category according to the pixel gray of each gray pixel point belonging to each category after clustering, and repeatedly performing the step of determining the pixel gray of each gray pixel point and the gray distance between the clustering centers until the gray distortion meets the clustering ending condition, and ending clustering.
And S230, generating a gradient image corresponding to the medical image according to the gray level of each pixel, and determining a gradient edge in the gradient image according to the pixel gradient of each gradient pixel point in the gradient image.
And S240, detecting a target region from each candidate region according to the coupling degree between the region boundary and the gradient edge of each candidate region.
According to the technical scheme of the embodiment of the invention, the gray class splitting point (namely the initial clustering center) is determined in the pixel gray level sequencing result of each gray pixel point through the preset class number, and each gray pixel point is clustered based on the clustering center and each pixel gray level, so that the effect of self-adaptive gray level clustering on each gray pixel point based on the more accurate initial clustering center is realized.
In order to better understand the specific implementation process of the above clustering, the LBG algorithm is exemplarily described below as an example. Illustratively, the LBG algorithm is a relatively classical algorithm for image compression based on a vector quantization mode, and the optimal solution is sought by using the lloyd iteration, so that the training vector set can be effectively divided. The steps for designing vector quantization are as follows: suppose the training vector set is X ═ X0,x1,...xM-1}，
1) Setting an initial codebook, B0＝{y0 (0),y1 (0),...yN-1 (0)Let the iteration number n equal to 0, average distortion D0→ infinity, and a relative error threshold ε (0 ≦ ε ≦ 1);
2) according to the nearest neighbor condition, the codebook B is codednAs centroids (i.e., clustering)Center), divides the training vector set X into N cells Sn＝{S0 n,S1 n,...SN-1 n},Si nSatisfy the requirement of
3) The average distortion (i.e. the gray level distortion) generated after cell division is calculated, and the average distortion per cell is defined as shown in the following formula, wherein the calculation process of min is the process of cell division, yj nIs the centroid before update of the jth cell, DnIs all xiY of the cell divided by itj nThe average of the gray scale distances between;
4) judging whether the relative error between the average distortion of the current iteration and the average distortion of the previous iteration is smaller than epsilon or not;
5) if so, stopping the whole iterative algorithm, otherwise, adding the numerical values of the corresponding dimensions of all vectors in each cell according to a centroid condition, then dividing the numerical values by the number of all vectors in the cell to serve as the centroid of the cell, updating the code book by using the redetermined centroid (namely code words) of each cell, enabling the iteration number n to be n +1, and then jumping to the step 2).
On the basis, in combination with application scenarios possibly related to the embodiment of the invention, each gray pixel point can be used as a vector to perform training based on an LBG algorithm. Because the LBG algorithm is relatively dependent on the initial codebook, the pixel gray levels of all gray level pixel points can be sequenced, gray level category splitting points are determined in a gray level sequencing result based on the preset category number, and then the gray level category splitting points can be used as the initial codebook to carry out LBG iteration.
EXAMPLE III
Fig. 3 is a flowchart of a region detection method according to a third embodiment of the present invention. The present embodiment is optimized based on the above technical solutions. In this embodiment, optionally, for the current region in each candidate region, detecting a target region from each candidate region according to the degree of coupling between the region boundary and the gradient edge of each candidate region, which may specifically include: acquiring first similarity of each current pixel point in the current boundary of the current region and each edge pixel point in the gradient edge at the pixel position, and second similarity of each current pixel point and each target pixel point in the target boundary of the candidate region detected as the target region at the pixel position; and judging whether the current area is the target area or not according to the first similarity or the first similarity and the second similarity so as to realize the detection of the target area. The same or corresponding terms as those in the above embodiments are not explained in detail herein.
Referring to fig. 3, the method of this embodiment may specifically include the following steps:
s310, clustering the gray pixel points according to the pixel gray of the gray pixel points in the medical image, and taking the region where the gray pixel points belonging to the same category are as a candidate region.
And S320, generating a gradient image corresponding to the medical image according to the gray level of each pixel, and determining a gradient edge in the gradient image according to the pixel gradient of each gradient pixel point in the gradient image.
S330, aiming at the current area in each candidate area, acquiring first similarity of each current pixel point in the current boundary of the current area and each edge pixel point in the gradient edge at the pixel position, and second similarity of each current pixel point and each target pixel point in the target boundary of the candidate area detected as the target area at the pixel position.
The current region is a candidate region for detecting the current region at the current moment, the current boundary is a region boundary of the current region, the current pixel points are pixel points on the current boundary, the first similarity can represent the similarity of each current pixel point and each edge pixel point on the pixel position, and the first similarity can be determined by the first number of each current pixel point and the number ratio of the second number of the current pixel points which can be the same as or similar to the edge pixel points on the pixel position. Accordingly, since it is possible to detect whether the current region is the target region after each time the current region is screened out from the candidate regions, it means that there may be a candidate region detected as the target region after the current region is updated, and the candidate region is the previous current region. The target boundary is a region boundary of a candidate region detected as the target region, the target pixel points are pixel points on the target boundary, and similarly, the second similarity can represent the similarity of each target pixel point and each edge pixel point on the pixel position.
S340, judging whether the current area is the target area or not according to the first similarity or the first similarity and the second similarity so as to realize the detection of the target area.
For example, as shown in fig. 4a, if the first similarity is determined by a ratio of a first number of each current pixel to a second number of current pixels that are the same as or similar to the pixel position of a certain edge pixel, the first similarity may be 50%; of course, if the first similarity is not very high, then the current region cannot be directly considered as the target region, because a part of the current boundary of the current region may coincide with the gradient edge and another part of the current boundary coincides with the target boundary, and the current region at this time may also be used as the target region, for example, as shown in fig. 4b, if the second similarity is represented in the same manner as the first similarity, then the first similarity and the second similarity are both 25%. That is, whether the current region is the target region may be determined according to the first similarity, or the first similarity and the second similarity. It should be noted that each current region can be detected by the above steps, thereby achieving the effect of detecting the target region from each candidate region.
In practical application, optionally, on the basis of any one of the above technical solutions, the detection process of the target region may be to determine a region detection order of each candidate region according to a region attribute of the target region to be detected, and screen out a current region from each candidate region according to the region detection order, for example, taking the case where the target region is determined to be a high attenuation region according to the region attribute, and a candidate region corresponding to the high attenuation region is usually a low gray level candidate region, so that detection may be started from a candidate region with a low gray level; judging whether the current area is used as a target area according to the coupling degree between the area boundary and the gradient edge of the current area, and detecting the target area according to a judgment result; taking a next region located in the current region in the region detection sequence as the current region, and repeatedly performing the step of determining whether to use the current region as the target region according to the coupling degree between the region boundary and the gradient edge of the current region until the detected candidate region and/or the gradient edge satisfy a preset determination end condition, where the determination end condition may include that the number of the detected candidate regions is greater than a preset number threshold, the gradient edge no longer exists, and the like, and is not specifically limited herein.
According to the technical scheme of the embodiment of the invention, whether the current area is the target area or not is judged according to the first similarity of each current pixel point in the current boundary and each edge pixel point in the gradient edge at the pixel position and the second similarity of each current pixel point and each target pixel point in the target boundary of the candidate area detected as the target area at the pixel position, so that the effect of accurately detecting the target area under different conditions is realized.
On this basis, in order to better understand the specific implementation process of the target area detection, the following description is made by way of example with reference to specific examples. Exemplarily, taking the target area as a high attenuation area as an example, gradually traversing each class with low-to-high gray levels in the LBG classification, taking a candidate area of the lowest class which is not detected currently as a current area, comparing the current boundary of the current area with a gradient edge, and if the number of current pixel points which are the same as or similar to each edge pixel point on the gradient edge in the pixel position of the current boundary reaches a certain proportion of the number of all current pixel points, considering that the current area belongs to one part of the high attenuation area; if the above condition is not satisfied, it may be further determined whether a portion of the current boundary is closer to the gradient edge and another portion is closer to the boundary of the region (i.e., the target boundary) that has been detected as the high attenuation region, and if so, the current region may be determined as the high attenuation region. Further, edge pixel points on the gradient edge close to the high attenuation region are assigned to be 0, that is, edge pixel points occupied by the target boundary are set to be 0. The reason for this is that the former should theoretically complete the detection of the target region at this time, but the former is not yet completed in practical application, and may be that an error occurs in the region detection process and the loss should be stopped in time; the latter means that all high attenuation regions have been detected completely, at which point the detection can be stopped.
Correspondingly, the detection process of the direct exposure area and the high attenuation area can change the area detection sequence from low to high into the area detection sequence from high to low, change the lowest class into the highest class, change the class where the relatively larger area gray scale in the LBG classification is traversed into the class where the relatively smaller area gray scale in the LBG classification is traversed, and have the same steps, which is not described herein again.
It should be noted that, in practical applications, it is not necessary to determine in advance whether a high attenuation region or a direct exposure region exists in a medical image, but only when a detection target is the high attenuation region, detection is performed based on a detection step related to the high attenuation region, and when the detection target is the direct exposure region, detection is performed based on a detection step related to the direct exposure region. That is, the region detection method can extract a high attenuation region, a direct exposure region, or a high attenuation region and a direct exposure region from a medical image.
Example four
Fig. 5 is a block diagram of a region detection apparatus according to a fourth embodiment of the present invention, which is configured to execute the region detection method according to any of the embodiments. The device and the region detection method of each embodiment belong to the same inventive concept, and details that are not described in detail in the embodiment of the region detection device may refer to the embodiment of the region detection method. Referring to fig. 5, the apparatus may specifically include: a candidate region determination module 410, a gradient edge determination module 420, and a target region detection module 430.
The candidate region determining module 410 is configured to cluster the gray pixel points according to the pixel gray levels of the gray pixel points in the medical image, and use a region where the gray pixel points belonging to the same category are located as a candidate region;
a gradient edge determining module 420, configured to generate a gradient image corresponding to the medical image according to the gray level of each pixel, and determine a gradient edge in the gradient image according to the pixel gradient of each gradient pixel point in the gradient image;
and the target region detection module 430 is configured to detect a target region from each candidate region according to a coupling degree between a region boundary and a gradient edge of each candidate region.
Optionally, the candidate region determining module 410 may specifically include:
the gray class splitting point determining unit is used for sequencing the pixel gray of each gray pixel point in the medical image and determining a gray class splitting point in a gray sequencing result based on the preset class number;
and the gray pixel point clustering unit is used for taking the gray category splitting point as an initial clustering center and clustering each gray pixel point based on the clustering center and the pixel gray of each gray pixel point.
On this basis, optionally, the gray pixel clustering unit may be specifically configured to:
aiming at each gray pixel point, determining the pixel gray of the gray pixel point and the gray distance between the clustering centers, and clustering the gray pixel points to the category of the clustering center corresponding to the minimum gray distance;
determining gray level distortion according to the minimum gray level distance corresponding to each gray level pixel point, and judging whether the gray level distortion meets a preset clustering end condition;
if not, determining the clustering center of each category again according to the pixel gray levels of the clustered gray level pixel points belonging to the category aiming at each category;
and repeating the step of determining the pixel gray of the gray pixel points and the gray distance between the clustering centers until the gray distortion meets the clustering ending condition, and ending the clustering.
Optionally, the gradient edge determining module 420 may specifically include:
the gray class region screening unit is used for screening a gray class region corresponding to the region attribute of the target region to be detected from each candidate region according to the region gray of each candidate region;
and the first gradient edge determining unit is used for determining the gradient edge in the gradient image according to the pixel gradient of the gradient pixel point corresponding to the pixel point in each region in the gray class region in the gradient image.
Optionally, the gradient edge determining module 420 may specifically include:
the binary image generating unit is used for sequencing the gradient pixel points according to the pixel gradients of the gradient pixel points in the gradient image, screening the gradient pixel points according to the sequencing positions of the gradient pixel points in the sequencing result, and generating a binary image corresponding to the gradient image;
and the second gradient edge determining unit is used for taking the binary edge formed by each binary pixel point in the binary image as the gradient edge of the gradient image.
Optionally, for the current area in each candidate area, the target area detecting module 430 may include:
a similarity obtaining unit, configured to obtain first similarities of each current pixel in a current boundary of the current region and each edge pixel in a gradient edge at a pixel position, and second similarities of each current pixel and each target pixel in a target boundary of a candidate region detected as the target region at a pixel position;
and the first target area detection unit is used for judging whether the current area is the target area or not according to the first similarity or the first similarity and the second similarity so as to realize the detection of the target area.
Optionally, the target area detecting module 430 may specifically include:
the current region screening unit is used for determining the region detection sequence of each candidate region according to the region attribute of the target region to be detected and screening the current region from each candidate region according to the region detection sequence;
the second target area detection unit is used for judging whether the current area is used as the target area according to the coupling degree between the area boundary and the gradient edge of the current area and realizing the detection of the target area according to the judgment result;
and the iteration execution unit is used for taking the next region positioned in the current region in the region detection sequence as the current region, and repeatedly executing the step of judging whether the current region is taken as the target region according to the coupling degree between the region boundary and the gradient edge of the current region until the detected candidate region and/or the gradient edge meet the preset judgment end condition.
In the area detection device provided by the fourth embodiment of the present invention, the candidate area determination module clusters the gray pixels according to the pixel gray levels of the gray pixels in the medical image, and the area where the gray pixels belonging to the same category are located is used as the candidate area; the gradient edge determining module generates a gradient image corresponding to the medical image according to the gray level of each pixel, and determines a gradient edge in the gradient image according to the pixel gradient of each gradient pixel point in the gradient image; and the target area detection module detects the target area from each candidate area according to the coupling degree between the area boundary and the gradient edge of each candidate area. The device solves the problem of low detection precision of the target region caused by uncertain gray level difference between the target region and human tissues, image noise, gray level transition inside the target region and the like aiming at the clustering of the regions, and realizes the effect of accurate detection of the target region by comparing the complete region boundary of the candidate region obtained by the method with the gradient edge of the region boundary with high probability as the target region.
The area detection device provided by the embodiment of the invention can execute the area detection method provided by any embodiment of the invention, and has corresponding functional modules and beneficial effects of the execution method.
It should be noted that, in the embodiment of the area detection apparatus, the included units and modules are only divided according to functional logic, but are not limited to the above division as long as the corresponding functions can be implemented; in addition, specific names of the functional units are only for convenience of distinguishing from each other, and are not used for limiting the protection scope of the present invention.
EXAMPLE five
Fig. 6 is a schematic structural diagram of an area detection apparatus according to a fifth embodiment of the present invention, and as shown in fig. 6, the apparatus includes a memory 510, a processor 520, an input device 530, and an output device 540. The number of processors 520 in the device may be one or more, and one processor 520 is taken as an example in fig. 6; the memory 510, processor 520, input device 530, and output device 540 in the apparatus may be connected by a bus or other means, such as by bus 550 in fig. 6.
The memory 510 is used as a computer-readable storage medium for storing software programs, computer-executable programs, and modules, such as program instructions/modules corresponding to the region detection method in the embodiment of the present invention (for example, the candidate region determining module 410, the gradient edge determining module 420, and the target region detecting module 430 in the region detection apparatus). The processor 520 executes various functional applications of the device and data processing by executing software programs, instructions, and modules stored in the memory 510, that is, implements the area detection method described above.
The memory 510 may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function; the storage data area may store data created according to use of the device, and the like. Further, the memory 510 may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid state storage device. In some examples, memory 510 may further include memory located remotely from processor 520, which may be connected to devices through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The input device 530 may be used to receive input numeric or character information and generate key signal inputs related to user settings and function control of the device. The output device 540 may include a display device such as a display screen.
EXAMPLE six
An embodiment of the present invention provides a storage medium containing computer-executable instructions, which when executed by a computer processor, perform a region detection method, the method including:
clustering each gray pixel point according to the pixel gray of each gray pixel point in the medical image, and taking the area where each gray pixel point belonging to the same category is as a candidate area;
generating a gradient image corresponding to the medical image according to the gray level of each pixel, and determining a gradient edge in the gradient image according to the pixel gradient of each gradient pixel point in the gradient image;
and detecting a target region from each candidate region according to the coupling degree between the region boundary and the gradient edge of each candidate region.
Of course, the storage medium provided by the embodiment of the present invention contains computer-executable instructions, and the computer-executable instructions are not limited to the operations of the method described above, and may also perform related operations in the area detection method provided by any embodiment of the present invention.
From the above description of the embodiments, it is obvious for those skilled in the art that the present invention can be implemented by software and necessary general hardware, and certainly, can also be implemented by hardware, but the former is a better embodiment in many cases. With this understanding, the technical solutions of the present invention may be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as a floppy disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a FLASH Memory (FLASH), a hard disk or an optical disk of a computer, and includes instructions for enabling a computer device (which may be a personal computer, a server, or a network device) to execute the methods according to the embodiments of the present invention.
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.
Claims (10)
1. A method of area detection, comprising:
clustering each gray pixel point according to the pixel gray of each gray pixel point in the medical image, and taking the area where each gray pixel point belonging to the same category is located as a candidate area;
generating a gradient image corresponding to the medical image according to the gray level of each pixel, and determining a gradient edge in the gradient image according to the pixel gradient of each gradient pixel point in the gradient image;
and detecting a target region from each candidate region according to the coupling degree between the region boundary of each candidate region and the gradient edge.
2. The method of claim 1, wherein clustering each gray pixel point according to the pixel gray of each gray pixel point in the medical image comprises:
the method comprises the steps of sequencing pixel gray levels of gray level pixel points in a medical image, and determining gray level category splitting points in a gray level sequencing result based on the preset category number;
and taking the gray class splitting point as an initial clustering center, and clustering each gray pixel point based on the clustering center and the pixel gray of each gray pixel point.
3. The method of claim 2, wherein clustering each of the plurality of gray pixels based on the cluster center and the pixel gray of each of the plurality of gray pixels comprises:
determining the pixel gray of each gray pixel point and the gray distance between the clustering centers for each gray pixel point, and clustering the gray pixel points to the category of the clustering center corresponding to the minimum gray distance;
determining gray level distortion according to the minimum gray level distance corresponding to each gray level pixel point, and judging whether the gray level distortion meets a preset clustering end condition;
if not, re-determining the clustering center of each category according to the pixel gray levels of the gray pixel points belonging to the category after clustering aiming at each category;
and repeating the step of determining the pixel gray of the gray pixel points and the gray distance between the clustering centers until the gray distortion meets the clustering end condition and the clustering is ended.
4. The method of claim 1, wherein determining a gradient edge within the gradient image according to pixel gradients of gradient pixel points in the gradient image comprises:
screening out a gray class region corresponding to the region attribute of the target region to be detected from each candidate region according to the region gray of each candidate region;
and determining a gradient edge in the gradient image according to the pixel gradient of the gradient pixel point corresponding to the pixel point in each area in the gray class area in the gradient image.
5. The method of claim 1, wherein determining a gradient edge within the gradient image according to pixel gradients of gradient pixel points in the gradient image comprises:
sorting each gradient pixel point according to the pixel gradient of each gradient pixel point in the gradient image, screening each gradient pixel point according to the sorting position of each gradient pixel point in the sorting result, and generating a binary image corresponding to the gradient image;
and taking the binary edge formed by each binary pixel point in the binary image as the gradient edge of the gradient image.
6. The method of claim 1, wherein the detecting a target region from each of the candidate regions according to the degree of coupling between the region boundary and the gradient edge of each of the candidate regions for a current region of the candidate regions comprises:
acquiring first similarity of each current pixel point in the current boundary of the current region and each edge pixel point in the gradient edge at a pixel position, and second similarity of each current pixel point and each target pixel point in the target boundary of the candidate region detected as a target region at the pixel position;
and judging whether the current area is the target area or not according to the first similarity or the first similarity and the second similarity so as to realize the detection of the target area.
7. The method of claim 1, wherein the detecting a target region from each of the candidate regions according to the degree of coupling between the region boundary of each candidate region and the gradient edge comprises:
determining the region detection sequence of each candidate region according to the region attribute of a target region to be detected, and screening out the current region from each candidate region according to the region detection sequence;
judging whether the current area is used as the target area or not according to the coupling degree between the area boundary of the current area and the gradient edge, and realizing the detection of the target area according to the judgment result;
and taking the next region located in the current region in the region detection sequence as the current region, and repeatedly executing the step of judging whether the current region is taken as the target region according to the coupling degree between the region boundary of the current region and the gradient edge until the detected candidate region and/or the gradient edge meet the preset judgment end condition.
8. An area detecting apparatus, comprising:
the candidate region determining module is used for clustering each gray pixel point according to the pixel gray of each gray pixel point in the medical image and taking the region where each gray pixel point belonging to the same category is as a candidate region;
the gradient edge determining module is used for generating a gradient image corresponding to the medical image according to each pixel gray level and determining a gradient edge in the gradient image according to the pixel gradient of each gradient pixel point in the gradient image;
and the target region detection module is used for detecting a target region from each candidate region according to the coupling degree between the region boundary of each candidate region and the gradient edge.
9. An area detecting apparatus, characterized by comprising:
one or more processors;
a memory for storing one or more programs;
when executed by the one or more processors, cause the one or more processors to implement the region detection method of any one of claims 1-7.
10. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the region detection method according to any one of claims 1 to 7.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011630482.0A CN114693907A (en) | 2020-12-31 | 2020-12-31 | Region detection method, device, equipment and storage medium |
| PCT/CN2021/112733 WO2022033598A1 (en) | 2020-08-14 | 2021-08-16 | Breast x-ray radiography acquisition method and apparatus, and computer device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011630482.0A CN114693907A (en) | 2020-12-31 | 2020-12-31 | Region detection method, device, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114693907A true CN114693907A (en) | 2022-07-01 |
Family
ID=82133734
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011630482.0A Pending CN114693907A (en) | 2020-08-14 | 2020-12-31 | Region detection method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114693907A (en) |
-
2020
- 2020-12-31 CN CN202011630482.0A patent/CN114693907A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107480677B (en) | Method and device for identifying interest region in three-dimensional CT image | |
| CN110648337A (en) | Hip joint segmentation method, hip joint segmentation device, electronic apparatus, and storage medium | |
| CN112102237A (en) | Brain tumor recognition model training method and device based on semi-supervised learning | |
| CN110517262B (en) | Target detection method, device, equipment and storage medium | |
| US20080260254A1 (en) | Automatic 3-D Object Detection | |
| US20080097942A1 (en) | System and Method for Automated Suspicious Object Boundary Determination | |
| CN111161279A (en) | Medical image segmentation method and device and server | |
| US9317926B2 (en) | Automatic spinal canal segmentation using cascaded random walks | |
| CN114066886B (en) | Bone segmentation boundary determining method and device, electronic equipment and storage medium | |
| JP5462866B2 (en) | Image classification based on image segmentation | |
| JP6422746B2 (en) | Nuclear medicine diagnosis apparatus, medical image processing apparatus, source position estimation method, and source position estimation program | |
| US20210327583A1 (en) | Determination of a growth rate of an object in 3d data sets using deep learning | |
| US11062486B2 (en) | Methods and apparatus for deep learning based data transfer between imaging systems | |
| CN111292298A (en) | Breast cancer pathological typing determination method, device and storage medium | |
| CN115187512B (en) | Method, system, device and medium for predicting invasion risk of large blood vessel of hepatocellular carcinoma | |
| CN114693907A (en) | Region detection method, device, equipment and storage medium | |
| CN114037719B (en) | Bone region determination method and device, electronic equipment and storage medium | |
| CN113160199B (en) | Image recognition method and device, computer equipment and storage medium | |
| CN112529918B (en) | Method, device and equipment for segmenting brain room area in brain CT image | |
| US7379576B2 (en) | Method and system for patient identification in 3D digital medical images | |
| CN111784705B (en) | Method, device, equipment and storage medium for delineating region of interest | |
| EP3494544B1 (en) | Robust pulmonary lobe segmentation | |
| US20220261985A1 (en) | System for determining the presence of features in a dataset | |
| CN115578554B (en) | Vascular focus identification method, device, electronic equipment and readable storage medium | |
| US11721022B2 (en) | Apparatus and method for automated analyses of ultrasound images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |