Automatic nipple positioning method for breast ultrasonic three-dimensional volume detection
By automatically detecting the nipple position in the three-dimensional volumetric data of breast ultrasound, using a nipple target detection model and removing abnormal data, the problem of time-consuming and reliance on human experience in traditional methods is solved, and efficient and accurate nipple localization is achieved.
Patent Information
- Application Number
- CN202411492628.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-24
- Publication Date
- 2024-11-22
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Traditional three-dimensional volume ultrasound methods for locating the nipple in the breast rely on manual annotation, which is time-consuming and dependent on the doctor's experience, lacking automation and accuracy.
A nipple target detection model was used to automatically detect the nipple position in the three-dimensional volume data of breast ultrasound. By filtering and removing abnormal data, the mean value was calculated to obtain the final nipple position coordinates.
It improves the efficiency and accuracy of nipple positioning, and reduces manual intervention and time consumption.
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of three-dimensional ultrasonic detection, and in particular to an automatic nipple positioning method for breast ultrasonic three-dimensional volume detection. Background Art
[0002] In order to more completely scan breast tissue during breast examination, breast tissue is usually divided into four quadrants: outer upper, inner upper, outer lower, and inner lower. With the nipple as the center, horizontal and vertical lines are added to determine the four quadrants. Breast diseases usually occur in the above four quadrants, so accurately locating the nipple position and using it to determine the interval where breast diseases occur is of great clinical significance.
[0003] Traditional nipple localization based on three-dimensional volumetric ultrasound breast data requires doctors to manually mark the position of the nipple, which is more dependent on the experience of clinicians and requires doctors to judge the coronal images frame by frame, which is very time-consuming. Summary of the invention
[0004] The object of the present invention is to provide a method for automatically locating the nipple for breast ultrasound three-dimensional volume detection, and more specifically to provide a method for automatically determining the nipple position coordinates in breast ultrasound three-dimensional volume data.
[0005] To achieve the above object, the present invention adopts the following technical solution: a nipple automatic positioning method for breast ultrasound three-dimensional volume detection, comprising the following steps: S01. First, parse out several frames of two-dimensional coronal images from the acquired three-dimensional breast ultrasound volume data, mark the frame numbers in sequence, record the frame number of the two-dimensional coronal images as N, and remove the two-dimensional coronal images without nipple features, leaving the two-dimensional coronal images with frame numbers N1~N2.
[0006] S02. Use the trained nipple target detection model to perform nipple detection on the two-dimensional coronal images with frame numbers N1~N2 obtained in step S01, and screen out the two-dimensional coronal images with obvious nipple features. The frame numbers of the two-dimensional coronal images with obvious nipple features are denoted as T1~T2, and the frame number of the two-dimensional coronal images with obvious nipple features is denoted as n.
[0007] S03, performing nipple position detection on the two-dimensional coronal image with frame numbers T1-T2 obtained in step S02, and obtaining n coordinate points, which are recorded as (x1, y1), (x2, y2), ... (x n ,y n ), and for x1, x2, ... x n The abnormal data of y1, y2, ... y nThe abnormal data is removed from (x1, y1), (x2, y2), ... (x n ,y n ) to select the ones that pass through x1, x2, ... x n and y1, y2, ...y n After the abnormal data is removed, the coordinate points where both x and y exist are recorded as (x'1, y'1), (x'2, y'2), ... (x' m , y' m ).
[0008] S04, according to the coordinate points (x'1, y'1), (x'2, y'2), ... (x' m , y' m ), and the coordinates of the nipple position (x a ,y a ), where x a =(x'1+x'2+……+x' m ) / m,y a =(y'1+y'2+……+y' m ) / m.
[0009] Specifically, in step S01, when removing the two-dimensional coronal image in which the nipple feature does not exist, the following method is specifically used: S11. Calculate the pixel value variance of the N frames of two-dimensional coronal images obtained, and arrange them in the order of frame numbers. Select the frame number where the pixel value variance starts to stabilize, and record it as N. D , the frame number is the value of N1, that is, N1=N D .
[0010] S12. Then, according to the acquisition depth H of the acquired breast ultrasound three-dimensional volume data and the number of frames N of the two-dimensional coronal plane image, the number of frames represented by the unit length E=N / H is calculated, and the actual length F of the nipple is obtained from the acquired breast ultrasound three-dimensional volume data, thereby obtaining the value of N2: N2=N1+E*F.
[0011] Specifically, in step S03, x1, x2, ... x n When removing abnormal data, all x1, x2, ... x n Calculate the Z-score to eliminate outliers that deviate from the overall trend based on the standard deviation; n When removing abnormal data, all y1, y2, ... y n Z-score calculation is performed to eliminate outliers that deviate from the overall trend based on the standard deviation.
[0012] The beneficial effect of the present invention is that a two-dimensional coronal image with obvious nipple features is determined from the two-dimensional coronal image parsed from the breast ultrasound three-dimensional volume data, and after multiple nipple position coordinates are identified therefrom, the data of abnormal results are eliminated from the nipple position coordinates to obtain the final nipple position coordinates, thereby improving the efficiency and accuracy of nipple position positioning. DETAILED DESCRIPTION
[0013] Embodiment 1, the present invention adopts the following technical scheme: a method for automatic nipple positioning for breast ultrasound three-dimensional volume detection, comprising the following steps: S01. First, parse out several frames of two-dimensional coronal images from the acquired three-dimensional breast ultrasound volume data, mark the frame numbers in sequence, record the frame number of the two-dimensional coronal images as N, and remove the two-dimensional coronal images without nipple features, leaving the two-dimensional coronal images with frame numbers N1~N2.
[0014] S02. Use the trained nipple target detection model to perform nipple detection on the two-dimensional coronal images with frame numbers N1~N2 obtained in step S01, and screen out the two-dimensional coronal images with obvious nipple features. The frame numbers of the two-dimensional coronal images with obvious nipple features are denoted as T1~T2, and the frame number of the two-dimensional coronal images with obvious nipple features is denoted as n.
[0015] S03, performing nipple position detection on the two-dimensional coronal image with frame numbers T1-T2 obtained in step S02, and obtaining n coordinate points, which are recorded as (x1, y1), (x2, y2), ... (x n ,y n ), and for x1, x2, ... x n The abnormal data of y1, y2, ... y n The abnormal data is removed from (x1, y1), (x2, y2), ... (x n ,y n ) to select the ones that pass through x1, x2, ... x n and y1, y2, ...y n After the abnormal data is removed, the coordinate points where both x and y exist are recorded as (x'1, y'1), (x'2, y'2), ... (x' m , y' m ).
[0016] S04, according to the coordinate points (x'1, y'1), (x'2, y'2), ... (x' m , y' m ), and the coordinates of the nipple position (xa ,y a ), where x a =(x'1+x'2+……+x' m ) / m,y a =(y'1+y'2+……+y' m ) / m.
[0017] Specifically, in step S01, when removing the two-dimensional coronal image in which the nipple feature does not exist, the following method is specifically used: S11. Calculate the pixel value variance of the N frames of two-dimensional coronal images obtained, and arrange them in the order of frame numbers. Select the frame number where the pixel value variance starts to stabilize, and record it as N. D , the frame number is the value of N1, that is, N1=N D .
[0018] S12. Then, according to the acquisition depth H of the acquired breast ultrasound three-dimensional volume data and the number of frames N of the two-dimensional coronal plane image, the number of frames represented by the unit length E=N / H is calculated, and the actual length F of the nipple is obtained from the acquired breast ultrasound three-dimensional volume data, thereby obtaining the value of N2: N2=N1+E*F.
[0019] Specifically, in step S03, x1, x2, ... x n When removing abnormal data, all x1, x2, ... x n Calculate the Z-score to eliminate outliers that deviate from the overall trend based on the standard deviation; n When removing abnormal data, all y1, y2, ... y n Perform Z-score calculation to eliminate outliers that deviate from the overall trend based on the standard deviation. Z-score calculation is Z-score calculation, also known as standard score calculation, which can calculate outliers that deviate from the overall trend in a group of data based on the standard deviation of the data, thereby eliminating outliers and improving the accuracy of nipple position detection.
[0020] Of course, the above are only preferred embodiments of the present invention, and are not intended to limit the scope of use of the present invention. Therefore, any equivalent changes made to the principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for automatic nipple positioning for breast ultrasound three-dimensional volume detection, It is characterized in that The steps include: S01, firstly parse a plurality of frames of two-dimensional coronal images from the acquired three-dimensional breast ultrasound volume data, mark the frame numbers in sequence, record the number of frames of the two-dimensional coronal images as N, and remove the two-dimensional coronal images without nipple features, leaving the two-dimensional coronal images with frame numbers N1-N2; S02, using the trained nipple target detection model to perform nipple detection on the two-dimensional coronal images with frame numbers N1-N2 obtained in step S01, and screen out two-dimensional coronal images with obvious nipple features, the frame numbers of the two-dimensional coronal images with obvious nipple features are denoted as T1-T2, and the number of frames of the two-dimensional coronal images with obvious nipple features is denoted as n; S03, performing nipple position detection on the two-dimensional coronal image with frame numbers T1-T2 obtained in step S02, and obtaining n coordinate points, recorded as (x 1 ,y 1 )、(x 2 ,y 2 ),……(x n ,y n ), and for x 1 、x 2 ,……x n The abnormal data of y 1 ,y 2 ,……y n The abnormal data is removed from (x 1 ,y 1 )、(x 2 ,y 2 ),……(x n ,y n ) to filter out the 1 、x 2 ,……x n and 1 ,y 2 ,……y n The coordinate point where both x and y exist after the abnormal data are removed is recorded as (x' 1 , y' 1 )、(x' 2 , y' 2 ),……(x' m , y' m ); S04, according to the coordinate point (x' 1 , y' 1 )、(x' 2 , y' 2 ),……(x' m , y' m ), and the coordinates of the nipple position (x a ,y a ), where x a =(x' 1 +x' 2 +……+x' m ) / m,y a =(y' 1 +y' 2 +……+y' m ) / m.
2. The method for automatic nipple positioning for breast ultrasound three-dimensional volume detection according to claim 1, It is characterized in that In step S01, when removing the two-dimensional coronal images in which the nipple features are not present, the following method is specifically used: S11. Calculate the pixel value variance of the N frames of two-dimensional coronal images obtained, and arrange them in the order of frame numbers. Select the frame number where the pixel value variance starts to stabilize, and record it as N. D , the frame number is the value of N1, that is, N1=N D ; S12. Then, according to the acquisition depth H of the acquired breast ultrasound three-dimensional volume data and the number of frames N of the two-dimensional coronal plane image, the number of frames represented by the unit length E=N / H is calculated, and the actual length F of the nipple is obtained from the acquired breast ultrasound three-dimensional volume data, thereby obtaining the value of N2: N2=N1+E*F.
3. The method for automatic nipple positioning for breast ultrasound three-dimensional volume detection according to claim 1, It is characterized in that In step S03, 1 、x 2 ,……x n When removing abnormal data, all x 1 、x 2 ,……x n Calculate the Z-score to remove outliers that deviate from the overall trend based on the standard deviation; 1 ,y 2 ,……y n When removing abnormal data, all y 1 ,y 2 ,……y n Z-score calculation is performed to eliminate outliers that deviate from the overall trend based on the standard deviation.
Citation Information
Patent Citations
Nipple position automatic detection method based on ABUS coronal plane image
CN110246125A
Focus detection method based on three-dimensional ultrasonic image
CN111340780A
Nipple detection method and device, computer equipment and storable medium
CN114299027A
System and method for automatically detecting nipple from 3D automatic breast ultrasound image
KR101494975B1
