阅读说明：本技术 一种基于超声造影识别肝脏局灶性病变的系统 (System for discernment liver focal lesion based on ultrasonic contrast ) 是由 王宁 张东海 于 2019-11-05 设计创作，主要内容包括：本发明公开了一种基于超声造影识别肝脏局灶性病变的系统,包括录像采集模块、图像提取模块、图像预处理模块、分类标注模块、特征提取模块、数据处理模块、数据输出模块,本发明首先采用基于边缘增强与色彩平衡相结合的方式对造影图像进行去噪增强处理,从而在保留造影图像色彩的同时增强图像的有用信息,然后基于图像灰度直方图对造影图像进行特征提取,以获取合理的量化参数,从而可以最大程度上抑制探头移动和病人呼吸运动导致的病灶运动带来的诊断误差,以科学、客观、清晰、有效地反映病灶特征,以提高超声造影技术对肝脏局灶性病变特征识别的准确率与效率,从而为超声造影技术更快更好的发展提供新的途径。(The invention discloses a system for identifying liver focal lesion based on ultrasonic radiography, which comprises a video acquisition module, an image extraction module, an image preprocessing module, a classification marking module, a feature extraction module, a data processing module and a data output module, wherein the system firstly adopts a mode of combining edge enhancement and color balance to carry out denoising enhancement processing on a radiography image so as to enhance the useful information of the image while keeping the color of the radiography image, then carries out feature extraction on the radiography image based on an image gray histogram to obtain reasonable quantitative parameters, thereby inhibiting the diagnosis error caused by the movement of a probe and the movement of the focus caused by the respiratory movement of a patient to the maximum extent, reflecting the focus features scientifically, objectively, clearly and effectively so as to improve the accuracy and efficiency of the ultrasonic radiography technology for identifying the liver focal lesion features, thereby providing a new way for the ultrasonic contrast technology to develop more quickly and better.)

1. The utility model provides a system based on liver focal lesion is discerned to ultrasonic contrast, its characterized in that, includes video recording collection module, image extraction module, image preprocessing module, categorised mark module, characteristic extraction module, data processing module, data output module, wherein:

the video acquisition module is used for acquiring an ultrasonic radiography video from ultrasonic radiography equipment;

the image extraction module is used for extracting multi-frame ultrasonic contrast images from the ultrasonic contrast video according to a time sequence;

the image preprocessing module is used for preprocessing the ultrasonic contrast image data so as to obtain contrast data after color enhancement and dividing a rectangular region of interest into a plurality of rectangular sub-regions of interest;

the classification labeling module is used for performing classification labeling on the rectangular sub-regions of interest;

the characteristic extraction module is used for extracting the texture characteristics of each rectangular interesting sub-region obtained by the classification and labeling module;

the data processing module constructs an artificial neural network based on the textural features extracted by the feature extraction module and the corresponding recognition results thereof, and recognizes the image to be processed based on the artificial neural network;

and the data output module is used for outputting the processing result of the data processing module.

2. The system of claim 1, wherein the feature extraction module performs texture feature extraction on each rectangular sub-region of interest and estimates the motion trajectory of the particles, thereby deriving the weight redistribution on the particles.

3. The system of claim 1, further comprising an image rejection module for rejecting the significantly abnormal ultrasound contrast image frames, wherein the module calculates an average value of the original time intensity curve of the region of interest, multiplies the average value by a certain weighting factor to serve as a threshold for subsequent processing by the module, and then performs a difference between the average value and all the original time intensity values, compares the obtained result with the threshold, and regards the contrast image frames corresponding to the data smaller than the threshold as abnormal image frames, and the abnormal image frames are replaced by the average value.

4. The system of claim 1, wherein the ultrasound contrast image data comprises static image data and/or dynamic image data.

5. The system of claim 4, further comprising processing the dynamic image using smoothing filtering and time-series flow averaging.

6. The system of claim 1, further comprising an image cropping and magnifying module for cropping and magnifying the ultrasound contrast image according to the region of interest in the selected ultrasound contrast image.

7. The system of claim 1, further comprising a data storage module for storing the processing result outputted from the data processing module.

8. The system of claim 1, further comprising a labeling module for temporally and/or textually labeling the ultrasound contrast image.

9. The system of claim 1, further comprising a printing module for printing the output ultrasound contrast image processing result to form an ultrasound contrast report.

Technical Field

The invention relates to the technical field of ultrasound, in particular to a system for identifying liver focal lesion based on ultrasonic radiography.

Background

The liver focal lesion is one of the common clinical liver lesions, and patients affected by the disease can continuously show the symptoms of liver function decline and the like, and can seriously damage the liver function after the disease progresses. Because most of patients with liver focal lesion lack of obvious lesion range, typical early symptom expression is difficult to find, so that the problems of misdiagnosis, missed diagnosis and the like are easy to occur after the patients visit, the effective treatment time is endangered, and even the state of the patients is damaged due to the progress of the disease. Therefore, the most effective means for treating liver cancer is early detection and early treatment. Although CT, MRI and especially PET imaging have good effect on early diagnosis of liver cancer, ultrasonic imaging has become the first choice method for liver cancer screening at present due to the advantages of high safety, high speed, low cost and the like. Particularly, the emergence of the innovative technology of ultrasonic contrast imaging breaks through the limitation that the traditional B ultrasonic image can only describe structural information, and can carry out differential diagnosis on the focal hepatic nodules by observing the blood perfusion condition of the liver, thereby realizing the early detection on the malignant tumor of the liver.

The ultrasonic contrast technology mainly utilizes intravenous injection of microbubble contrast agent to scan the interface echo acoustic impedance difference of an object, so that the tumor focus is highlighted from the surrounding tissues, and the ultrasonic contrast technology can be used as a conventional means for screening liver cancer in liver cancer diagnosis and becomes one of the hot spots for liver disease diagnosis research. However, the current identification link of the characteristics in the ultrasonic contrast imaging technology mainly depends on naked eyes, and the accuracy of the identification link depends on the analysis level, patience and experience of different image analysts. In addition, human visual system processing information has the defects of inaccuracy and uncertainty, so that different image analysts have subjective differences in the identification of the contrast image characteristics. In addition, when the number of identification times is large, the image analyst is difficult to avoid phenomena such as visual fatigue and slow response, and the error rate is improved. Meanwhile, the data specification, resolution, size and the like acquired by different devices are difficult to be drawn neatly, and when different doctors use the device probes to acquire the radiography motion sequence images, the problems of frequent shaking of the acquired sequence images, disappearance of targets in a focus area and the like occur due to different methods, shaking of the probes, tissue deformation caused by human motion and respiration and violent shaking. In addition, in the ultrasonic imaging process, due to the uneven characteristic of human cell vascular tissues, speckle noise can be generated by the superposed scattered echoes, and the quality and the resolution of a contrast image can be seriously influenced, so that the subsequent extraction and analysis of the characteristics of a target in a focal zone are not facilitated. Therefore, in order to improve the accuracy and efficiency of the ultrasound contrast technology in identifying the liver focal lesion features, a liver focal lesion automatic image identification system which has an objective conclusion, a stable state and a comprehensive and clear image needs to be developed, so that a new way is provided for the faster and better development of the ultrasound contrast technology.

Disclosure of Invention

The invention aims to solve the problems in the prior art and provides a system for identifying liver focal lesion based on ultrasonic radiography.

In order to achieve the purpose, the invention adopts the technical scheme that:

the utility model provides a system based on liver focal lesion is discerned to ultrasonic contrast, includes video recording collection module, image extraction module, image preprocessing module, categorised mark module, characteristic extraction module, data module, data output module, wherein:

the video acquisition module is used for acquiring an ultrasonic radiography video from ultrasonic radiography equipment;

the image extraction module is used for extracting multi-frame ultrasonic contrast images from the ultrasonic contrast video according to a time sequence;

the image preprocessing module is used for preprocessing the ultrasonic contrast image data so as to obtain contrast data after color enhancement and dividing a rectangular region of interest into a plurality of rectangular sub-regions of interest;

the classification labeling module is used for performing classification labeling on the rectangular sub-regions of interest;

the characteristic extraction module is used for extracting the texture characteristics of each rectangular interesting sub-region obtained by the classification and labeling module;

the data processing module constructs an artificial neural network based on the textural features extracted by the feature extraction module and the corresponding recognition results thereof, and recognizes the image to be processed based on the artificial neural network;

and the data output module is used for outputting the processing result of the data processing module.

The image preprocessing module is used for preprocessing an ultrasonic contrast image, and the specific process method is as follows:

step a) first of all with the formula

Calculating the chromaticity, wherein K is the number of spectral channels, and K is 3 in the RGB color space;

step b) color space mapping is carried out on the formula to obtain the formula

Wherein S is _i(x, y) denotes an image of the ith channel, C _iA color recovery factor representing the ith channel for adjusting the scale of the 3 channel colors, f (×) representing a mapping function of the color space, β, α representing the gain normality and the controlled nonlinear intensity, respectively;

step c) multiplying the enhanced image by a color recovery factor to obtain the color R of the recovered original image _MsRCR(x,y)＝C(x,y)R _MSR(x, y) to display relatively dark areas present in the image to compensate for the image distortion;

step d) converting the RGB value range in the above formula to a display field of [0,255], wherein each color channel is adjusted by the absolute minimum and maximum of three color bands, and the calculation formula is as follows:

step e) clipping and correcting the extreme color values obtained in the step to obtain good contrast, so as to output an enhanced image, wherein a calculation formula is as follows:

R _MSRCR(x,y)＝g(R _MSRCR(x,y)-b]

where g denotes a gain coefficient and b denotes a deviation.

Furthermore, before preprocessing the ultrasonic contrast image, the ultrasonic contrast video is subjected to noise reduction processing, so that effective information is provided for subsequent image feature extraction. The algorithm controls diffusion increment from the gradient of the image, controls diffusion on a normal line and a tangent line, and enables the whole image to keep edge information. The calculation formula is as follows:

f ₁＝1/(1+C _a|v _n| ²)

f ₃＝1-1/(1+R|v _n| ²)

in the above calculation formula, f ₁Denotes a diffusion control coefficient, f ₂Representing the edge control coefficient, f ₃Denotes the edge enhancement coefficient, C _aShowing control of the anisotropic diffusion, C _bThe information indicating the original image is maintained, and σ indicates a parameter of a gaussian filter function.

As a further limitation of the above scheme, before extracting the texture feature of each rectangular sub-region of interest, the feature extraction module needs to estimate the motion trajectory of the particles by using a global optical flow method to obtain the redistribution of the weight to the particles.

Wherein, the motion trail of the particles is estimated by adopting a global optical flow method to obtain the redistribution of the weight to the particles, and the specific algorithm is as follows:

s1, converting the image subjected to denoising enhancement by the image preprocessing module from an RGB space to an HSV space, dividing the image into a plurality of regions according to color information, and converting each color C, wherein the conversion formulas of the space H, S, V in the HSV are respectively as follows:

H _c＝X _c(R _c)

V＝max(R,G,B)

when V ≠ 0

And S2, calculating the optical flow weight to obtain the weight value of each particle, and then normalizing the weight value.

The method for extracting the texture features of each rectangular sub-region of interest by the feature extraction module comprises the following specific processes:

step 1): converting a color image in the original image area of the candidate sample into a gray image, and acquiring a gray value of the original image area by a formula X of 0.297R +0.585G +0.118B, wherein X, R, G, B respectively represents the gray value and red, green and blue component values;

step 2) calculating the occurrence times of gray level images in different gray levels, and extracting a gray level histogram;

step 3) passing through a formula

Calculating the mean value of the gray level histogram and passing through a formula

Obtaining n-order moment (n is 1, 2, 3, 4) of histogram to obtain 4 characteristic vectors, and passing m ₁、m ₂、m ₃、m ₄4 are provided withThe feature vector represents a texture feature of the image region;

wherein f (x) represents the probability of the gray level of the image region in different gray levels, x represents the gray level value, u represents the mean value of the gray level value, m ₁、m ₂、m ₃、m ₄Respectively representing the dispersion, variance, skewness and peak value of the gray value of the image, and m ₂The method is used for measuring the gray contrast and is mainly used for describing the relative smoothness degree of a histogram and the depth condition of textures in an image; m is ₃Representing histogram skewness, namely the symmetry of gray values and mean values, and being used for describing the texture gray level fluctuation condition in an image area; m is ₄The relative flatness and the distribution aggregation of the histogram are shown, and the contrast of texture gray scale in the image area can be reflected.

As a further limitation of the above scheme, the method further comprises an image rejection module for rejecting the ultrasound contrast image frame with obvious abnormality, wherein the module calculates an average value of original time intensity curves of the region of interest, multiplies the average value by a certain weighting coefficient to serve as a threshold value for subsequent processing of the module, then calculates differences between the average value and all original time intensity values, compares the obtained result with the threshold value, regards the contrast image frame corresponding to the data smaller than the threshold value as an abnormal image frame, and replaces the abnormal image frame data with the average value.

As a further limitation of the above aspect, the ultrasound contrast image data includes still image data and/or dynamic image data.

As a further limitation of the above solution, the method further includes processing the dynamic image by using smoothing filtering and time-series flow averaging.

As a further limitation of the above scheme, the ultrasound imaging apparatus further includes an image cropping and enlarging module, configured to crop and enlarge the acquired ultrasound contrast image according to the region of interest in the selected ultrasound contrast image.

As a further limitation of the above scheme, the system further comprises a data storage module, which is used for storing the processing result output by the data processing module.

As a further limitation of the above solution, the system further comprises a labeling module for performing temporal and/or textual labeling on the ultrasound contrast image.

As a further limitation of the above solution, the ultrasound imaging apparatus further includes a printing module, configured to print the output ultrasound imaging processing result to form an ultrasound imaging report.

Compared with the prior art, the invention has the beneficial effects that:

(1) the invention adopts a mode of combining edge enhancement and color balance to carry out denoising enhancement processing on the contrast image, thereby enhancing useful information of the image while keeping the color of the contrast image, then carrying out feature extraction on the contrast image based on an image gray histogram to obtain reasonable quantitative parameters, thereby inhibiting the diagnosis error brought by focus motion caused by probe movement and patient respiratory motion to the maximum extent, reflecting the focus features scientifically, objectively, clearly and effectively, thereby improving the accuracy and efficiency of the ultrasonic contrast technology for identifying the liver focal lesion features and providing a new way for the faster and better development of the ultrasonic contrast technology.

(2) According to the invention, the motion trail of the particles is estimated by adopting a global optical flow method, and the redistribution of the weight to the particles is obtained, so that the sampling accuracy of the ultrasonic contrast image is greatly improved, and the accuracy of the ultrasonic contrast technology on the identification of the liver focal lesion features is further improved.

(3) The ultrasonic contrast images are extracted from the ultrasonic contrast video according to the time sequence, so that the analysis of the ultrasonic contrast video file is realized, the images can be arranged according to the time sequence of the images strictly, and the strict requirement of the ultrasonic contrast on the time sequence is met; by cutting and amplifying the region of interest in the image, the image change of the region of interest can still be clearly observed when a plurality of pictures are printed; the time and the character marking are carried out on each image, the strict requirement of ultrasonic radiography on the clear marking of the time is met, and the image preprocessing module, the feature extraction module and the data processing module are cooperated, so that the recognition degree and the universality of the ultrasonic radiography in the identification of liver focal lesion are greatly improved.

Drawings

Fig. 1 is a schematic structural diagram of a system for identifying liver focal lesions based on ultrasound contrast.

Fig. 2 is a flowchart of an algorithm for calculating particle weights according to an embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more clearly and completely understood, the technical solutions of the present invention will be described below with reference to the accompanying drawings, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments; all other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The present invention will be described in further detail below with reference to specific embodiments and with reference to the attached drawings.