A method and system for identifying brittle tilapia based on M-mode ultrasound

Abstract

The present invention belongs to the technical field of image processing and ultrasonic inspection, and proposes a method and system for identifying brittle tilapia based on M-mode ultrasound, which is specifically as follows: the tilapia is subjected to inspection treatment; the ultrasonic image of the tilapia is collected by using M-mode ultrasound, and a diastolic image is obtained; the diastolic image is preprocessed and venous pixels are identified, and diastolic muscle energy analysis is performed based on the diastolic image and its venous pixels to obtain a muscle energy evaluation value; finally, the state of the brittle tilapia is identified based on the muscle energy evaluation value. The incompatibility between the heart function and the brittle process in the fish body is accurately and quickly identified based on the ultrasonic image, and the individual differences in the compensatory ability or tolerance of the cardiovascular system of the fish body to the brittle feed are effectively quantified, providing a mathematical basis for the brittle process of tilapia breeding, thereby improving the identification effect of tilapia that is not adapted to brittle breeding, and reducing the risk of tilapia death or low activity due to insufficient tolerance.

Description

Technical Field

[0001] The invention belongs to the technical field of image processing and ultrasonic inspection, and particularly relates to a method and system for identifying brittle tilapia based on M-mode ultrasound. Background Art

[0002] In the process of brittle breeding of tilapia, the brittle-fed tilapia often die or lack of activity in the late stage of brittle breeding, which seriously reduces the production efficiency of the farm. The death of brittle tilapia is directly related to the feeding method of brittle feed. Brittle feeding can easily cause heart dysfunction in tilapia. The compensatory ability or tolerance of the cardiovascular system of different fish to brittle feed varies from person to person, so the degree of heart dysfunction varies from individual to individual. However, heart dysfunction cannot be simply monitored by blood flow velocity. This is because artificially broiled embrittled fish are embrittled by feeding feed containing broad beans or broad bean extracts. While the muscles are embrittled, on the one hand, the hemolytic reaction caused by broad beans destroys a large number of blood cells, causing the body to compensatorily produce more red blood cells and white blood cells, causing blood viscosity to increase; on the other hand, while the muscles of tilapia are embrittled, the vascular endothelial cells are simultaneously embrittled and hardened, reducing the inner diameter of the blood vessels and resulting in a decrease in vascular shear force. These factors together lead to a decrease in blood flow velocity; however, in the later stages of embrittlement breeding, some fish can no longer compensate for the hemolytic reaction caused by broad beans, resulting in a large number of blood cells being destroyed and the number of intact blood cells dropping sharply, resulting in a decrease in blood viscosity. At this time, the blood flow velocity increases instead. Therefore, it is impossible to identify individual tilapia with heart dysfunction based on the blood flow velocity. The water and mucus on the surface of the fish are good coupling agents. It is feasible and convenient to perform ultrasonic testing on the quality of tilapia in water. The hemolytic reaction of broad beans itself causes insufficient blood supply to the myocardium, and the blood flow per unit time is reduced due to the reduction in blood flow velocity, which causes hypoxia in the body and leads to heart dysfunction. The myocardium is damaged due to ischemia and hypoxia, resulting in reduced myocardial contractility. Therefore, the degree of heart dysfunction of brittle tilapia can be effectively quantified by M-mode ultrasound technology. Based on this, the present invention proposes a method and system for identifying brittle tilapia based on M-mode ultrasound. Summary of the invention

[0003] The purpose of the present invention is to propose a method and system for identifying brittle tilapia based on M-mode ultrasound to solve one or more technical problems existing in the prior art and at least provide a beneficial choice or create conditions.

[0004] In order to achieve the above object, according to one aspect of the present invention, a method for identifying brittle tilapia based on M-mode ultrasound is provided, and the method comprises the following steps:

[0005] S100, placing the tilapia under inspection;

[0006] S200, using M-mode ultrasound to collect ultrasound images of tilapia and obtain diastolic images;

[0007] S300, preprocessing the diastolic image and identifying venous pixels;

[0008] S400, performing diastolic muscle energy analysis according to the diastolic image and its venous pixels to obtain a muscle energy evaluation value;

[0009] S500, identifying the state of the brittle tilapia according to the muscle energy evaluation value.

[0010] Further, in step S100, the method of subjecting the tilapia to inspection treatment is: transferring the tilapia to a water tank filled with oxygenated fresh water and anesthetics, wherein the anesthetics include any one of 20% benzocaine ethanol solution 1-2 ml / l0L water, chlorobutanol or 30% benzocaine ethanol solution 1 ml / L water; when the tilapia body does not move but the gill cover continues to move, laying the tilapia on its side until the inspection treatment is completed.

[0011] Furthermore, in step S200, the method of collecting diastolic images of tilapia using M-mode ultrasound is: placing an M-mode ultrasound probe between the pectoral fins of the tilapia, collecting a number of ultrasound images along the ventral midline, and recording the images in which the heart is in the end diastolic phase as diastolic images.

[0012] Furthermore, in step S300, the method for preprocessing the diastolic image and identifying venous pixels is: performing median filtering denoising on the diastolic image, and enhancing the contrast of the diastolic image in combination with a histogram equalization algorithm; identifying the vascular area in the diastolic image through a pre-trained UNet convolutional neural network, and marking the pixels in the vascular area as venous pixels.

[0013] Further, in step S400, the diastolic muscle energy analysis is performed based on the diastolic image and its venous pixels, and the method for obtaining the muscle energy evaluation value is: the blood vessel region of the processed image is divided into regions by a superpixel segmentation algorithm to obtain a domain, and the domain in which all pixels in the region are venous pixels is recorded as a quasi-venous domain, and the remaining domains adjacent to the quasi-venous domain are recorded as boundary domains;

[0014] For any boundary domain, the nearest quasi-vein domain is taken as its inverse mapping domain, and the grayscale value of the centroid position of the inverse mapping domain is the inverse mapping grayscale; the proportion of vein pixels in the boundary domain is recorded as the vein recognition ratio, and the difference between the inverse mapping grayscale and the average value of the corresponding grayscale values ​​of each non-vein pixel is calculated, and the ratio of the difference obtained and the vein recognition ratio is the inverse mapping attenuation gradient Dcg;

[0015] Initialize the muscle energy search domain corresponding to the current boundary domain as the current boundary domain, search for each boundary domain adjacent to the muscle energy search domain and merge them into the muscle energy search domain, iterate the search and merge until no boundary domain is merged into the muscle energy search domain;

[0016] All boundary domains obtained by the search are recorded as search domains, the distance between the search domain and the current boundary domain is recorded as search distance dsp, and each search domain is written into a sequence in ascending order of distance and recorded as search domain sequence;

[0017] The iteration number when any search domain is merged into the muscle energy search domain is recorded as TN, and the total number of iterations is recorded as the search degree max.TN. The attenuation gradient contraction condition is: Dcg p ≥(TN / max.TN)Dcg p +(1-TN / max.TN)Dcg st e -dsp ; Among them, Dcg p and Dcg st Represent the inverse mapping attenuation gradient of the current boundary analysis domain and the search domain respectively, max.TN refers to; e is a natural constant;

[0018] Starting from the first element of the search domain sequence, search for the first element that does not meet the attenuation gradient contraction condition as the muscle energy boundary domain; muscle energy boundary domain, the gradient change rate of the current boundary domain is Rvt_Dcg = (Dcg p -Dcg st ) / Dcg st ×Dsp;

[0019] The muscle energy evaluation value is calculated based on the gradient change rate: HTfrK=mean{rp(max.TN i1 ) Rvt_Dcg i1}, where i1 is the serial number of the boundary domain, and all boundary domains correspond to max.TN in the search degree i1 The frequency of occurrence of the corresponding value is rp(max.TN i1 ), max.TN i1 and Rvt_Dcg i1 are the search degree and gradient change rate of the i1th boundary domain respectively.

[0020] When the diastolic image is analyzed by the attenuation gradient contraction condition, if the inverse mapping attenuation gradient is unevenly distributed in the image, resulting in a large difference in the distribution density of the boundary analysis domains of different search degrees, the muscle energy evaluation value data finally obtained will be suspected of being distorted. This is because this method is highly dependent on the accuracy of the recognition model of the venous pixels, resulting in the calculated fragile muscle energy evaluation value having a chance defect caused by the model quality. At present, there is no feasible technology to enhance the chance defect problem caused by this method. In order to eliminate this influence, the present invention proposes a more preferred solution:

[0021] Preferably, in step S400, the method of performing diastolic muscle energy analysis based on the diastolic image and its venous pixels to obtain the muscle energy evaluation value is:

[0022] The diastolic image is divided into several regions by a superpixel segmentation algorithm and recorded as domains, wherein the domain with venous pixels is recorded as the venous domain of the diastolic image, and the remaining regions are recorded as the venous domain of the diastolic image;

[0023] The domains pointed to by each pixel in the eight neighborhoods of a pixel are recorded as a set of TPRZ. When the length of the TPRZ of the pixel is ≥ 2, the pixel is defined as a boundary point. Any domain is taken as the current domain, and the TPRZs of each boundary point in the current domain are formed into a union. In this union, each domain except the current domain is taken as an adjacent domain. The line segment between the two pixels farthest apart in the domain is taken as the domain long axis, and the midpoint of the domain long axis is taken as the domain center. The average value of the grayscale value of each pixel in the domain is recorded as the domain fixed value, the average value of the grayscale value of each venation pixel is calculated and recorded as the venation fixed value, and the average value of the grayscale value of the remaining pixels is recorded as the off-venation fixed value.

[0024] Take any context analysis domain as the current context analysis domain, obtain the domain values ​​of each adjacent analysis domain of the current context analysis domain, and take the adjacent analysis domains with the maximum domain value and the minimum domain value as the positive feedback neighborhood and the negative feedback neighborhood respectively; obtain the distance value between the domain center of the positive feedback neighborhood and each pixel in the current context analysis domain, and take the pixel with the maximum distance value as the positive feedback edge point, obtain the distance value between the domain center of the negative feedback neighborhood and each pixel in the current context analysis domain, and take the pixel with the maximum distance value as the negative feedback edge point, take the negative feedback edge point as the center of the circle, and draw a circle with the distance between the negative feedback edge point and the domain center of the current context analysis domain as the radius and record it as the negative feedback domain, take the positive feedback edge point as the center of the circle, and draw a circle with the distance between the positive feedback edge point and the domain center of the current context analysis domain as the radius and record it as the positive feedback domain, then the feedback domain of the current context analysis domain is the intersection of the negative feedback domain and the positive feedback domain, and the feedback domain is used to calculate the fragile muscle energy assessment value.

[0025] The main purpose of setting the feedback domain is to exclude those areas that are not affected by cardiac dysfunction, thereby reducing the interference of non-pathological signals on the analysis results. For example, changes in blood flow velocity are not always directly related to cardiac dysfunction, and the setting of the feedback domain is derived from the guidance of vein pixels located by UNet convolutional neural network training, so the quantitative results of the areas that further reflect cardiac dysfunction will be more representative.

[0026] Furthermore, the method for calculating the fragile muscle energy evaluation value using the feedback domain is as follows: define the proportion of pixels in all pixels of any domain that appear in the feedback domain as the infiltration value; set a ratio value as the infiltration ratio EZRt, and its value range EZRt∈[20%,50%]; when the infiltration value of the domain is greater than the infiltration ratio, or when the domain center of the domain is in the feedback domain, the domain is defined as being in the feedback domain; the off-pulse domain and the venous domain in the feedback domain are respectively regarded as the off-pulse bundle domain and the venous bundle domain;

[0027] Regarding the setting of the infiltration ratio, the principle is to increase the specificity of the test to reduce misjudgment. Therefore, when there are large individual pathological differences in the tested fish population, or some individuals have weak pathological reactions to crispy feed, a lower infiltration ratio can be set to help capture minor abnormalities in these individuals and avoid missed detections. However, too high an infiltration ratio may lead to missed detection of some minor abnormalities or early symptoms.

[0028] The feedback domain is screened: the first screening condition is: if the number of vascular bundle domains in the current vascular analysis domain is less than 2, the current vascular analysis domain and its corresponding feedback domain are eliminated; the second screening condition is: if the proportion of pixels belonging to the vascular analysis domain in the current vascular analysis domain is less than the infiltration ratio, the current vascular analysis domain and its corresponding feedback domain are eliminated;

[0029] The difference between the average value of the domain setting of each choroid bundle domain under the feedback domain and the choroid setting is recorded as the obstruction level DrbL; the fragile muscle energy assessment value HTfrK of the diastolic image is obtained by quantifying the obstruction level, and its calculation method is:

[0030] ;

[0031] Where j1 is the multiplicative variable, NMLY is the number of context domains, and DrbL j1 is the level of the j1th context domain, LBP j1 is the sum of the grayscale values ​​of the same pixels in the j1th feedback domain and the corresponding context analysis domain, XAP j1 is the sum of the grayscale values ​​of the same pixels in the j1th feedback domain and the adjacent domain of the corresponding context domain, LTP j1 is the number of pixels on the long axis of the j1th context analysis domain, and exp() is an exponential function with the natural number e as the base.

[0032] Beneficial effects: The brittle muscle energy assessment value can accurately identify the incompatibility between the heart function and the brittle process in the fish body based on ultrasound images, and effectively quantify the individual differences in the fish's cardiovascular system's compensatory ability or tolerance to brittle feed, providing a mathematical basis for the brittle process of tilapia breeding, thereby improving the identification effect of tilapia that are not adapted to brittle feeding.

[0033] Further, in step S500, the method of identifying the state of brittle tilapia according to the muscle energy evaluation value is: when the brittle muscle energy evaluation value obtained by the individual is higher than that of the individuals under the same breeding environment, it is considered that the individual has a high risk of heart dysfunction during the brittle breeding process, otherwise it is identified that the risk of brittle heart dysfunction is low.

[0034] Preferably, all undefined variables in the present invention, if not clearly defined, can be manually set thresholds.

[0035] The present invention also provides an M-mode ultrasound-based brittle tilapia identification system, the M-mode ultrasound-based brittle tilapia identification system comprising: a processor, a memory, and a computer program stored in the memory and executable on the processor, the processor implementing the steps in the M-mode ultrasound-based brittle tilapia identification method when executing the computer program, the M-mode ultrasound-based brittle tilapia identification system can be run in computing devices such as desktop computers, notebook computers, PDAs, and cloud data centers, and the executable system may include, but is not limited to, a processor, a memory, and a server cluster, and the processor executes the computer program in the following system units:

[0036] A fish body preparation unit for testing, used for preparing tilapia for testing;

[0037] An ultrasonic data acquisition unit, used for acquiring ultrasonic images of tilapia using M-mode ultrasound and obtaining diastolic images;

[0038] An image preprocessing unit, used for preprocessing the diastolic image and identifying venous pixels;

[0039] A brittleness risk assessment unit is used to perform diastolic muscle energy analysis based on the diastolic image and its venous pixels to obtain a muscle energy assessment value;

[0040] The risk identification unit is used to identify the status of brittle tilapia based on the muscle energy assessment value.

[0041] The beneficial effects of the present invention are as follows: the present invention provides a method and system for quickly detecting brittle fish meat based on M-type ultrasound images, which can accurately and quickly identify the incompatibility between the heart function and the brittle process in the fish body according to the ultrasound image, effectively quantify the individual differences in the compensatory ability or tolerance of the cardiovascular system of the fish body to the brittle feed, and provide a mathematical basis for the brittle process of tilapia breeding, thereby improving the identification effect of tilapia that is not suitable for brittle breeding, reducing the risk of tilapia death or low activity due to insufficient tolerance, and thus ensuring the production efficiency and resource utilization of the farm. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The above and other features of the present invention will become more obvious by describing in detail the embodiments shown in the accompanying drawings. The same reference numerals in the accompanying drawings of the present invention represent the same or similar elements. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other accompanying drawings can be obtained based on these accompanying drawings without creative work. In the accompanying drawings:

[0043] Figure 1 Shown is a flow chart of a method for identifying brittle tilapia based on M-mode ultrasound;

[0044] Figure 2 Shown is a structural diagram of an embrittled tilapia identification system based on M-mode ultrasound. DETAILED DESCRIPTION

[0045] The following will be combined with the embodiments and drawings to clearly and completely describe the concept, specific structure and technical effects of the present invention, so as to fully understand the purpose, scheme and effect of the present invention. It should be noted that the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0046] like Figure 1 The figure shows a flow chart of a method for identifying brittle tilapia based on M-mode ultrasound. Figure 1 A method for identifying brittle tilapia based on M-mode ultrasound according to an embodiment of the present invention is described, and the method comprises the following steps:

[0047] S100, placing the tilapia under inspection;

[0048] S200, using M-mode ultrasound to collect ultrasound images of tilapia and obtain diastolic images;

[0049] S300, preprocessing the diastolic image and identifying venous pixels;

[0050] S400, performing diastolic muscle energy analysis according to the diastolic image and its venous pixels to obtain a muscle energy evaluation value;

[0051] S500, identifying the state of the brittle tilapia according to the muscle energy evaluation value.

[0052] Further, in step S100, the method of subjecting the tilapia to inspection treatment is: transferring the tilapia to a water tank filled with oxygenated fresh water and anesthetics, wherein the anesthetics include any one of 20% benzocaine ethanol solution 1-2 ml / l0L water, chlorobutanol or 30% benzocaine ethanol solution 1 ml / L water; when the tilapia body does not move but the gill cover continues to move, laying the tilapia on its side until the inspection treatment is completed.

[0053] Furthermore, in step S200, the method of collecting diastolic images of tilapia using M-mode ultrasound is: placing an M-mode ultrasound probe between the pectoral fins of the tilapia, collecting a number of ultrasound images along the ventral midline, and recording the images in which the heart is in the end diastolic phase as diastolic images.

[0054] Unlike mammals, the ventricles of fish are located in the sagittal midline and ventral to the atria and venous sinuses. Therefore, the best scanning plane for fish echocardiography is the ventral midline, which can obtain longitudinal images including all chambers.

[0055] Furthermore, in step S300, the method for preprocessing the diastolic image and identifying venous pixels is: performing median filtering denoising on the diastolic image, and enhancing the contrast of the diastolic image in combination with a histogram equalization algorithm; identifying the vascular area in the diastolic image through a pre-trained UNet convolutional neural network, and marking the pixels in the vascular area as venous pixels.

[0056] Further, in step S400, the diastolic muscle energy analysis is performed based on the diastolic image and its venous pixels, and the method for obtaining the muscle energy evaluation value is: the vascular region of the processed image is divided into regions by a superpixel segmentation algorithm to obtain a domain, the domain in which all pixels in the region are venous pixels is recorded as a quasi-venous domain, the remaining domains adjacent to the quasi-venous domain are recorded as boundary domains, and the pixel points in the eight neighborhoods of each pixel in the boundary domain that have non-venous pixels are recorded as boundary points; the non-vascular region is grid-divided so that the standard grid area is the same as the average area of ​​each quasi-venous domain;

[0057] For any boundary domain, the nearest quasi-vein domain is taken as its inverse mapping domain, and the grayscale value of the centroid position of the inverse mapping domain is the inverse mapping grayscale; the proportion of vein pixels in the boundary domain is recorded as the vein recognition ratio, and the difference between the inverse mapping grayscale and the average value of the corresponding grayscale values ​​of each non-vein pixel is calculated, and the ratio of the difference obtained and the vein recognition ratio is the inverse mapping attenuation gradient Dcg;

[0058] Initialize the muscle energy search domain corresponding to the current boundary domain as the current boundary domain, search for each boundary domain adjacent to the muscle energy search domain and merge them into the muscle energy search domain, iterate the search and merge until no boundary domain is merged into the muscle energy search domain;

[0059] All boundary domains obtained by the search are recorded as search domains, the distance between the search domain and the current boundary domain is recorded as search distance dsp, and each search domain is written into a sequence in ascending order of distance and recorded as search domain sequence;

[0060] The iteration number when any search domain is merged into the muscle energy search domain is recorded as TN, and the total number of iterations is recorded as the search degree max.TN. The attenuation gradient contraction condition is: Dcg p ≥(TN / max.TN)Dcg p +(1-TN / max.TN)Dcg st e -dsp ; Among them, Dcg p and Dcg st Represent the inverse mapping attenuation gradient of the current boundary analysis domain and the search domain respectively, max.TN refers to; e is a natural constant;

[0061] Starting from the first element of the search domain sequence, search for the first element that does not meet the attenuation gradient contraction condition as the muscle energy boundary domain; muscle energy boundary domain, the gradient change rate of the current boundary domain is Rvt_Dcg = (Dcg p -Dcg st ) / Dcg st ×Dsp;

[0062] The muscle energy evaluation value is calculated based on the gradient change rate: HTfrK=mean{rp(max.TN i1 ) Rvt_Dcg i1}, where i1 is the serial number of the boundary domain, and all boundary domains correspond to max.TN in the search degree i1 The frequency of occurrence of the corresponding value is rp(max.TN i1 ), max.TN i1 and Rvt_Dcg i1 are the search degree and gradient change rate of the i1th boundary domain respectively.

[0063] The distance between any two regions is the distance between the center points of the two regions; when there are overlapping pixels on the boundaries of any two regions, the two regions are defined as adjacent; the iteration number refers to the number of iterations in which it is added to the muscle energy search domain. The centroid is the weighted average position of the grayscale values ​​of all pixels in the region;

[0064] Preferably, in step S400, the diastolic muscle energy analysis is performed based on the diastolic image and its venous pixels, and the method for obtaining the muscle energy evaluation value is: the diastolic image is divided into a plurality of regions by a superpixel segmentation algorithm and recorded as analysis domains, the analysis domain in which the venous pixels exist is recorded as the venous analysis domain of the diastolic image, and the remaining regions are recorded as the venous separation analysis domain of the diastolic image;

[0065] Each pixel in the diastolic image points to a unique domain, that is, the domain to which the pixel belongs;

[0066] The domains pointed to by each pixel in the eight neighborhoods of a pixel are recorded as a set of TPRZ. When the length of the TPRZ of the pixel is ≥ 2, the pixel is defined as a boundary point. Any domain is taken as the current domain, and the TPRZs of each boundary point in the current domain are formed into a union. In this union, each domain except the current domain is taken as an adjacent domain. The line segment between the two pixels farthest apart in the domain is taken as the domain long axis, and the midpoint of the domain long axis is taken as the domain center. The average value of the grayscale value of each pixel in the domain is recorded as the domain fixed value, the average value of the grayscale value of each venation pixel is calculated and recorded as the venation fixed value, and the average value of the grayscale value of the remaining pixels is recorded as the off-venation fixed value.

[0067] If there are pixels of different resolution domains on any straight line in the diastolic image, it is defined that the straight line passes through the resolution domain to which these pixels belong;

[0068] Take any context analysis domain as the current context analysis domain, obtain the domain values ​​of each adjacent analysis domain of the current context analysis domain, and take the adjacent analysis domains with the maximum domain value and the minimum domain value as the positive feedback neighborhood and the negative feedback neighborhood respectively; obtain the distance value between the domain center of the positive feedback neighborhood and each pixel in the current context analysis domain, and take the pixel with the maximum distance value as the positive feedback edge point, obtain the distance value between the domain center of the negative feedback neighborhood and each pixel in the current context analysis domain, and take the pixel with the maximum distance value as the negative feedback edge point, take the negative feedback edge point as the center of the circle, and draw a circle with the distance between the negative feedback edge point and the domain center of the current context analysis domain as the radius and record it as the negative feedback domain, take the positive feedback edge point as the center of the circle, and draw a circle with the distance between the positive feedback edge point and the domain center of the current context analysis domain as the radius and record it as the positive feedback domain, then the feedback domain of the current context analysis domain is the intersection of the negative feedback domain and the positive feedback domain, and the feedback domain is used to calculate the fragile muscle energy assessment value.

[0069] Furthermore, the method for calculating the fragile muscle energy evaluation value using the feedback domain is as follows: define the proportion of pixels in all pixels of any domain that appear in the feedback domain as the infiltration value; set a ratio value as the infiltration ratio EZRt, and its value range EZRt∈[20%,50%]; when the infiltration value of the domain is greater than the infiltration ratio, or when the domain center of the domain is in the feedback domain, the domain is defined as being in the feedback domain; the off-pulse domain and the venous domain in the feedback domain are respectively regarded as the off-pulse bundle domain and the venous bundle domain;

[0070] The feedback domain is screened: the first screening condition is: if the number of vascular bundle domains in the current vascular analysis domain is less than 2, the current vascular analysis domain and its corresponding feedback domain are eliminated; the second screening condition is: if the proportion of pixels belonging to the vascular analysis domain in the current vascular analysis domain is less than the infiltration ratio, the current vascular analysis domain and its corresponding feedback domain are eliminated;

[0071] The above elimination means that it is no longer used as the basic object for subsequent calculation of the fragile muscle energy evaluation value. The context analysis domain and the feedback domain have a one-to-one correspondence.

[0072] The difference between the average value of the domain setting of each choroid bundle domain under the feedback domain and the choroid setting is recorded as the obstruction level DrbL; the fragile muscle energy assessment value HTfrK of the diastolic image is obtained by quantifying the obstruction level, and its calculation method is:

[0073] ;

[0074] Where j1 is the multiplicative variable, NMLY is the number of context domains, and DrbL j1 is the level of the j1th context domain, LBP j1 is the sum of the grayscale values ​​of the same pixels in the j1th feedback domain and the corresponding context analysis domain, XAP j1 is the sum of the grayscale values ​​of the same pixels in the j1th feedback domain and the adjacent domain of the corresponding context domain, LTP j1 is the number of pixels on the long axis of the j1th context analysis domain, and exp() is an exponential function with the natural number e as the base.

[0075] Further, in step S500, the method of identifying the state of brittle tilapia according to the muscle energy evaluation value is: when the brittle muscle energy evaluation value obtained by the individual is higher than that of the individuals under the same breeding environment, it is considered that the individual has a high risk of heart dysfunction during the brittle breeding process, otherwise it is identified that the risk of brittle heart dysfunction is low.

[0076] The embodiment of the present invention provides a system for identifying brittle tilapia based on M-mode ultrasound, such as Figure 2 The figure shows a structural diagram of an M-mode ultrasound-based brittle tilapia identification system of the present invention. The M-mode ultrasound-based brittle tilapia identification system of this embodiment includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the above-mentioned M-mode ultrasound-based brittle tilapia identification method embodiment are implemented.

[0077] The system comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to run in the following units of the system:

[0078] A fish body preparation unit for testing, used for preparing tilapia for testing;

[0079] An ultrasonic data acquisition unit, used for acquiring ultrasonic images of tilapia using M-mode ultrasound and obtaining diastolic images;

[0080] An image preprocessing unit, used for preprocessing the diastolic image and identifying venous pixels;

[0081] A brittleness risk assessment unit is used to perform diastolic muscle energy analysis based on the diastolic image and its venous pixels to obtain a muscle energy assessment value;

[0082] The risk identification unit is used to identify the status of brittle tilapia based on the muscle energy assessment value.

[0083] The brittle tilapia identification system based on M-mode ultrasound can be run in computing devices such as desktop computers, laptops, PDAs, and cloud servers. The brittle tilapia identification system based on M-mode ultrasound can be operated on systems that may include, but are not limited to, processors and memories. Those skilled in the art can understand that the example is only an example of a brittle tilapia identification system based on M-mode ultrasound, and does not constitute a limitation on a brittle tilapia identification system based on M-mode ultrasound. It may include more or fewer components than the example, or combine certain components, or different components. For example, the brittle tilapia identification system based on M-mode ultrasound may also include input and output devices, network access devices, buses, etc.

[0084] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the operating system of the embrittled tilapia identification system based on M-mode ultrasound, and uses various interfaces and lines to connect various parts of the entire operating system of the embrittled tilapia identification system based on M-mode ultrasound.

[0085] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the brittle tilapia identification system based on M-mode ultrasound by running or executing the computer program and / or module stored in the memory and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0086] Although the description of the present invention has been quite detailed and has been described in particular with respect to several described embodiments, it is not intended to be limited to any of these details or embodiments or any particular embodiment, so as to effectively cover the intended scope of the present invention. In addition, the present invention is described above with the embodiments foreseeable by the inventors, and its purpose is to provide a useful description, and those non-substantial changes to the present invention that are not currently foreseen may still represent equivalent changes of the present invention.

Claims

1. A method for identifying brittle tilapia based on M-mode ultrasound, characterized in that: The method comprises the following steps: S100, placing the tilapia under inspection; S200, using M-mode ultrasound to collect ultrasound images of tilapia and obtain diastolic images; S300, preprocessing the diastolic image and identifying venous pixels; S400, performing diastolic muscle energy analysis according to the diastolic image and its venous pixels to obtain a muscle energy evaluation value; S500, identifying the state of brittle tilapia according to the muscle energy evaluation value; The method for analyzing diastolic muscle energy according to the diastolic image and its choroid pixels to obtain the muscle energy evaluation value is as follows: the vascular area of ​​the processed image is divided into regions to obtain the analysis domain, the quasi-choroid analysis domain and the boundary analysis domain are separated from the analysis domain according to the choroid pixels, the inverse mapping attenuation gradient of the boundary analysis domain is calculated using the grayscale value of the quasi-choroid analysis domain, the attenuation gradient contraction condition is constructed according to the inverse mapping attenuation gradient to locate the muscle energy boundary domain, the gradient change rate is calculated according to the inverse mapping attenuation gradient of the muscle energy boundary domain, and the muscle energy evaluation value is calculated by combining the various gradient change rates.

2. A method for identifying brittle tilapia based on M-mode ultrasound according to claim 1, characterized in that: In step S100, the method for subjecting the tilapia to inspection treatment is as follows: the tilapia is transferred to a water tank filled with oxygenated fresh water and an anesthetic, wherein the anesthetic comprises any one of 20% benzocaine ethanol solution 1-2 ml / 10L water, chlorobutanol or 30% benzocaine ethanol solution 1 ml / L water; when the tilapia body does not move but the gill cover continues to move, the tilapia is laid flat on its side to complete the inspection treatment.

3. A method for identifying brittle tilapia based on M-mode ultrasound according to claim 1, characterized in that: In step S200, the method for acquiring diastolic images of tilapia using M-mode ultrasound is as follows: an M-mode ultrasound probe is placed between the pectoral fins of the tilapia, a number of ultrasound images are acquired along the ventral midline, and images in which the heart is in the end diastolic phase are recorded as diastolic images.

4. A method for identifying brittle tilapia based on M-mode ultrasound according to claim 3, characterized in that: In step S300, the method for preprocessing the diastolic image and identifying the venous pixels is as follows: performing median filtering denoising on the diastolic image, and enhancing the contrast of the diastolic image in combination with a histogram equalization algorithm; identifying the vascular area in the diastolic image through a pre-trained UNet convolutional neural network, and marking the pixels in the vascular area as venous pixels.

5. A method for identifying brittle tilapia based on M-mode ultrasound according to claim 1, characterized in that: In step S400, the diastolic muscle energy analysis is performed based on the diastolic image and its venous pixels, and the method for obtaining the muscle energy evaluation value is: the blood vessel area of ​​the processed image is divided into regions by a superpixel segmentation algorithm to obtain a domain, and the domain in which all pixels in the region are venous pixels is recorded as a quasi-venous domain, and the remaining domains adjacent to the quasi-venous domain are recorded as boundary domains; For any boundary domain, the nearest quasi-vein domain is taken as its inverse mapping domain, and the grayscale value of the centroid position of the inverse mapping domain is the inverse mapping grayscale; the proportion of vein pixels in the boundary domain is recorded as the vein recognition ratio, and the difference between the inverse mapping grayscale and the average value of the corresponding grayscale values ​​of each non-vein pixel is calculated, and the ratio of the difference obtained and the vein recognition ratio is the inverse mapping attenuation gradient Dcg; Initialize the muscle energy search domain corresponding to the current boundary domain as the current boundary domain, search for each boundary domain adjacent to the muscle energy search domain and merge them into the muscle energy search domain, iterate the search and merge until no boundary domain is merged into the muscle energy search domain; All boundary domains obtained by the search are recorded as search domains, and the distance between the search domain and the current boundary domain is recorded as search distance dsp. Each search domain is written into a sequence in ascending order of distance and recorded as a search domain sequence; The iteration number when any search domain is merged into the muscle energy search domain is recorded as TN, and the total number of iterations is recorded as the search degree max.TN. The attenuation gradient contraction condition is: Dcg p ≥(TN / max.TN)Dcg p +(1-TN / max.TN)Dcg st e -dsp ; Among them, Dcg p and Dcg st Represent the inverse mapping attenuation gradient of the current boundary analysis domain and the search domain respectively; e is a natural constant; Starting from the first element of the search domain sequence, search for the first element that does not meet the attenuation gradient contraction condition as the muscle energy boundary domain; the gradient change rate of the current boundary domain is Rvt_Dcg = (Dcg p -Dcg st ) / Dcg st ×Dsp; The muscle energy evaluation value is calculated based on the gradient change rate: HTfrK=mean{rp(max.TN i1 ) Rvt_Dcg i1 }, where i1 is the serial number of the boundary domain, and all boundary domains correspond to max.TN in the search degree i1 The frequency of occurrence of the corresponding value is rp(max.TN i1 ), max.TN i1 and Rvt_Dcg i1 are the search degree and gradient change rate of the i1th boundary domain respectively.

6. A method for identifying brittle tilapia based on M-mode ultrasound according to claim 1, characterized in that: In step S400, the method for performing diastolic muscle energy analysis based on the diastolic image and its venous pixels to obtain the muscle energy evaluation value can be replaced by: dividing the diastolic image into a plurality of regions by a superpixel segmentation algorithm and recording them as analysis domains, recording the analysis domains in which venous pixels exist as the venous analysis domains of the diastolic image, and recording the remaining regions as the venous separation analysis domains of the diastolic image; The domains pointed to by each pixel in the eight neighborhoods of a pixel are recorded as a set of TPRZ. When the length of the TPRZ of the pixel is ≥ 2, the pixel is defined as a boundary point. Any domain is taken as the current domain, and the TPRZs of each boundary point in the current domain are formed into a union. In this union, each domain except the current domain is taken as an adjacent domain. The line segment between the two pixels farthest apart in the domain is taken as the domain long axis, and the midpoint of the domain long axis is taken as the domain center. The average value of the grayscale value of each pixel in the domain is recorded as the domain fixed value, the average value of the grayscale value of each venation pixel is calculated and recorded as the venation fixed value, and the average value of the grayscale value of the remaining pixels is recorded as the off-venation fixed value. Take any context analysis domain as the current context analysis domain, obtain the domain values ​​of each adjacent analysis domain of the current context analysis domain, and take the adjacent analysis domains with the maximum domain value and the minimum domain value as the positive feedback neighborhood and the negative feedback neighborhood respectively; obtain the distance value between the domain center of the positive feedback neighborhood and each pixel in the current context analysis domain, and take the pixel with the maximum distance value as the positive feedback edge point, obtain the distance value between the domain center of the negative feedback neighborhood and each pixel in the current context analysis domain, and take the pixel with the maximum distance value as the negative feedback edge point, take the negative feedback edge point as the center of the circle, and draw a circle with the distance between the negative feedback edge point and the domain center of the current context analysis domain as the radius and record it as the negative feedback domain, take the positive feedback edge point as the center of the circle, and draw a circle with the distance between the positive feedback edge point and the domain center of the current context analysis domain as the radius and record it as the positive feedback domain, then the feedback domain of the current context analysis domain is the intersection of the negative feedback domain and the positive feedback domain, and the feedback domain is used to calculate the fragile muscle energy assessment value.

7. A method for identifying brittle tilapia based on M-mode ultrasound according to claim 6, characterized in that: The method of calculating the fragile muscle energy evaluation value using the feedback domain is as follows: define the proportion of pixels that appear in the feedback domain among all pixels in any analysis domain as the penetration value; Set a ratio value as the infiltration ratio EZRt, and its value range is EZRt∈[20%,50%]; when the infiltration value of the domain is greater than the infiltration ratio, or when the domain center of the domain is in the feedback domain, the domain is defined as being in the feedback domain; the off-pulse domain and the vein domain in the feedback domain are respectively regarded as the off-pulse bundle domain and the vein bundle domain; The feedback domain is screened: the first screening condition is: if the number of vascular bundle domains in the current vascular analysis domain is less than 2, the current vascular analysis domain and its corresponding feedback domain are eliminated; the second screening condition is: if the proportion of pixels belonging to the vascular analysis domain in the current vascular analysis domain is less than the infiltration ratio, the current vascular analysis domain and its corresponding feedback domain are eliminated; The difference between the average value of the domain setting of each choroid bundle domain under the feedback domain and the choroid setting is recorded as the hindrance level DrbL; the fragile muscle energy assessment value of the diastolic image is obtained by quantifying the hindrance level.

8. A method for identifying brittle tilapia based on M-mode ultrasound according to claim 1, characterized in that: In step S500, the method for identifying the state of brittle tilapia according to the muscle energy evaluation value is: when the brittle muscle energy evaluation value obtained by the individual is higher than that of the individuals under the same breeding environment, it is considered that the individual has a high risk of heart dysfunction during the brittle breeding process, otherwise it is identified that the risk of brittle heart dysfunction is low.

9. A system for identifying brittle tilapia based on M-mode ultrasound, characterized in that: The brittle tilapia identification system based on M-mode ultrasound comprises: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the brittle tilapia identification method based on M-mode ultrasound described in any one of claims 1 to 8 are implemented. The brittle tilapia identification system based on M-mode ultrasound runs on a desktop computer, a laptop computer, a PDA, and a computing device in a cloud data center.

Images