Method for evaluating the efficacy of drugs for benign prostatic hyperplasia based on ultrasound images

Abstract

The present invention relates to the field of edge detection technology, and more specifically, to a method for evaluating the efficacy of prostate hyperplasia drugs based on ultrasound images. The method comprises: obtaining a prostate ultrasound image from each examination of a prostate patient; determining an abnormal hyperplasia region based on the distribution of edge corner points in the prostate ultrasound image; matching the abnormal hyperplasia regions in prostate ultrasound images from two adjacent examinations to obtain a matching region of the abnormal hyperplasia region; determining the degree of hyperplasia detail preservation based on the differences in pixel values, edge corner points, and vascular morphology between the abnormal hyperplasia region and its matching region; correcting the Gaussian filter standard deviation in Canny edge detection of the prostate ultrasound image to obtain a corrected Gaussian filter standard deviation and determine the true edge of the abnormal hyperplasia region; and evaluating drug efficacy based on the true edge of the abnormal hyperplasia region. The present invention can improve the accuracy of hyperplasia region positioning and accurately evaluate the efficacy of prostate hyperplasia drugs.

Description

Technical Field

[0001] The present invention relates to the technical field of edge detection, and in particular to a method for evaluating the efficacy of a prostate hyperplasia drug based on ultrasound images. Background Art

[0002] Benign Prostatic Hyperplasia (BPH) is a common disease in middle-aged and elderly men. It does not resolve on its own or with medication. Its primary manifestation is prostate tissue hyperplasia, which leads to urinary tract obstruction and symptoms such as frequent urination, urgency, and dysuria. With the aging population, the incidence of BPH is increasing annually. Early diagnosis and accurate evaluation of the efficacy of BPH medications are crucial to improving patients' quality of life.

[0003] In existing methods, the evaluation of the efficacy of prostate hyperplasia drugs mainly relies on the changes in the hyperplasia area in the patient's prostate ultrasound image while taking prostate hyperplasia drugs. When the hyperplasia area is significantly reduced, it indicates that the prostate hyperplasia drug is effective. When the hyperplasia area increases or the changes are not obvious, it indicates that the prostate hyperplasia drug is not effective. This is conducive to timely adjustment of the treatment drugs for prostate hyperplasia patients and improve the treatment effect of prostate hyperplasia. However, in existing methods, the prostate ultrasound image has interference from low contrast and resolution anisotropy. As a result, the Canny edge detection algorithm is prone to smoothing out the edges of small protrusions or depressions in the hyperplasia area in the prostate ultrasound image during the Gaussian filtering process, resulting in inaccurate positioning of the hyperplasia area, making it impossible to accurately analyze the changes in the hyperplasia area, affecting the accurate evaluation of the efficacy of prostate hyperplasia drugs. Summary of the Invention

[0004] In order to solve the technical problem that the Canny edge detection algorithm is prone to smoothing out the edges of tiny protrusions or depressions in the hyperplastic area of ​​the prostate ultrasound image during Gaussian filtering due to the interference of low contrast and resolution anisotropy in the prostate ultrasound image, thereby resulting in inaccurate positioning of the hyperplastic area, the purpose of the present invention is to provide a method for evaluating the efficacy of prostate hyperplasia drugs based on ultrasound images. The technical solution adopted is as follows:

[0005] An embodiment of the present invention provides a method for evaluating the efficacy of a drug for benign prostatic hyperplasia based on ultrasound images, the method comprising the following steps:

[0006] Obtain every prostate ultrasound image from every examination of a prostate patient;

[0007] According to the distribution of edge corner points in each prostate ultrasound image, the abnormal proliferation area in each prostate ultrasound image is obtained;

[0008] Matching the abnormal proliferation area in each examination with the prostate ultrasound image of the previous examination to obtain the matching area of ​​each abnormal proliferation area; obtaining the degree of proliferation detail preservation of each prostate ultrasound image based on the distribution differences of pixel values ​​and edge corner points between each abnormal proliferation area in each prostate ultrasound image and its matching area, as well as the distribution differences of vascular morphology;

[0009] Based on the degree of hyperplasia detail preservation, the Gaussian filter standard deviation in the Canny edge detection process of each prostate ultrasound image is corrected, and the corrected Gaussian filter standard deviation in the Canny edge detection process of each prostate ultrasound image is obtained to determine the true edge of the abnormal hyperplasia area in the prostate ultrasound image of each examination;

[0010] Evaluation of the efficacy of benign prostatic hyperplasia drugs based on the true margins of the abnormally proliferative area.

[0011] Furthermore, the method for obtaining the abnormal proliferation area is:

[0012] For any prostate ultrasound image and any edge corner point in the prostate ultrasound image, a preset number of other edge corner points in the prostate ultrasound image that are closest to the edge corner point are used as neighborhood corner points of the edge corner point;

[0013] Obtain the variance of the Euclidean distance between the edge corner point and each of its neighboring corner points as the distribution characteristic value of the edge corner point;

[0014] For any two edge corner points in the prostate ultrasound image, the sum of the difference in the distribution characteristic values ​​of the two edge corner points and the Euclidean distance between the two edge corner points is used as the distance measurement value of the two edge corner points;

[0015] Clustering the edge corner points in the prostate ultrasound image using a K-means clustering algorithm based on the distance measurement value between any two edge corner points in the prostate ultrasound image to obtain a cluster of corner point clusters in the prostate ultrasound image;

[0016] According to the distribution of edge corner points in each corner point cluster, the abnormality degree of each corner point cluster is obtained;

[0017] When the abnormality level is greater than the preset abnormality threshold, the corresponding corner cluster is regarded as an abnormal cluster;

[0018] For any abnormal cluster, the edge corner points in the abnormal cluster are connected by a region growing algorithm to form a complete region, which is regarded as an abnormal proliferation region in the prostate ultrasound image.

[0019] Furthermore, the method for obtaining the abnormality degree is:

[0020] For any corner point cluster, obtain the Euclidean distance between any two edge corner points in the corner point cluster as the first distance;

[0021] The result of normalizing the product of the variance of all first distances and the inverse of the mean of all first distances is taken as the abnormality degree of the corner cluster.

[0022] Furthermore, the method for obtaining the matching area is:

[0023] The prostate ultrasound images obtained in each examination are arranged according to the order of acquisition to form an image sequence for each examination; wherein the process of acquiring the prostate ultrasound images in each examination is completely consistent;

[0024] For any prostate ultrasound image in the image sequence of any examination, the prostate ultrasound image at the same position as the prostate ultrasound image in the image sequence of the previous examination is used as the matching image of the prostate ultrasound image;

[0025] For any abnormal proliferation area in the prostate ultrasound image, the abnormal proliferation area is matched with the abnormal proliferation area in the matching image using the DeepSORT (DeepLearning Simple Online and Realtime Tracking) target tracking algorithm to obtain a matching area of ​​the abnormal proliferation area.

[0026] Furthermore, the method for obtaining the degree of preservation of the proliferation details is:

[0027] Obtaining a proliferation quantification factor for each abnormal proliferation region according to the distribution differences of pixel values ​​and edge corner points in each abnormal proliferation region and its matching region;

[0028] According to the distribution difference of the vascular morphology in each abnormal proliferation area and its matching area, the blood flow dynamic change value of each abnormal proliferation area is obtained;

[0029] The degree of hyperplasia detail preservation of each prostate ultrasound image is obtained based on the correlation between the hyperplasia quantification factor and the blood flow dynamic change value in each prostate ultrasound image.

[0030] Furthermore, the method for obtaining the proliferation quantification factor is:

[0031] For any abnormal proliferation area, obtaining the average grayscale value of all pixels in the abnormal proliferation area as the first grayscale value;

[0032] Obtaining an average of the grayscale values ​​of all pixels in the matching area of ​​the abnormal proliferation area as a second grayscale value;

[0033] using the difference between the first grayscale value and the second grayscale value as a first growth rate analysis value;

[0034] The difference between the number of edge corner points in the abnormal proliferation area and the matching area is used as the second proliferation velocity analysis value;

[0035] The sum of the first proliferation rate analysis value and the second proliferation rate analysis value is used as a proliferation quantification factor of the abnormal proliferation area.

[0036] Furthermore, the method for obtaining the blood flow dynamic change value is:

[0037] For any abnormal proliferation area, the color Doppler ultrasound image of the abnormal proliferation area is divided into multiple local areas;

[0038] For any local area, a straight line is fitted according to the positions of all edge pixels in the local area by the least square method, which serves as the reference straight line of the local area;

[0039] Obtaining a mean value of the goodness of fit between each edge line in the local area and the reference straight line as a first indicator of the local area;

[0040] Obtaining a ratio of the number of edge pixels in the local area to the number of all pixels as a second indicator of the local area;

[0041] The product of the normalized and negatively correlated result of the first index and the second index is used as the blood vessel analysis value of the local area;

[0042] The mean of the vascular analysis values ​​of all local areas is used as the blood flow dynamic characteristic value of the abnormal proliferation area;

[0043] The difference between the blood flow dynamic characteristic value of the abnormal proliferation area and the matching area is used as the blood flow dynamic change value of the abnormal proliferation area.

[0044] Furthermore, the method for obtaining the degree of preservation of the proliferation details is:

[0045] For any prostate ultrasound image, the proliferation quantification factor and the blood flow dynamic change value of each abnormal proliferation area in the prostate ultrasound image are sorted according to the order of the same abnormal proliferation areas, and the proliferation quantification factor sequence and the blood flow dynamic change value sequence are obtained respectively;

[0046] The result of normalizing the Pearson correlation coefficient of the proliferation quantification factor sequence and the blood flow dynamic change value sequence is used as the degree of preservation of the proliferation details of the prostate ultrasound image.

[0047] Furthermore, the method for obtaining the modified Gaussian filter standard deviation is:

[0048] For any prostate ultrasound image, the product of the degree of hyperplasia detail preservation of the prostate ultrasound image and the Gaussian filter standard deviation during the Canny edge detection process of the prostate ultrasound image is used as the modified Gaussian filter standard deviation during the Canny edge detection process of the prostate ultrasound image.

[0049] Furthermore, the method for obtaining the edge corner points is:

[0050] The edge corner points in each prostate ultrasound image are obtained using the Harris corner detection algorithm.

[0051] The present invention has the following beneficial effects:

[0052] The present invention first obtains the abnormal hyperplasia area in each prostate ultrasound image based on the distribution of edge corner points in each prostate ultrasound image, preparing for the subsequent accurate and efficient analysis of the loss of hyperplasia details in each prostate ultrasound image; in order to accurately analyze the loss of hyperplasia details, each examination is matched with the abnormal hyperplasia area in the prostate ultrasound image of the adjacent previous examination, and the matching area of ​​each abnormal hyperplasia area is obtained, and the area corresponding to the same hyperplasia site is accurately determined, which is conducive to a more comprehensive analysis of hyperplasia details; and then, based on the distribution differences in pixel values ​​and edge corner points between each abnormal hyperplasia area in each prostate ultrasound image and its matching area, as well as the distribution differences in vascular morphology, the degree of hyperplasia detail retention of each prostate ultrasound image is obtained, accurately reflecting the retention of hyperplasia details in each prostate ultrasound image. This is beneficial for the subsequent accurate adjustment of the filtering degree in the process of identifying the edge of the abnormal hyperplasia area through the Canny edge detection algorithm; and then the Gaussian filter standard deviation in the Canny edge detection process of each prostate ultrasound image is corrected based on the degree of hyperplasia detail retention, and the corrected Gaussian filter standard deviation in the Canny edge detection process of each prostate ultrasound image is obtained, so that the true edge of the abnormal hyperplasia area in the prostate ultrasound image of each examination can be accurately determined, and then the abnormal hyperplasia area of ​​each examination can be accurately obtained, so that the changes in the abnormal hyperplasia area can be accurately analyzed, which is beneficial for the accurate evaluation of the efficacy of prostate hyperplasia drugs, and then the efficacy of prostate hyperplasia drugs is accurately evaluated based on the true edge of the abnormal hyperplasia area, so that doctors can adjust the treatment plan of prostate patients in time, effectively improving the treatment effect of prostate patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0054] Figure 1 A schematic flow chart of a method for evaluating the efficacy of a drug for benign prostatic hyperplasia based on ultrasound images provided by one embodiment of the present invention;

[0055] Figure 2 A flow chart of a method for obtaining the degree of preservation of proliferation details provided by one embodiment of the present invention;

[0056] Figure 3 This is a structural diagram of a prostate hyperplasia drug efficacy evaluation system based on ultrasound images provided by one embodiment of the present invention;

[0057] Figure 4 A schematic diagram of a computer device provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0058] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation, structure, features and effects of the ultrasound-based benign prostatic hyperplasia drug efficacy evaluation method proposed by the present invention. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0059] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0060] The specific scheme of the method for evaluating the efficacy of benign prostatic hyperplasia drugs based on ultrasound images provided by the present invention is described in detail below with reference to the accompanying drawings.

[0061] Example 1:

[0062] The present invention proposes a method for evaluating the efficacy of prostate hyperplasia drugs based on ultrasound images. Figure 1 , which shows a schematic flow chart of a method for evaluating the efficacy of a drug for benign prostatic hyperplasia based on ultrasound images according to an embodiment of the present invention, the method comprising the following steps:

[0063] Step S1: Acquire each prostate ultrasound image of each prostate patient during each examination.

[0064] Specifically, this embodiment uses a prostate patient and a benign prostatic hyperplasia drug as an example for analysis. It should be noted that subsequent prostate patients refer to this prostate patient, and benign prostatic hyperplasia drugs refer to this type of benign prostatic hyperplasia drug. To analyze the efficacy of the benign prostatic hyperplasia drug taken by the prostate patient, this embodiment first uses the examination when the doctor first prescribes the benign prostatic hyperplasia drug to the prostate patient as the first examination. The prostate patient has not yet taken the benign prostatic hyperplasia drug at this examination. It is known that the treatment of benign prostatic hyperplasia is a relatively long process. In order to reasonably and timely analyze the efficacy of the benign prostatic hyperplasia drug and enable the doctor to adjust the prostate patient's treatment plan in a timely manner, this embodiment sets the prostate patient to come to the hospital for an examination every other month. During each examination, a high-resolution ultrasound device is used to obtain the prostate ultrasound image of each examination of the prostate patient. At the same time, it is set that the prostate patient needs to come to the hospital for two more examinations starting from the first examination to analyze the efficacy of the benign prostatic hyperplasia drug. The implementer can set the time interval between two adjacent examinations of the prostate patient and the number of examinations for analyzing the efficacy of the benign prostatic hyperplasia drug according to actual conditions, and there is no limitation here.

[0065] To accurately examine prostate hyperplasia (BPH) in prostate patients, this embodiment automatically performs ultrasound scans of the patient's prostate area at intervals of a specified length during each examination. Consequently, multiple prostate ultrasound images are acquired during each examination, enabling more accurate analysis of BPH. In this embodiment, the specified length is set to 0.5 mm. The user may adjust the specified length based on actual circumstances and is not limited here.

[0066] To avoid interference and accurately and efficiently identify hyperplastic areas within the prostate region, this embodiment uses the Otsu threshold segmentation algorithm to automatically distinguish the prostate region from the background region in prostate ultrasound images. The Otsu threshold segmentation algorithm is well-known and will not be described in detail here. It should be noted that subsequent analysis will focus solely on the prostate region within the prostate ultrasound image.

[0067] Step S2: Obtain the abnormal proliferation area in each prostate ultrasound image according to the distribution of edge corner points in each prostate ultrasound image.

[0068] It is known that prostatic hyperplasia is a localized proliferation of prostate tissue, which is usually manifested as a local protrusion or depression of the glandular contour. The morphological changes in the hyperplastic area in the prostate ultrasound image will cause the edge to be more irregular than the edge of normal tissue. Therefore, this embodiment first obtains the edge line in each prostate ultrasound image through the Canny edge detection algorithm, and then uses the Harris corner detection algorithm to extract the edge corner points on the edge line. The denser and more irregular the distribution of edge corner points on a certain edge line, the more likely the area corresponding to the edge line is an abnormal hyperplasia area. Therefore, this embodiment obtains the abnormal hyperplasia area in each prostate ultrasound image based on the distribution of edge corner points in each prostate ultrasound image. Among them, the Canny edge detection algorithm and the Harris corner detection algorithm are both well-known technologies and will not be described in detail.

[0069] Preferably, in a manner that can be implemented in this embodiment, the method for obtaining the abnormal hyperplasia area is: for any prostate ultrasound image and any edge corner point in the prostate ultrasound image, a preset number of other edge corner points in the prostate ultrasound image that are closest to the edge corner point are used as the neighborhood corner points of the edge corner point; this embodiment sets the preset number to 8, and the implementer can set the size of the preset number according to actual conditions, which is not limited here. The variance of the Euclidean distance between the edge corner point and each of its neighborhood corner points is obtained as the distribution characteristic value of the edge corner point; the larger the distribution characteristic value, the more irregular the distribution of the edge corner point, which indirectly indicates that the edge corner point is more likely to be the edge pixel point corresponding to the hyperplasia area. Among them, the method for obtaining the Euclidean distance is a well-known technology and will not be described in detail;

[0070] In order to filter out the hyperplasia area based on the edge corner points, it is necessary to first divide the edge corner points. It is known that the distances between the edge corner points corresponding to the same area are similar, and the distribution characteristic values ​​are also similar. Therefore, in this embodiment, for any two edge corner points in the prostate ultrasound image, the sum of the absolute value of the difference between the distribution characteristic values ​​of the two edge corner points and the Euclidean distance of the two edge corner points is used as the distance measurement value of the two edge corner points; when the distance measurement value is smaller, the two edge corner points are more likely to be edge pixel points corresponding to the same area; therefore, in this embodiment, according to the distance measurement value of any two edge corner points in the prostate ultrasound image, the edge corner points in the prostate ultrasound image are clustered by the K-means clustering algorithm to obtain the corner point clustering cluster in the prostate ultrasound image; it should be noted that this embodiment obtains the K value in the K-means clustering algorithm by the elbow method. Among them, the K-means clustering algorithm and the elbow method are both well-known technologies and will not be described in detail;

[0071] It is known that the distribution characteristics of the edge corner points in the proliferative area are irregular and dense. Therefore, this embodiment obtains the abnormality degree of each corner point cluster based on the distribution of the edge corner points in each corner point cluster. The greater the abnormality degree, the more likely the area corresponding to the corner point cluster is a proliferative area. The abnormality degree is obtained by: for any corner point cluster, the Euclidean distance between any two edge corner points in the corner point cluster is obtained as the first distance. When the variance of all the first distances is larger, it means that the distribution of the edge corner points in the corner point cluster is more irregular. The area corresponding to the corner point cluster is more likely to be a proliferation area; at the same time, when the mean of all first distances is smaller, it means that the edge corner points in the corner point cluster are more densely distributed, further indicating that the area corresponding to the corner point cluster is more likely to be a proliferation area; furthermore, this embodiment normalizes the product of the variance of all first distances and the inverse of the mean of all first distances as the abnormality degree of the corner point cluster; this embodiment normalizes the product of the variance of all first distances and the inverse of the mean of all first distances using the norm normalization function;

[0072] It is known that the greater the degree of abnormality, the more likely the area corresponding to the corresponding corner point cluster is a hyperplastic area. Therefore, this embodiment sets the preset abnormality threshold at 0.5. The implementer can set the preset abnormality threshold according to actual conditions, and this is not limited here. When the abnormality is greater than the preset abnormality threshold, the corresponding corner point cluster is identified as an abnormal cluster, that is, the corner point cluster corresponding to the hyperplastic area is screened out. Considering that the process of obtaining edge lines using the Canny edge detection algorithm and processing through Gaussian filtering can easily smooth out some edges of the hyperplastic area, this embodiment, for any abnormal cluster, connects the edge corner points in the abnormal cluster using the region growing algorithm to form a complete region, which is identified as an abnormal hyperplastic area in the prostate ultrasound image. The region growing algorithm is a well-known technology and will not be described in detail.

[0073] At this point, the abnormal hyperplasia region in each prostate ultrasound image is obtained, that is, the hyperplasia site in each prostate ultrasound image is determined. It should be noted that in this embodiment, subsequent analysis is not performed directly on prostate ultrasound images that do not have abnormal hyperplasia regions.

[0074] Step S3: Match each examination with the abnormal hyperplasia area in the prostate ultrasound image of the adjacent previous examination to obtain the matching area of ​​each abnormal hyperplasia area; obtain the degree of hyperplasia detail preservation of each prostate ultrasound image based on the distribution differences of pixel values ​​and edge corner points between each abnormal hyperplasia area in each prostate ultrasound image and its matching area, as well as the distribution differences of vascular morphology.

[0075] Specifically, the abnormal hyperplasia region obtained in step S3 is not constructed based on the actual edge line of the abnormal hyperplasia region. There may be a deviation in the positioning of the abnormal hyperplasia region and the actual hyperplasia region, which in turn affects the subsequent accurate analysis of changes in the abnormal hyperplasia region and the accurate evaluation of the efficacy of benign prostatic hyperplasia drugs. Therefore, this embodiment requires analyzing the loss of hyperplasia details in each prostate ultrasound image, and then adjusting the degree of Gaussian filtering in the Canny edge detection algorithm. Further correction is made to the edge of each abnormal hyperplasia region to ensure that the positioning of the abnormal hyperplasia region and the actual hyperplasia region are more consistent.

[0076] In actual practice, when benign prostatic hyperplasia worsens, the grayscale value of the hyperplastic region will increase and the edge will become more irregular. Therefore, this embodiment determines the hyperplasia of the abnormal hyperplastic region by analyzing the grayscale changes and the changes in the number of edge corners in the abnormal hyperplastic region. On the other hand, when benign prostatic hyperplasia worsens, the local blood flow demand in the hyperplastic region increases, resulting in vascular dilation and increased blood flow, which will form a vascular morphology that appears smaller and denser. Therefore, this embodiment can also determine the hyperplasia of the abnormal hyperplastic region by analyzing the changes in the vascular morphology of the abnormal hyperplastic region. The more consistent the hyperplasia determined by the changes in the grayscale value and the number of edge corners of the abnormal hyperplastic region is with the hyperplasia determined by the changes in the vascular morphology, the more consistent the positioning of the abnormal hyperplastic region is with the actual hyperplastic region, the less the edge is smoothed, and the less hyperplastic details are lost. Furthermore, this embodiment first matches the abnormal hyperplasia region in each examination with the prostate ultrasound image of the immediately preceding examination to obtain a matching region for each abnormal hyperplasia region. The abnormal hyperplasia region and its matching region are essentially the same hyperplasia region in the two adjacent examinations. It should be noted that prostate hyperplasia does not disappear with drug treatment, so the same abnormal hyperplasia region in two adjacent examinations will definitely be successfully matched. The degree of hyperplasia detail preservation in each prostate ultrasound image is then determined based on the differences in pixel values ​​and edge corner point distributions, as well as the differences in vascular morphology distributions, between each abnormal hyperplasia region and its matching region in each prostate ultrasound image. The greater the degree of hyperplasia detail preservation, the less adjustment is required for the Gaussian filtering level in the Canny edge detection algorithm in the corresponding prostate ultrasound image.

[0077] Preferably, in one possible implementation of this embodiment, the matching region is obtained by arranging the prostate ultrasound images obtained for each examination according to the order of acquisition to form an image sequence for each examination; wherein the process of obtaining the prostate ultrasound images for each examination is completely consistent, and therefore the number of prostate ultrasound images corresponding to each examination is the same, and prostate ultrasound images at the same position in different image sequences correspond to prostate regions at the same position. For any prostate ultrasound image in the image sequence of any examination, the prostate ultrasound image at the same position as the prostate ultrasound image in the image sequence of the examination immediately preceding the examination is used as the matching image for the prostate ultrasound image;

[0078] For any abnormal hyperplasia area in the prostate ultrasound image, the abnormal hyperplasia area is matched with the abnormal hyperplasia area in the matching image using the DeepSORT target tracking algorithm to accurately obtain the matching area of ​​the abnormal hyperplasia area, wherein the abnormal hyperplasia area and its matching area correspond to the same hyperplasia site. The DeepSORT target tracking algorithm is a well-known technology and will not be described in detail. It should be noted that there is no adjacent previous examination for the first examination. Therefore, this embodiment does not obtain the matching area of ​​each abnormal hyperplasia area in each prostate ultrasound image of the first examination.

[0079] Preferably, in one possible implementation of this embodiment, the method for obtaining the degree of preservation of proliferation details can be found in Figure 2 , which shows a flow chart of a method for obtaining the degree of preservation of proliferation details provided by this embodiment, the method comprising the following steps:

[0080] Step S201: obtaining a proliferation quantization factor for each abnormal proliferation region according to the distribution differences of pixel values ​​and edge corner points in each abnormal proliferation region and its matching region.

[0081] When an abnormally proliferative region has a larger grayscale value and more edge corners than its matching region, it indicates that the abnormally proliferative region exhibits more severe hyperplasia. Furthermore, this embodiment obtains a hyperplasia quantization factor for each abnormally proliferative region based on the difference in pixel values ​​and edge corner distribution between each abnormally proliferative region and its matching region. The larger the hyperplasia quantization factor, the more severe the hyperplasia exhibited by the corresponding abnormally proliferative region.

[0082] In one implementation of this embodiment, a method for obtaining a proliferation quantization factor is as follows: for any abnormal proliferation region, obtaining the average grayscale value of all pixels in the abnormal proliferation region as a first grayscale value; obtaining the average grayscale value of all pixels in a matching region of the abnormal proliferation region as a second grayscale value; using the difference between the first grayscale value and the second grayscale value as a first proliferation rate analysis value; using the difference between the number of edge corner points in the abnormal proliferation region and its matching region as a second proliferation rate analysis value; when the first proliferation rate analysis value and the second proliferation rate analysis value are both larger, it indicates that the abnormal proliferation region is changing towards severe proliferation; in order to accurately obtain the proliferation information representing the abnormal proliferation region, the sum of the first proliferation rate analysis value and the second proliferation rate analysis value is used as the proliferation quantization factor of the abnormal proliferation region.

[0083] At this point, the proliferation quantification factor of each abnormal proliferation area is obtained.

[0084] Step S202: obtaining a blood flow dynamic change value of each abnormal proliferation region according to the distribution difference of the blood vessel morphology in each abnormal proliferation region and its matching region.

[0085] When a certain abnormal hyperplasia region exhibits more bends and irregularities in its blood vessels compared to its matching region, it indicates that the hyperplasia in the abnormal hyperplasia region is more severe. Furthermore, this embodiment obtains a blood flow dynamic change value for each abnormal hyperplasia region based on the distribution difference in blood vessel morphology between each abnormal hyperplasia region and its matching region. The larger the blood flow dynamic change value, the more severe the hyperplasia in the corresponding abnormal hyperplasia region.

[0086] In one possible implementation of this embodiment, the method for obtaining the blood flow dynamic change value is as follows: it is known that the vascular morphological changes corresponding to the hyperplastic area are small and dense. Therefore, for any abnormal hyperplastic area, this embodiment obtains a color Doppler ultrasound image of the abnormal hyperplastic area, because the vascular morphology can be accurately analyzed through the color Doppler ultrasound image, and then divides the color Doppler ultrasound image of the abnormal hyperplastic area into multiple local areas; this embodiment sets the size of the local area to , the implementer can set the size of the local area according to the actual situation, and it is not limited here. It should be noted that when dividing the local area, if the remaining boundary is less than a complete local area, the local area at the boundary may partially overlap with the adjacent local area. In order to accurately analyze the distribution of blood vessels in the abnormal proliferation area, and then analyze each local area separately, for any local area, a straight line is fitted according to the positions of all edge pixels in the local area by the least squares method, which serves as the reference straight line for the local area; when the goodness of fit between a certain edge line and the reference straight line is smaller, the greater the degree of curvature of the edge line, and when the degree of curvature of the edge lines in the local area is greater, the more obvious the blood flow dynamics represented by the local area, and then this embodiment obtains the mean of the goodness of fit between each edge line in the local area and the reference straight line as the first indicator of the local area; the smaller the first indicator, the more obvious the blood flow dynamics represented by the local area. Among them, the method of fitting a straight line and obtaining the goodness of fit by the least squares method are both well-known technologies and will not be described in detail;

[0087] To more accurately analyze the blood flow dynamics in the local area, the density of the edge lines in the local area is further analyzed, and the ratio of the number of edge pixels in the local area to the total number of pixels is obtained as the second index of the local area. The larger the second index, the denser the blood vessels in the local area are, and the more obvious the blood flow dynamics represented by the local area.

[0088] In order to accurately represent the blood flow dynamics of the local area, this embodiment further multiplies the product of the normalized and negatively correlated result of the first indicator and the second indicator as the blood vessel analysis value of the local area; the larger the blood vessel analysis value, the more obvious the blood flow dynamics of the local area.

[0089] The calculation formula for the vascular analysis value is: Where, is the vascular analysis value of the ath local area; is the first index of the a-th local area; is the number of edge pixels in the a-th local area; is the number of all pixels in the a-th local area; is the second index of the a-th local area; norm is the normalization function.

[0090] In order to comprehensively analyze the blood flow dynamics of the abnormal proliferation area, the average of the vascular analysis values ​​of all local areas is used as the blood flow dynamic characteristic value of the abnormal proliferation area; the larger the blood flow dynamic characteristic value, the more obvious the blood flow dynamics of the abnormal proliferation area.

[0091] To analyze whether the abnormal hyperplasia region is showing signs of severe hyperplasia, the difference between the blood flow dynamics characteristic values ​​of the abnormal hyperplasia region and its matching region is used as the blood flow dynamics change value of the abnormal hyperplasia region. The larger the blood flow dynamics change value, the more severe the abnormal hyperplasia region's information is.

[0092] At this point, the blood flow dynamic change value of each abnormal proliferation area is obtained.

[0093] Step S203: Obtaining the degree of hyperplasia detail preservation of each prostate ultrasound image based on the correlation between the hyperplasia quantification factor and the blood flow dynamic change value in each prostate ultrasound image.

[0094] When the proliferation quantification factor and the blood flow dynamic change value of each abnormal proliferation area in a prostate ultrasound image show the same change trend, it means that the details of the proliferation area in the prostate ultrasound image are lost less. In this embodiment, the degree of proliferation detail preservation of each prostate ultrasound image is obtained based on the correlation between the proliferation quantification factor and the blood flow dynamic change value in each prostate ultrasound image.

[0095] In one possible implementation of this embodiment, the method for obtaining the degree of hyperplasia detail preservation is as follows: for any prostate ultrasound image, the hyperplasia quantification factor and blood flow dynamic change value of each abnormal hyperplasia region in the prostate ultrasound image are sorted according to the same order of abnormal hyperplasia regions, thereby obtaining a hyperplasia quantification factor sequence and a blood flow dynamic change value sequence, respectively. It should be noted that the order of the abnormal hyperplasia regions in the prostate ultrasound image can be randomly set and is not limited herein, but it must be ensured that data at the same position in the hyperplasia quantification factor sequence and the blood flow dynamic change value sequence correspond to the same abnormal hyperplasia region. The Pearson correlation coefficient between the hyperplasia quantification factor sequence and the blood flow dynamic change value sequence is then obtained. A larger Pearson correlation coefficient indicates a more consistent trend between the hyperplasia quantification factor and the blood flow dynamic change value of the abnormal hyperplasia region, indirectly reflecting a lesser loss of hyperplasia detail in the prostate ultrasound image. Furthermore, in this embodiment, the Pearson correlation coefficient of the hyperplasia quantification factor sequence and the blood flow dynamic change value sequence is normalized to determine the degree of hyperplasia detail preservation in the prostate ultrasound image. In this embodiment, the Pearson correlation coefficient of the proliferation quantification factor sequence and the blood flow dynamic change value sequence is normalized by the norm normalization function. The method for obtaining the Pearson correlation coefficient is a well-known technology and will not be described in detail.

[0096] At this point, the degree of preservation of hyperplasia details of each prostate ultrasound image is obtained.

[0097] Step S4: Based on the degree of hyperplasia detail preservation, the Gaussian filter standard deviation in the Canny edge detection process of each prostate ultrasound image is corrected, the corrected Gaussian filter standard deviation in the Canny edge detection process of each prostate ultrasound image is obtained, and the true edge of the abnormal hyperplasia area in the prostate ultrasound image of each examination is determined.

[0098] It is known that when the edge line in the prostate ultrasound image is obtained by the Canny edge detection algorithm, the prostate ultrasound image will be Gaussian filtered. When the Gaussian filter standard deviation is larger, the filtering degree is greater, and the edge details are more likely to be lost; when the Gaussian filter standard deviation is smaller, the filtering degree is smaller, and more edge details are retained; in order to avoid the edge of the hyperplasia area being lost, resulting in inaccurate identification of the hyperplasia area, this embodiment corrects the Gaussian filter standard deviation in the Canny edge detection process of each prostate ultrasound image based on the degree of hyperplasia detail retention, obtains the corrected Gaussian filter standard deviation in the Canny edge detection process of each prostate ultrasound image, and then accurately determines the true edge of the abnormal hyperplasia area in the prostate ultrasound image of each examination.

[0099] Preferably, in one possible implementation of this embodiment, the modified Gaussian filter standard deviation is obtained by multiplying the degree of hyperplasia detail preservation in any prostate ultrasound image by the Gaussian filter standard deviation during Canny edge detection for the prostate ultrasound image, and using this as the modified Gaussian filter standard deviation during Canny edge detection for the prostate ultrasound image. It should be noted that the Gaussian filter standard deviation during Canny edge detection for the prostate ultrasound image is known.

[0100] At this point, the modified Gaussian filter standard deviation of each prostate ultrasound image in the Canny edge detection process is obtained, and the true edge of the abnormal hyperplasia area in each prostate ultrasound image of each examination is accurately determined, which is conducive to the subsequent timely and accurate evaluation of the efficacy of prostate hyperplasia drugs.

[0101] Step S5: Evaluate the efficacy of the benign prostatic hyperplasia drug based on the true margin of the abnormal hyperplasia area.

[0102] Specifically, this embodiment constructs a 3D model based on the true edges of the abnormally proliferative regions in each prostate ultrasound image for each examination. The volume of each abnormally proliferative region is then calculated based on the cumulative volume of all abnormally proliferative regions for each examination, representing the total volume of the hyperplasia for each examination. It should be noted that the total volume of the hyperplasia for the initial examination is not calculated.

[0103] When the total volume of hyperplasia detected in the second test is greater than the total volume of hyperplasia detected in the third test, it indicates that the treatment effect of prostatic hyperplasia is better, indirectly reflecting the better efficacy of the prostatic hyperplasia drug. Furthermore, this embodiment obtains the difference between the total volume of hyperplasia detected in the second test and the total volume of hyperplasia detected in the third test as the volume change value; then, the ratio of the volume change value to the total volume of hyperplasia detected in the second test is used as the prostatic hyperplasia drug efficacy assessment value. The larger the prostatic hyperplasia drug efficacy assessment value, the better the prostatic hyperplasia drug efficacy. Therefore, this embodiment sets the preset assessment threshold value to 0.3. The implementer can set the preset assessment threshold value according to actual circumstances and is not limited here. When the prostatic hyperplasia drug efficacy assessment value is greater than the preset assessment threshold value, it indicates that the prostatic hyperplasia drug treatment is effective and the prostate patient can continue to take the prostatic hyperplasia drug. When the prostatic hyperplasia drug efficacy assessment value is less than or equal to the preset assessment threshold value, it indicates that the volume change in the hyperplasia area is small, indirectly indicating that the prostatic hyperplasia drug efficacy is limited. In this case, the doctor needs to adjust the prostate patient's medication, optimize the prostate patient's drug treatment plan, and improve the prostate patient's treatment effect.

[0104] In summary, this embodiment obtains a prostate ultrasound image of a prostate patient for each examination; obtains an abnormal hyperplasia area based on the distribution of edge corners in the prostate ultrasound image; matches the abnormal hyperplasia areas in the prostate ultrasound images of two adjacent examinations to obtain a matching area of ​​the abnormal hyperplasia area; obtains the degree of hyperplasia detail retention based on the distribution differences in pixel values, edge corners, and vascular morphology between the abnormal hyperplasia area and its matching area, corrects the Gaussian filter standard deviation in Canny edge detection of the prostate ultrasound image, obtains a corrected Gaussian filter standard deviation, and determines the true edge of the abnormal hyperplasia area. By correcting the Gaussian filter standard deviation in Canny edge detection, the present invention effectively improves the accuracy of hyperplasia area positioning, which is conducive to accurately evaluating the efficacy of prostate hyperplasia drugs.

[0105] Example 2:

[0106] The present invention also proposes a prostate hyperplasia drug efficacy evaluation system based on ultrasound images, please refer to Figure 3 , which shows a structural diagram of a prostate hyperplasia drug efficacy evaluation system based on ultrasound images provided by an embodiment of the present invention. The system includes: an image acquisition module 10, an abnormal hyperplasia area acquisition module 20, a hyperplasia detail retention degree acquisition module 30, a real edge acquisition module 40 and an evaluation module 50.

[0107] The image acquisition module 10 is used to acquire each prostate ultrasound image of a prostate patient during each examination.

[0108] The abnormal proliferation region acquisition module 20 is used to acquire the abnormal proliferation region in each prostate ultrasound image according to the distribution of edge corner points in each prostate ultrasound image.

[0109] The hyperplasia detail retention degree acquisition module 30 is used to match the abnormal hyperplasia area in each examination with the prostate ultrasound image of the adjacent previous examination to obtain the matching area of ​​each abnormal hyperplasia area; based on the distribution differences of pixel values ​​and edge corner points between each abnormal hyperplasia area in each prostate ultrasound image and its matching area, as well as the distribution differences of vascular morphology, the hyperplasia detail retention degree of each prostate ultrasound image is obtained.

[0110] a true edge acquisition module 40 for correcting the Gaussian filter standard deviation during Canny edge detection of each prostate ultrasound image based on the degree of hyperplasia detail preservation, obtaining the corrected Gaussian filter standard deviation during Canny edge detection of each prostate ultrasound image, and determining the true edge of the abnormal hyperplasia region in each prostate ultrasound image examined;

[0111] The evaluation module 50 evaluates the efficacy of the benign prostatic hyperplasia drug based on the actual margin of the abnormal hyperplasia area.

[0112] It should be noted that the system provided in the above embodiment is merely exemplified by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above. In addition, the above embodiment provides a prostate hyperplasia drug efficacy evaluation system based on ultrasound images and a prostate hyperplasia drug efficacy evaluation method based on ultrasound images. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0113] Example 3:

[0114] The present invention also proposes a device for evaluating the efficacy of a drug for benign prostatic hyperplasia based on ultrasound images. The device includes a memory and a processor. The memory stores executable program code, and the processor is configured to call and execute the executable program code to perform a method for evaluating the efficacy of a drug for benign prostatic hyperplasia based on ultrasound images, as provided in an embodiment of the present application. The device can be a chip, component, or module. The chip may include a connected processor and memory. The memory is configured to store instructions. When the processor calls and executes the instructions, the chip can execute the method for evaluating the efficacy of a drug for benign prostatic hyperplasia based on ultrasound images, as provided in the above embodiment.

[0115] In addition, the present invention also protects a computer device, see Figure 4The computer device includes a memory 401, a processor 402, and a computer program 403 stored in the memory 401 and running on the processor 402, wherein when the processor 402 executes the computer program 403, the computer device can execute any of the ultrasound image-based prostate hyperplasia drug efficacy evaluation methods introduced above.

[0116] Example 4:

[0117] The present invention also provides a computer-readable storage medium, which stores computer program code. When the computer program code is run on a computer, the computer executes the above-mentioned related method steps to implement a method for evaluating the efficacy of prostate hyperplasia drugs based on ultrasound images provided in the above embodiment.

[0118] Example 5:

[0119] The present invention also provides a computer program product. When the computer program product is run on a computer, it enables the computer to execute the above-mentioned related steps to implement the ultrasound image-based prostate hyperplasia drug efficacy evaluation method provided in the above embodiment.

[0120] Among them, the device, computer-readable storage medium, computer program product or chip provided in this embodiment are all used to execute the corresponding methods provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

[0121] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0122] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A method for evaluating the efficacy of prostate hyperplasia drugs based on ultrasound images, characterized in that: The method comprises the following steps: Obtain every prostate ultrasound image from every examination of a prostate patient; According to the distribution of edge corner points in each prostate ultrasound image, the abnormal proliferation area in each prostate ultrasound image is obtained; Matching the abnormal proliferation area in each examination with the prostate ultrasound image of the previous examination to obtain the matching area of ​​each abnormal proliferation area; obtaining the degree of proliferation detail preservation of each prostate ultrasound image based on the distribution differences of pixel values ​​and edge corner points between each abnormal proliferation area in each prostate ultrasound image and its matching area, as well as the distribution differences of vascular morphology; Based on the degree of hyperplasia detail preservation, the Gaussian filter standard deviation in the Canny edge detection process of each prostate ultrasound image is corrected, and the corrected Gaussian filter standard deviation in the Canny edge detection process of each prostate ultrasound image is obtained to determine the true edge of the abnormal hyperplasia area in the prostate ultrasound image of each examination; Evaluate the efficacy of BPH drugs based on the true margins of the abnormally proliferative area; The method for obtaining the degree of preservation of the proliferation details is as follows: Obtaining a proliferation quantification factor for each abnormal proliferation region according to the distribution differences of pixel values ​​and edge corner points in each abnormal proliferation region and its matching region; According to the distribution difference of the vascular morphology in each abnormal proliferation area and its matching area, the blood flow dynamic change value of each abnormal proliferation area is obtained; The degree of hyperplasia detail preservation of each prostate ultrasound image is obtained based on the correlation between the hyperplasia quantification factor and the blood flow dynamic change value in each prostate ultrasound image.

2. The method for evaluating the efficacy of prostate hyperplasia drugs based on ultrasound images according to claim 1, characterized in that: The method for obtaining the abnormal proliferation area is: For any prostate ultrasound image and any edge corner point in the prostate ultrasound image, a preset number of other edge corner points in the prostate ultrasound image that are closest to the edge corner point are used as neighborhood corner points of the edge corner point; Obtain the variance of the Euclidean distance between the edge corner point and each of its neighboring corner points as the distribution characteristic value of the edge corner point; For any two edge corner points in the prostate ultrasound image, the sum of the difference in the distribution characteristic values ​​of the two edge corner points and the Euclidean distance between the two edge corner points is used as the distance measurement value of the two edge corner points; Clustering the edge corner points in the prostate ultrasound image using a K-means clustering algorithm based on the distance measurement value between any two edge corner points in the prostate ultrasound image to obtain a cluster of corner point clusters in the prostate ultrasound image; According to the distribution of edge corner points in each corner point cluster, the abnormality degree of each corner point cluster is obtained; When the abnormality level is greater than the preset abnormality threshold, the corresponding corner cluster is regarded as an abnormal cluster; For any abnormal cluster, the edge corner points in the abnormal cluster are connected by a region growing algorithm to form a complete region, which is regarded as an abnormal proliferation region in the prostate ultrasound image.

3. The method for evaluating the efficacy of prostate hyperplasia drugs based on ultrasound images according to claim 2, wherein: The method for obtaining the abnormality degree is: For any corner point cluster, obtain the Euclidean distance between any two edge corner points in the corner point cluster as the first distance; The result of normalizing the product of the variance of all first distances and the inverse of the mean of all first distances is taken as the abnormality degree of the corner cluster.

4. The method for evaluating the efficacy of prostate hyperplasia drugs based on ultrasound images according to claim 1, wherein: The method for obtaining the matching area is: The prostate ultrasound images obtained in each examination are arranged according to the order of acquisition to form an image sequence for each examination; wherein the process of acquiring the prostate ultrasound images in each examination is completely consistent; For any prostate ultrasound image in the image sequence of any examination, the prostate ultrasound image at the same position as the prostate ultrasound image in the image sequence of the previous examination is used as the matching image of the prostate ultrasound image; For any abnormal proliferation region in the prostate ultrasound image, the abnormal proliferation region is matched with the abnormal proliferation region in the matching image by using the DeepSORT target tracking algorithm to obtain a matching region of the abnormal proliferation region.

5. The method for evaluating the efficacy of prostate hyperplasia drugs based on ultrasound images according to claim 1, wherein: The method for obtaining the proliferation quantification factor is: For any abnormal proliferation area, obtaining the average grayscale value of all pixels in the abnormal proliferation area as the first grayscale value; Obtaining an average of the grayscale values ​​of all pixels in the matching area of ​​the abnormal proliferation area as a second grayscale value; using the difference between the first grayscale value and the second grayscale value as a first growth rate analysis value; The difference between the number of edge corner points in the abnormal proliferation area and the matching area is used as the second proliferation velocity analysis value; The sum of the first proliferation rate analysis value and the second proliferation rate analysis value is used as a proliferation quantification factor of the abnormal proliferation area.

6. The method for evaluating the efficacy of prostate hyperplasia drugs based on ultrasound images according to claim 1, wherein: The method for obtaining the blood flow dynamic change value is: For any abnormal proliferation area, the color Doppler ultrasound image of the abnormal proliferation area is divided into multiple local areas; For any local area, a straight line is fitted according to the positions of all edge pixels in the local area by the least square method, which serves as the reference straight line of the local area; Obtaining a mean value of the goodness of fit between each edge line in the local area and the reference straight line as a first indicator of the local area; Obtaining a ratio of the number of edge pixels in the local area to the number of all pixels as a second indicator of the local area; The product of the normalized and negatively correlated result of the first index and the second index is used as the blood vessel analysis value of the local area; The mean of the vascular analysis values ​​of all local areas is used as the blood flow dynamic characteristic value of the abnormal proliferation area; The difference between the blood flow dynamic characteristic value of the abnormal proliferation area and the matching area is used as the blood flow dynamic change value of the abnormal proliferation area.

7. The method for evaluating the efficacy of prostate hyperplasia drugs based on ultrasound images according to claim 1, wherein: The method for obtaining the degree of preservation of the proliferation details is as follows: For any prostate ultrasound image, the proliferation quantification factor and the blood flow dynamic change value of each abnormal proliferation area in the prostate ultrasound image are sorted according to the order of the same abnormal proliferation areas, and the proliferation quantification factor sequence and the blood flow dynamic change value sequence are obtained respectively; The result of normalizing the Pearson correlation coefficient of the proliferation quantification factor sequence and the blood flow dynamic change value sequence is used as the degree of preservation of the proliferation details of the prostate ultrasound image.

8. The method for evaluating the efficacy of prostate hyperplasia drugs based on ultrasound images according to claim 1, wherein: The method for obtaining the modified Gaussian filter standard deviation is: For any prostate ultrasound image, the product of the degree of hyperplasia detail preservation of the prostate ultrasound image and the Gaussian filter standard deviation during the Canny edge detection process of the prostate ultrasound image is used as the modified Gaussian filter standard deviation during the Canny edge detection process of the prostate ultrasound image.

9. The method for evaluating the efficacy of prostate hyperplasia drugs based on ultrasound images according to claim 1, wherein: The method for obtaining the edge corner points is: The edge corner points in each prostate ultrasound image are obtained using the Harris corner detection algorithm.

Images