Kidney stone self-service examination sheet generation method and system based on medical ultrasonic equipment

Abstract

The embodiment of the invention discloses a kidney stone self-service examination sheet generation method and system based on medical ultrasonic equipment, and the method comprises the steps: 1, obtaining an ultrasonic image collected by a user in a self-service manner through the medical ultrasonic equipment, the number of high-echo targets in the ultrasonic image, and the size data of each high-echo target, dividing each frame of ultrasonic image into a preset number of ultrasonic image sets; 2, segmenting each frame of ultrasonic image; 3, screening out a qualified ultrasonic image according to the total number of pixel points of the high-echo target area of the ultrasonic image and the average gray value; 4, judging the position of a high-echo target area; and 5, generating a check list, and storing and printing the check list. The self-service ultrasonic examination system facilitates self-service ultrasonic examination of a patient, greatly relieves the problem of insufficient hands of ultrasonic doctors in a hospital, shortens the examination time of the patient, and improves the doctor seeing efficiency of the patient.

Description

Technical Field

[0001] The present invention relates to the field of medical equipment, and in particular to a method and system for generating a self-service kidney stone examination form based on medical ultrasonic equipment. Background Art

[0002] At present, it takes about 10 minutes for doctors in the hospital's ultrasound room to examine patients. However, some hospitals are short of ultrasound doctors, so patients often have to wait in line for one or two hours before they can get their turn for an ultrasound examination. After the ultrasound examination, patients have to wait for about half an hour for the examination form to come out, and then go to the outpatient doctor to queue up for a follow-up visit. In addition, ultrasound examinations involve patient privacy. In some cases, it is inconvenient for doctors of different genders to perform the examination, which further leads to a shortage of doctors performing ultrasound examinations.

[0003] As a result, some patients who come for ultrasound examinations cannot be triaged in time and cannot receive treatment in time, which also leads to an increase in negative emotions among patients and aggravates the conflicts between doctors and patients. For example, patients with kidney stones may experience pain levels of more than level 7, and in severe cases may even experience vomiting and fainting. Ultrasound examination is a must-do examination for patients with kidney stones, and outpatient doctors need to confirm whether the patient's pain is caused by kidney stones based on the ultrasound examination form. Summary of the invention

[0004] The technical problem to be solved by the embodiments of the present invention is to provide a method and system for generating a self-service kidney stone examination form based on a medical ultrasonic device, so that patients can complete the ultrasonic examination by themselves, greatly alleviating the problem of insufficient ultrasound doctors in the hospital.

[0005] In order to solve the above technical problems, the embodiment of the present invention proposes a method for generating a self-service kidney stone examination form based on a medical ultrasonic device, comprising: Step 1: Obtain the ultrasound image collected by the user through the medical ultrasound equipment, the number of high-echo targets in the ultrasound image, and the size data of each high-echo target, pre-process each frame of the ultrasound image, and then divide the pre-processed frames of the ultrasound image into preset ultrasound image sets according to the time axis of the ultrasound image; Step 2: Input each frame of ultrasound image in each ultrasound image set into a preset segmentation model, segment each frame of ultrasound image, and segment out a high-echo target area in each frame of ultrasound image; Step 3: sorting the total number of pixels and the average grayscale value of the high echo target area of ​​each frame of the ultrasound image in each ultrasound image set; taking the ultrasound image whose total ranking of the number of pixels and the grayscale value ranking in each ultrasound image set is before the preset position as a qualified ultrasound image; Step 4: Input the qualified ultrasound image into the preset stone location detection network to determine the location of the high-echo target area; Step 5: Generate an inspection sheet based on the qualified ultrasound image, the number of high-echo targets, the location information of the high-echo target area, and the size data of each high-echo target, and save and print the inspection sheet.

[0006] Accordingly, an embodiment of the present invention further provides a self-service kidney stone checklist generation system based on a medical ultrasound device, comprising: Acquisition module: obtains the ultrasound images collected by the user through the medical ultrasound equipment, the number of high-echo targets in the ultrasound images, and the size data of each high-echo target, pre-processes each frame of the ultrasound image, and then divides the pre-processed frames of the ultrasound image into preset ultrasound image sets according to the time axis of the ultrasound image; Segmentation module: each frame of ultrasound image in each ultrasound image set is input into a preset segmentation model, each frame of ultrasound image is segmented, and a high-echo target area in each frame of ultrasound image is segmented; Screening module: sorting the total number of pixels and the average grayscale value of the high echo target area of ​​each frame of ultrasound image in each ultrasound image set respectively; taking the ultrasound image whose total ranking of the number of pixels and the grayscale value ranking in each ultrasound image set is before the preset position as a qualified ultrasound image; Position determination module: inputs qualified ultrasound images into a preset stone position detection network to determine the position of the high-echo target area; Generation module: Generates an inspection sheet according to qualified ultrasound images, the number of high-echo targets, the location information of the high-echo target area, and the size data of each high-echo target, and saves and prints the inspection sheet.

[0007] The beneficial effects of the present invention are as follows: the present invention can be directly applied to existing medical ultrasound equipment, making it convenient for patients to conduct ultrasound examinations by themselves, replacing ultrasound doctors in screening qualified ultrasound images and making examination sheets, which greatly alleviates the problem of insufficient ultrasound doctors in hospitals, shortens the examination time of patients, and improves the efficiency of patients' medical treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 It is a flowchart of a method for generating a self-service kidney stone examination form based on a medical ultrasonic device according to an embodiment of the present invention.

[0009] Figure 2 It is a schematic diagram of the structure of the stone position detection network according to an embodiment of the present invention. DETAILED DESCRIPTION

[0010] It should be noted that, in the absence of conflict, the embodiments in the present application and the features in the embodiments may be combined with each other. The present invention is further described in detail below in conjunction with the drawings and specific embodiments.

[0011] In the embodiments of the present invention, if there are directional indications (such as up, down, left, right, front, back, etc.), they are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0012] In addition, in the present invention, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" or "second" may explicitly or implicitly include at least one of the features.

[0013] Please refer to Figure 1 The method for generating a self-service kidney stone checklist based on medical ultrasonic equipment in an embodiment of the present invention includes steps 1 to 5.

[0014] Step 1: Obtain the ultrasound images collected by the user through the medical ultrasound equipment, the number of high-echo targets in the ultrasound images, and the size data of each high-echo target, pre-process each frame of the ultrasound image, and then divide the pre-processed frames of the ultrasound image into preset ultrasound image sets according to the time axis of the ultrasound image. The preprocessing in this field is generally to convert to a uniform size and remove noise and artifacts in each frame of the ultrasound image. The present invention preferably has 4 ultrasound image sets. The present invention is divided into 4 ultrasound image sets according to the time axis, and then high-quality qualified ultrasound images are selected from the 4 ultrasound image sets respectively, so that the selected qualified ultrasound images can contain multiple angles, which is more convenient for outpatient doctors to diagnose according to the examination list.

[0015] High-echo targets are stones. The reflection characteristics of stones under ultrasound are significantly different from those of surrounding tissues. Therefore, stones usually appear as high-echo points or lumps on ultrasound images, with different sizes and shapes. Ultrasound images, the number of high-echo targets, and the size data of high-echo targets can all be collected from existing medical ultrasound equipment. The work of ultrasound doctors is mainly to screen qualified ultrasound images and determine the location of stones. Ultrasound doctors do not make diagnoses. The diagnosis is performed by outpatient doctors based on the examination list.

[0016] The user of the embodiment of the present invention is preferably the patient himself, and may also be a nurse in the hospital to alleviate the problem of insufficient ultrasound doctors in the hospital.

[0017] Step 2: Input each frame of ultrasound image in each ultrasound image set into a preset segmentation model, segment each frame of ultrasound image, and segment out a high-echo target area in each frame of ultrasound image.

[0018] As an implementation mode, the preset segmentation model is a SAM model, including an image encoder, a hint encoder and a mask decoder, and the preset segmentation model is trained according to the following steps: Several ultrasound images containing stones are obtained and the stone areas in the ultrasound images are annotated to construct a stone area data set; and a preset segmentation model is trained using the data in the stone area data set.

[0019] The present invention adopts the SAM model to identify and outline the stone area in the ultrasound image, which can help doctors quickly and accurately identify and locate the stone area, and facilitate the stone position detection network to more accurately determine the position of the stone in the abdomen.

[0020] Step 3: Sort the total number of pixels and the average grayscale value of the high echo target area of ​​each frame of ultrasound image in each ultrasound image set. The ultrasound image whose total ranking of the number of pixels and the grayscale value ranking in each ultrasound image set is before the preset position is regarded as a qualified ultrasound image.

[0021] The higher the ranking of the total number of pixels, the larger the display area of ​​the stones is at the angle of the frame of ultrasound image, and the more convenient it is for outpatient doctors to view. Since stones usually appear as high-echo points or lumps on ultrasound images, the higher the average grayscale value, the more obvious the area in the ultrasound image is, and the more convenient it is for outpatient doctors to view. Therefore, the present invention selects the ultrasound image with the top total ranking as a qualified ultrasound image, which is also in line with the ultrasound department's habit of screening high-quality ultrasound images. The total ranking of the present invention = (ranking of the number of pixels + ranking of grayscale value) / 2. The preset position is set according to the number of qualified ultrasound images actually required by the examination form. For example, if the examination form only needs 4 images, that is, only 1 image needs to be selected from a set of ultrasound images, then the preset position can be set to 2.

[0022] Step 4: Input the qualified ultrasound image into the preset stone location detection network to determine the location of the high-echo target area.

[0023] The present invention can be directly applied to medical ultrasound equipment, and can directly process ultrasound images collected by users through medical ultrasound equipment, the number of high-echo targets in the ultrasound images, and the size data of each high-echo target. It can automatically complete the image screening and position determination work of ultrasound doctors, and greatly improve the inspection efficiency.

[0024] As an implementation method, the preset stone location detection network is trained according to the following steps: Acquire a number of ultrasound images containing stones in different positions and annotate the positions of the stones in the ultrasound images to construct a stone position data set; A preset stone position detection network is built based on a deep learning framework, and the preset stone position detection network is trained using data in the stone position dataset.

[0025] The preset stone location detection network adopts the YOLOV4 detection network architecture, including the input end, BackBone benchmark network, Neck network and Head network. The BackBone benchmark network uses the Mish activation function instead of the original ReLU activation function and adds the Dropblock module.

[0026] YOLOV4 detection network architecture Figure 2 As shown, the present invention adopts the YOLOV4 detection network architecture to quickly and accurately determine whether the stones are in the renal pelvis, renal calyx, ureter (anterior, middle, and posterior segments) or bladder, etc., and then can accurately generate an examination list, replace the work of ultrasound doctors, and assist outpatient doctors in diagnosis.

[0027] Existing deep learning algorithms are mainly used in the medical field for the diagnosis of diseases, but the existing technology cannot improve the accuracy of disease diagnosis. The present invention uses deep learning algorithms for location judgment. When the training data set is large enough, its accuracy can be infinitely close to 100%, which can fully undertake inspection work.

[0028] Step 5: Generate an inspection sheet according to the qualified ultrasound image, the number of high echo targets, the location information of the high echo target area, and the size data of each high echo target, and save and print the inspection sheet. The format of the inspection sheet can adopt an existing format.

[0029] As an implementation mode, before step 1, an instruction step is also included: instructing the user to self-collect ultrasound images through the medical ultrasound device by voice, text, image and / or animation. The instruction step of the present invention instructs the user to self-operate the medical ultrasound device, for example, by displaying the operation process such as how to align the ultrasound probe of the medical ultrasound device with the abdomen for scanning by voice and text; or by instructing the user to self-operate the medical ultrasound device by image, animation, or video, for example, by playing the use process such as how to align the ultrasound probe of the medical ultrasound device with the abdomen for scanning by image, animation, or video.

[0030] The self-help kidney stone checklist generation system based on medical ultrasound equipment in the embodiment of the present invention includes: Acquisition module: obtains the ultrasound images collected by the user through the medical ultrasound equipment, the number of high-echo targets in the ultrasound images, and the size data of each high-echo target, pre-processes each frame of the ultrasound image, and then divides the pre-processed frames of the ultrasound image into preset ultrasound image sets according to the time axis of the ultrasound image; Segmentation module: each frame of ultrasound image in each ultrasound image set is input into a preset segmentation model, each frame of ultrasound image is segmented, and a high-echo target area in each frame of ultrasound image is segmented; Screening module: sorting the total number of pixels and the average grayscale value of the high echo target area of ​​each frame of ultrasound image in each ultrasound image set respectively; taking the ultrasound image whose total ranking of the number of pixels and the grayscale value ranking in each ultrasound image set is before the preset position as a qualified ultrasound image; Position determination module: inputs qualified ultrasound images into a preset stone position detection network to determine the position of the high-echo target area; Generation module: Generates an inspection sheet according to qualified ultrasound images, the number of high-echo targets, the location information of the high-echo target area, and the size data of each high-echo target, and saves and prints the inspection sheet.

[0031] As an implementation method, the preset stone location detection network is trained according to the following steps: Acquire a number of ultrasound images containing stones in different positions and annotate the positions of the stones in the ultrasound images to construct a stone position data set; A preset stone position detection network is built based on a deep learning framework, and the preset stone position detection network is trained using data in the stone position dataset.

[0032] As an implementation method, the preset stone position detection network adopts the YOLOV4 detection network architecture, including an input end, a BackBone reference network, a Neck network, and a Head network. The BackBone benchmark network uses the Mish activation function instead of the original ReLU activation function and adds the Dropblock module.

[0033] As an implementation mode, the preset segmentation model is a SAM model, including an image encoder, a hint encoder and a mask decoder, and the preset segmentation model is trained according to the following steps: Several ultrasound images containing stones are obtained and the stone areas in the ultrasound images are annotated to construct a stone area data set; and a preset segmentation model is trained using the data in the stone area data set.

[0034] As an embodiment, the self-service kidney stone checklist generation system based on medical ultrasound equipment also includes an instruction module: instructing the user to self-collect ultrasound images through the medical ultrasound equipment through voice, text, images and / or animation.

[0035] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for generating a self-service checklist for kidney stones based on medical ultrasound equipment. It is characterized in that include: Step 1: Obtain the ultrasound image collected by the user through the medical ultrasound equipment, the number of high-echo targets in the ultrasound image, and the size data of each high-echo target, pre-process each frame of the ultrasound image, and then divide the pre-processed frames of the ultrasound image into preset ultrasound image sets according to the time axis of the ultrasound image; Step 2: Input each frame of ultrasound image in each ultrasound image set into a preset segmentation model, segment each frame of ultrasound image, and segment out a high-echo target area in each frame of ultrasound image; Step 3: sorting the total number of pixels and the average grayscale value of the high echo target area of ​​each frame of the ultrasound image in each ultrasound image set; taking the ultrasound image whose total ranking of the number of pixels and the grayscale value ranking in each ultrasound image set is before the preset position as a qualified ultrasound image; Step 4: Input the qualified ultrasound image into the preset stone location detection network to determine the location of the high-echo target area; Step 5: Generate an inspection sheet based on the qualified ultrasound image, the number of high-echo targets, the location information of the high-echo target area, and the size data of each high-echo target, and save and print the inspection sheet.

2. The method for generating a self-help checklist for kidney stones based on medical ultrasonic equipment according to claim 1, It is characterized in that The preset stone location detection network is trained according to the following steps: Acquire a number of ultrasound images containing stones in different positions and annotate the positions of the stones in the ultrasound images to construct a stone position data set; A preset stone position detection network is built based on a deep learning framework, and the preset stone position detection network is trained using data in the stone position dataset.

3. The method for generating a self-help checklist for kidney stones based on medical ultrasound equipment as claimed in claim 2, It is characterized in that The preset stone location detection network adopts the YOLOV4 detection network architecture, including the input end, BackBone benchmark network, Neck network and Head network. The BackBone benchmark network uses the Mish activation function instead of the original ReLU activation function and adds the Dropblock module.

4. The method for generating a self-help checklist for kidney stones based on medical ultrasonic equipment according to claim 1, It is characterized in that The preset segmentation model is the SAM model, which includes an image encoder, a hint encoder, and a mask decoder. The preset segmentation model is trained according to the following steps: Several ultrasound images containing stones are obtained and the stone areas in the ultrasound images are annotated to construct a stone area data set; and a preset segmentation model is trained using the data in the stone area data set.

5. The method for generating a self-help checklist for kidney stones based on medical ultrasonic equipment as claimed in claim 1, It is characterized in that Also include the following steps before step 1: Instruct users to collect ultrasound images by themselves through medical ultrasound equipment through voice, text, images and / or animation.

6. A self-service kidney stone checklist generation system based on medical ultrasound equipment, It is characterized in that include: Acquisition module: obtains the ultrasound images collected by the user through the medical ultrasound equipment, the number of high-echo targets in the ultrasound images, and the size data of each high-echo target, pre-processes each frame of the ultrasound image, and then divides the pre-processed frames of the ultrasound image into preset ultrasound image sets according to the time axis of the ultrasound image; Segmentation module: each frame of ultrasound image in each ultrasound image set is input into a preset segmentation model, each frame of ultrasound image is segmented, and a high-echo target area in each frame of ultrasound image is segmented; Screening module: sorting the total number of pixels and the average grayscale value of the high echo target area of ​​each frame of ultrasound image in each ultrasound image set respectively; taking the ultrasound image whose total ranking of the number of pixels and the grayscale value ranking in each ultrasound image set is before the preset position as a qualified ultrasound image; Position determination module: inputs qualified ultrasound images into a preset stone position detection network to determine the position of the high-echo target area; Generation module: Generates an inspection sheet according to qualified ultrasound images, the number of high-echo targets, the location information of the high-echo target area, and the size data of each high-echo target, and saves and prints the inspection sheet.

7. The self-help kidney stone checklist generation system based on medical ultrasound equipment as claimed in claim 6, It is characterized in that The preset stone location detection network is trained according to the following steps: Acquire a number of ultrasound images containing stones in different positions and annotate the positions of the stones in the ultrasound images to construct a stone position data set; A preset stone position detection network is built based on a deep learning framework, and the preset stone position detection network is trained using data in the stone position dataset.

8. The self-help checklist generation system for kidney stones based on medical ultrasound equipment as claimed in claim 7, It is characterized in that The preset stone location detection network adopts the YOLOV4 detection network architecture, including the input end, BackBone benchmark network, Neck network and Head network. The BackBone benchmark network uses the Mish activation function instead of the original ReLU activation function and adds the Dropblock module.

9. The self-help kidney stone checklist generation system based on medical ultrasound equipment as claimed in claim 6, It is characterized in that The preset segmentation model is the SAM model, which includes an image encoder, a hint encoder, and a mask decoder. The preset segmentation model is trained according to the following steps: Several ultrasound images containing stones are obtained and the stone areas in the ultrasound images are annotated to construct a stone area data set; and a preset segmentation model is trained using the data in the stone area data set.

10. The self-help checklist generation system for kidney stones based on medical ultrasound equipment as claimed in claim 6, It is characterized in that It also includes an instruction module: which instructs the user to collect ultrasound images by self-service through medical ultrasound equipment through voice, text, image and / or animation.

Images