Focus puncture positioning method and device based on multi-source data analysis

Abstract

The invention discloses a focus puncture positioning method and device based on multi-source data analysis, and relates to the technical field of ultrasonic imaging, the general position of a focus is obtained through CT, MRI or ultrasonic image data, an image is obtained at a certain position to serve as a collection image, and a plurality of frames are continuously obtained to form a collection image sequence; dividing the image sequence into a plurality of sub-regions, and calculating a texture direction consistency index and a texture fluctuation index of each sub-region; according to the two indexes, the focus existence probability of the sub-region is calculated, the region with the maximum focus existence probability is recognized as a target sub-region, and accurate focus positioning is completed; therefore, when the puncture needle is close to the initial coverage area of the focus, the position of the real focus can be analyzed and judged according to the image obtained by the actual puncture needle, the influence of the fake focus is reduced, and the accuracy of focus puncture positioning is ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ultrasonic imaging, in particular to a lesion puncture positioning method and device based on multi-source data analysis. BACKGROUND

[0002] In modern minimally invasive intervention, lesion puncture positioning by multi-source data is widely used to improve the accuracy and safety of puncture operation. Usually, the initial coverage area of the lesion is obtained through CT, MRI or ultrasonic image data to provide preliminary navigation for puncture. Due to the resolution and real-time limitations of these CT, MRI or ultrasonic image data, it is difficult to accurately determine the specific boundary and position of the lesion. Therefore, real-time imaging devices such as ultrasonic probes at the end of the puncture needle are usually equipped during the puncture process to assist in dynamically adjusting the puncture path by real-time acquisition of local tissue images until the specific position of the lesion is found within the initial coverage area of the lesion, realizing more accurate lesion positioning.

[0003] However, real-time imaging of the puncture needle faces an important challenge: the texture and echo characteristics of some soft tissues and blood vessels around the lesion are highly similar to those of the lesion such as tumor, especially the blood vessel branches near the lesion often show extremely close signal characteristics; within the real-time image range of the puncture needle, some blood vessel structures may be located within the approximate area of the lesion determined by the previous multi-source image, causing the image recognition algorithm or the doctor to misjudge these blood vessel signals as lesions. Since these "false lesion" positions conform to the established lesion range, it is difficult to effectively distinguish them by spatial position, and the real lesion may be covered or submerged, which reduces the accuracy and reliability of puncture positioning. SUMMARY

[0004] The purpose of the present application is to solve the above-mentioned problems and provide a lesion puncture positioning method and device based on multi-source data analysis.

[0005] In the first aspect of the present application, a lesion puncture positioning method based on multi-source data analysis is first proposed, which comprises:

[0006] Obtain the initial coverage area of the lesion through CT, MRI or ultrasonic image data, and record the image obtained by the puncture needle at a certain position in the initial coverage area of the lesion as the acquisition image; and continuously acquire several frames of images at a certain position to obtain an acquisition image sequence;

[0007] Divide the images in the acquisition image sequence into several sub-regions, analyze the texture direction consistency information and texture fluctuation information in the acquisition image, and calculate the texture direction consistency index and texture fluctuation index of each sub-region;

[0008] Calculate the lesion existence probability of each sub-region according to the texture direction consistency index and the texture fluctuation index.

[0009] Analyze the probability of lesions in each sub-region, determine the sub-region with the direction of the lesion as the target sub-region, and determine the final location of the lesion.

[0010] Optionally, the calculation steps for the texture direction consistency index are as follows:

[0011] Each frame of the acquired image sequence is processed by grayscale conversion and normalization; each frame is divided into sub-regions of fixed size.

[0012] For each sub-region, the structure tensor method is used to extract the main texture direction angle θ. i,t θ i,t θ represents the main texture direction angle of the i-th sub-region in the t-th frame image. i,t ∈[0,π);

[0013] For each sub-region, calculate the standard deviation σθ of the main texture direction angle. i ;

[0014] The standard deviation is normalized and mapped to a texture orientation consistency index, calculated using the following formula: In the formula, TOC i For subregion R i Texture direction consistency index.

[0015] Optionally, the calculation steps for the texture fluctuation index are as follows:

[0016] Each frame of the acquired image sequence is processed by grayscale conversion and normalization; each frame is divided into sub-regions of fixed size.

[0017] Perform a two-dimensional Fourier transform on each sub-region to extract the frequency domain image of the region in frame t, and calculate the spectral energy distribution, i.e., the power spectrum, for each frame. Represents sub-region R in frame t. i The power spectrum;

[0018] Normalized spectral probability distribution Represents sub-region R in frame t i Normalized power spectrum at frequency coordinates (u, v);

[0019] Calculate the spectral entropy of the sub-region in frame t. For sub-region R in frame t i spectral entropy;

[0020] Calculate the spectral entropy of the sub-region The standard deviation is used as the texture fluctuation index.

[0021] Optionally, the step of calculating the lesion existence probability of each sub-region according to the texture direction consistency index and the texture fluctuation index, analyzing the probability of each sub-region existing the lesion, and determining the sub-region in the direction of the lesion as the target sub-region is:

[0022] The texture direction consistency index and the texture fluctuation index of each sub-region are normalized, the units are removed, and are mapped to the interval of 0-1, and the normalized texture direction consistency index and the texture fluctuation index are weighted and summed to obtain the lesion existence probability of each sub-region.

[0023] Optionally, the lesion existence probability of each sub-region is compared with a preset threshold value, if the lesion existence probability is less than the preset threshold value, it indicates that the probability of the sub-region existing the lesion is low, if the probability of all sub-regions existing the lesion is low, the puncture probe continues to detect in the initial coverage area of the lesion, avoids the current position, and determines the position of the lesion;

[0024] If the lesion existence probability is not less than the preset threshold value, it indicates that the probability of the sub-region existing the lesion is high, and the sub-region with the maximum lesion existence probability is recorded as the target sub-region, and the puncture probe moves in the direction of the target sub-region to determine the position of the lesion.

[0025] In the second aspect of the embodiment of the present application, a lesion puncture positioning device based on multi-source data analysis is provided, and the device comprises:

[0026] An image sequence module: obtaining the initial coverage area of the lesion through CT, MRI or ultrasonic image data, recording the image obtained by the puncture needle at a certain position in the initial coverage area of the lesion as a collection image, and continuously obtaining a plurality of frames of images at a certain position to obtain a collection image sequence;

[0027] An analysis module: dividing the images in the collection image sequence into a plurality of sub-regions, analyzing the texture direction consistency information and the texture fluctuation information in the collection image, and calculating the texture direction consistency index and the texture fluctuation index of each sub-region;

[0028] A probability module: calculating the lesion existence probability of each sub-region according to the texture direction consistency index and the texture fluctuation index;

[0029] A positioning module: analyzing the probability of each sub-region existing the lesion, determining the sub-region in the direction of the lesion as the target sub-region, and determining the final position of the lesion.

[0030] Optionally, the analysis module comprises:

[0031] A division module: performing grayscale and normalization processing on each frame of image in the collection image sequence; and dividing each frame of image into a sub-region with a fixed size;

[0032] Extraction module: using structure tensor method, extract the main texture direction angle θ i,t , θ i,t represents the main texture direction angle of the i-th sub-region in the t-th frame image, θ i,t ∈[0, π).

[0033] Calculation module: for each sub-region, calculate the standard deviation σθ i of the main texture direction angle.

[0034] Texture direction consistency module: normalize the standard deviation and map it to the texture direction consistency index, the calculation formula is: In the formula, TOC i is the texture direction consistency index of the sub-region R i .

[0035] Optionally, the analysis module further comprises:

[0036] Division module: for each frame of image in the collected image sequence, perform grayscale and normalization processing; divide each frame of image into sub-regions of fixed size;

[0037] Transformation module: perform two-dimensional Fourier transform on each sub-region to extract the frequency domain image of the region in the t-th frame, and calculate the frequency spectrum energy distribution of each frame, i.e. power spectrum represents the power spectrum of the sub-region R i in the t-th frame.

[0038] Probability distribution module: normalize the spectrum probability distribution represents the normalized power spectrum value of the sub-region R i at the frequency coordinates (u, v) in the t-th frame.

[0039] Spectrum entropy module: calculate the spectrum entropy of the sub-region in the t-th frame is the spectrum entropy of the sub-region R i in the t-th frame.

[0040] Texture fluctuation index module: calculate the standard deviation of the spectrum entropy of the sub-region , and take the standard deviation as the texture fluctuation index.

[0041] Optionally, the probability module is applied to:

[0042] Normalize the texture direction consistency index and the texture fluctuation index of each sub-region, remove the unit, map to the interval of 0-1, and weighted sum the normalized texture direction consistency index and the texture fluctuation index to obtain the lesion existence probability of each sub-region.

[0043] Optionally, the positioning module comprises:

[0044] The first positioning module: compare the lesion existence probability of each sub-region with the preset threshold value, if the lesion existence probability is less than the preset threshold value, it indicates that the probability of the sub-region existing the lesion is low, if the probability of all sub-regions existing the lesion is low, the puncture probe continues to detect in the initial covering area of the lesion, avoids the current position, and determines the position of the lesion;

[0045] The second positioning module: if the lesion existence probability is not less than the preset threshold value, it indicates that the probability of the sub-region existing the lesion is high, the sub-region with the maximum lesion existence probability is recorded as the target sub-region, the puncture probe moves to the direction of the target sub-region, and the position of the lesion is determined.

[0046] The beneficial effects of the present application are as follows:

[0047] The present application provides a lesion puncture positioning method and device based on multi-source data analysis, which obtains the initial covering area of the lesion through CT, MRI or ultrasonic image data, records the image obtained by the puncture needle at a certain position in the initial covering area of the lesion as the acquisition image, continuously acquires a plurality of image frames at a certain position to obtain an acquisition image sequence, divides the images in the acquisition image sequence into a plurality of sub-regions, analyzes the texture direction consistency information and texture fluctuation information in the acquisition image, calculates the texture direction consistency index and texture fluctuation index of each sub-region, calculates the lesion existence probability of each sub-region according to the texture direction consistency index and the texture fluctuation index, analyzes the probability of each sub-region existing the lesion, determines the sub-region in the direction of the lesion as the target sub-region, and determines the final position of the lesion. In this way, when the puncture needle approaches the initial covering area of the lesion, the position of the true lesion can be determined according to the actual image obtained by the puncture needle, the influence of the false lesion is reduced, and the accuracy of the lesion puncture positioning is ensured. BRIEF DESCRIPTION OF DRAWINGS

[0048] The present application will be further described below in conjunction with the drawings.

[0049] Figure 1 It is a flow chart of a lesion puncture positioning method based on multi-source data analysis;

[0050] Figure 2 It is a framework diagram of a lesion puncture positioning device based on multi-source data analysis. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0052] The embodiment of the present application provides a lesion puncture positioning method based on multi-source data analysis. Referring to Figure 1 , Figure 1 A flow chart of a lesion puncture positioning method based on multi-source data analysis is provided in the embodiment of the present application. The method comprises the following steps:

[0053] An initial coverage area of a lesion is obtained through CT, MRI or ultrasonic image data, an image obtained by a puncture needle at a certain position in the initial coverage area of the lesion is recorded as a collection image; and a plurality of frames of images are continuously obtained at the certain position to obtain a collection image sequence;

[0054] Images in the collection image sequence are divided into a plurality of sub-regions, texture direction consistency information and texture fluctuation information in the collection image are analyzed, and texture direction consistency indexes and texture fluctuation indexes of the sub-regions are calculated;

[0055] The lesion existence probabilities of the sub-regions are calculated according to the texture direction consistency indexes and the texture fluctuation indexes;

[0056] The existence probabilities of the sub-regions are analyzed, a sub-region in the direction of the lesion is determined as a target sub-region, and a final position of the lesion is determined.

[0057] According to the lesion puncture positioning method based on multi-source data analysis provided in the embodiment of the present application, when the puncture needle approaches the initial coverage area of the lesion, the position of the true lesion can be determined according to the image actually obtained by the puncture needle, the influence of the false lesion is reduced, and the accuracy of the lesion puncture positioning is ensured.

[0058] In one embodiment, an initial coverage area of a lesion is obtained through CT, MRI or ultrasonic image data, an image obtained by a puncture needle at a certain position in the initial coverage area of the lesion is recorded as a collection image; and a plurality of frames of images are continuously obtained at the certain position to obtain a collection image sequence;

[0059] Before the operation, the target part of the patient is systematically scanned by medical imaging equipment such as CT, MRI or ultrasound, and high-resolution, multi-slice image data is collected. CT scanning uses X-rays to perform tomographic imaging of tissues, which can clearly show the differences in tissue density, and is particularly suitable for identifying bones and calcified lesions; MRI uses a strong magnetic field and radio frequency pulses to respond to the hydrogen atoms in the human body. Internal tissue imaging reveals the contrast details of soft tissues, and can reflect soft tissue abnormalities such as tumors and edema; ultrasound imaging uses high-frequency sound waves to capture real-time echo information from tissues, making it easy to observe blood vessels and dynamic structures. The multi-modal image data collected is first pre-processed, including noise filtering, image registration and enhancement, to improve image quality and spatial consistency between different modalities. Medical image segmentation algorithms such as threshold-based segmentation, region growing, active contour models or deep learning-assisted segmentation methods are applied to accurately extract suspected lesion areas from the image. In combination with the doctor's clinical diagnosis experience, the spatial boundary of the lesion is labeled and confirmed, a three-dimensional reconstruction model is formed, and the initial coverage area of the lesion in the patient's body is determined. After determining the initial coverage area of the lesion, the puncture needle is moved to the initial coverage area of the lesion, and the specific position of the lesion in the initial coverage area of the lesion is accurately identified and located; the advantage of determining the initial coverage area of the lesion by preoperative CT, MRI or ultrasound medical imaging equipment is that the initial coverage area of the lesion is determined by preoperative influence, so that the puncture needle is moved to the initial coverage area of the lesion, greatly improving the overall efficiency of the puncture operation. After the spatial distribution of the lesion is determined, the doctor can plan the puncture path accordingly, without the need for extensive and blind exploration in the patient's body, thereby significantly shortening the time for intraoperative positioning; in addition, the preoperative rough positioning can also reduce unnecessary repeated puncture, effectively reducing the waste of medical resources, while reducing the risk of damage to the patient's tissues, improving the patient's comfort and safety.

[0060] It should be noted that the image obtained by the puncture needle at a certain position in the initial coverage area is referred to as the acquisition image; and a certain position is continuously acquired for several frames of images to obtain an acquisition image sequence;

[0061] Specifically, the puncture needle moves to the initial coverage area of the lesion, and when it reaches the initial coverage area of the lesion, due to the fact that the textures and echo characteristics of some soft tissues, blood vessels and lesions such as tumors in the image are highly similar, especially the blood vessel branches close to the lesion often show extremely close signal characteristics with the lesion; within the real-time image range of the puncture needle, some blood vessel structures may be located within the approximate area of the lesion determined by the previous multi-source image, causing the image recognition algorithm or the doctor to misjudge these blood vessel signals as lesions; in order to more accurately locate the lesion and reduce the distinction between the soft tissues, blood vessels and lesions such as tumors around the lesion, the true lesion position is determined by analyzing the image, and the puncture needle continuously and continuously acquires a plurality of image frames, generally 5-8 image frames, at a certain position in the initial coverage area to form an image sequence; the specific lesion position is determined by analyzing the image sequence, so that the lesion position can be more accurately determined based on dynamic changes; specifically, the specific steps of locating the lesion according to the image sequence are as follows:

[0062] In one embodiment, the images in the image sequence are divided into a plurality of sub-regions, the texture direction consistency information and the texture fluctuation information in the image sequence are analyzed, and the texture direction consistency index and the texture fluctuation index of each sub-region are calculated.

[0063] It should be noted that the images in the image sequence are divided into a plurality of sub-regions, and the texture change information of each sub-region is observed based on the image sequence to determine the probability that each sub-region contains lesion images, thereby performing accurate positioning.

[0064] Specifically, in one implementation, the calculation steps of the texture direction consistency index are as follows:

[0065] Each frame of image in the image sequence is subjected to grayscale and normalization processing; each frame of image is divided into a fixed size sub-region, and the i-th sub-region is denoted as R i .

[0066] The main texture direction angle θ i,t of each sub-region R i is extracted using the structure tensor method, θ i,t represents the main texture direction angle of the i-th sub-region in the t-th frame of image, and θ i,t ∈ [0, π), and the formula for calculation is as follows: In the formula, J xx represents the local average of the square of the gradient in the horizontal direction, i.e., the x direction, of the i-th sub-region image; J yy represents the local average of the square of the gradient in the vertical direction, i.e., the y direction, of the i-th sub-region image; and J xy represents the local average of the product of the gradients in the x and y directions of the i-th sub-region image.

[0067] For each sub-region R i , the fluctuation of the main texture direction angle of the corresponding region in all image frames is counted, the standard deviation of the main texture direction angle is calculated, and the formula is: In the formula, σθ i is the statistical deviation of the main texture direction angle of the sub-region R i ; μθ i is the mean of the main texture direction angle of the sub-region R i ; and T represents the total number of frames of images in the collected image sequence.

[0068] The standard deviation is normalized and mapped to the texture direction consistency index, and the formula is: In the formula, TOC i is the texture direction consistency index of the sub-region R i .

[0069] It should be noted that the texture direction consistency index of the sub-region is used to measure the stability of the main texture direction angle of the region in consecutive image frames; it reflects whether the texture direction of a fixed spatial position at multiple times (image frames) remains consistent. The index is obtained by calculating the standard deviation of the main texture direction angle of the sub-region in the image sequence, and then normalizing it, so the larger the value, the greater the fluctuation of the texture direction over time, and the more drastic the change in direction; the smaller the value, the more stable the texture direction, and the smaller the change. The reason why the smaller the texture direction consistency index, the greater the possibility of containing lesion tissue images in the corresponding sub-region, is that lesion tissues usually have the characteristics of structural disorder, irregularity, and abnormal morphology, and their texture directions often lack consistency and regularity in multiple image frames. Compared with normal tissues such as muscle, blood vessels, or fat tissue, these lesion regions have cell arrangement disorder, tissue boundary blur, or pathological blood flow changes, etc., which makes the main texture direction angle extracted from the image change between each frame, resulting in a large standard deviation of the main direction angle; therefore, the smaller the texture direction consistency index, the more drastic the fluctuation of the region in the texture direction, the higher the inconsistency, and it implies that the internal organization may have pathological changes or abnormal structures, which is a potential lesion region.

[0070] It should be noted that the benefit of analyzing the texture direction consistency index of the sub-region for judging whether the sub-region contains lesion tissue is that the temporal dynamic information in the image sequence can be used to reveal the stability and regularity of the tissue structure at the microscopic level, thereby providing a diagnostic clue that does not depend on color or morphological features. Compared with static analysis of a single frame image, this method is more sensitive to capture small abnormalities and potential lesions within the tissue by quantifying the change trend of the texture direction in consecutive frames, and is particularly suitable for identifying early lesions or lesions with blurred boundaries, which helps to achieve more accurate and robust lesion positioning and improve the success rate and safety of puncture guidance.

[0071] In an implementation manner, the calculation step of the texture fluctuation index is:

[0072] For each frame of image in the collected image sequence, grayscale and normalization processing is performed; each frame of image is divided into sub-regions of a fixed size, and the i-th sub-region is denoted as R i ;

[0073] For each sub-region R i , a two-dimensional Fourier transform is performed to extract the frequency domain image of the region in the t-th frame denotes the two-dimensional Fourier transform result of the sub-region R i in the t-th frame; (u, v) denotes the frequency coordinates; (I (t) |R i ) is the grayscale value matrix of the t-th frame image in the sub-region R i ;

[0074] The spectral energy distribution, i.e., the power spectrum P denotes the power spectrum of the sub-region R i in the t-th frame;

[0075] The normalized spectral probability distribution P denotes the normalized power spectrum value of the sub-region R i in the t-th frame at the frequency coordinates (u, v), representing the probability distribution; is the sum of the energy of all frequency points of the spectrum of the sub-region R i ;

[0076] The spectral entropy H i of the sub-region R in the t-th frame is calculated; In the formula, H i is the spectral entropy of the sub-region R i in the t-th frame, which measures the complexity and uncertainty of the spectrum;

[0077] The sub-region R i The mean value of the entire sequence of acquisition images The formula for calculation is: In the formula, T represents the total number of frames of images in the sequence of acquisition images;

[0078] The sub-region R i The spectral entropy of the sub-region R The standard deviation of the spectral entropy as the texture fluctuation index TFW i ;

[0079] It should be noted that the texture fluctuation index of the sub-region specifically refers to the degree of fluctuation of the spectral entropy value of the region in the continuous multi-frame image sequence collected by the puncture needle over time, which is usually quantified by the standard deviation of the spectral entropy. The spectral entropy reflects the complexity and uncertainty of the frequency components of the image region, and the smaller the fluctuation index, the more stable and consistent the texture features of the region in the entire image sequence, without significant structural changes or dynamic disturbances. Such stability is typically a characteristic of lesion tissue, as lesion tissue generally has a relatively uniform and fixed tissue structure, and its imaging texture changes little in a short period of time, and its texture features remain essentially constant with slight movement of the puncture needle or slight deformation of the tissue. In contrast, blood vessels and other non-lesion tissues are usually accompanied by blood flow and tissue dynamic changes, resulting in a large fluctuation in texture complexity and spectral entropy over time, with a higher texture fluctuation index. Therefore, sub-regions with lower texture fluctuation indices are more likely to be stable lesion tissue regions, while regions with higher fluctuation indices are likely to be blood vessels or other dynamic soft tissues. This distinction helps to more accurately lock the true lesion during the puncture process, reducing the risk of misjudging blood vessels and other structures as lesions, thereby improving the accuracy and safety of puncture positioning.

[0080] It should be noted that the stability difference of image texture over time can be used to effectively distinguish between structure-stable lesion tissue and dynamically changing non-lesion tissue such as blood vessels or normal soft tissue. This texture fluctuation-based analysis method does not rely on the static features of a single image, avoiding misjudgment problems caused by similar textures, improving the accuracy and reliability of lesion positioning. At the same time, by continuously monitoring the texture changes, it can reflect the small differences in tissue state in real time, providing more detailed reference for puncture needle path planning, which helps to reduce the risk and complications in the puncture process, and improve the safety and effectiveness of the overall treatment.

[0081] In one embodiment, the lesion presence probability of each sub-region is calculated according to the texture direction consistency index and the texture fluctuation index;

[0082] In one implementation, the calculation step of the lesion presence probability is:

[0083] The texture direction consistency index and the texture fluctuation index of each sub-region are normalized to remove units and mapped to the interval of 0-1, and the normalized texture direction consistency index and the normalized texture fluctuation index are weighted and summed to obtain the lesion existence probability of each sub-region; the calculation formula of the abnormal value is: In the formula, HY is the lesion existence probability, and cu and cr are the normalized texture direction consistency index and the normalized texture fluctuation index, respectively.

[0084] In one implementation, the step of determining the sub-region of the direction of the lesion as the target sub-region is:

[0085] The lesion existence probability of each sub-region is compared with a preset threshold value, if the lesion existence probability is less than the preset threshold value, it indicates that the probability of the sub-region existing the lesion is low; if the lesion existence probability of all sub-regions is low, the puncture probe continues to detect in the initial coverage area of the lesion, avoids the current position, and determines the position of the lesion until the position of the lesion is determined.

[0086] If the lesion existence probability is not less than the preset threshold value, it indicates that the probability of the sub-region existing the lesion is high; the sub-region with the maximum lesion existence probability is recorded as the target sub-region, the puncture probe moves in the direction of the target sub-region, and the position of the lesion is determined.

[0087] It should be noted that the commonly used normalization methods include Min-Max normalization, Z-Score standardization, etc., and the specific selection is determined according to the actual situation, which is not limited and described in detail; in addition, in general, the weight values of the normalized texture direction consistency index and the normalized texture fluctuation index are the same, and the weight sum is 1.

[0088] It should be noted that when comparing the lesion existence probability of each sub-region with the preset threshold, if the lesion existence probability of a certain sub-region is lower than the threshold, it means that the possibility of the lesion appearing in this region is small, while the possibility of normal tissue such as blood vessels is high, which indicates that the current position of the puncture probe is not the target region where the lesion is located. If the lesion existence probability of all sub-regions is detected to be lower than the threshold, it indicates that the lesion has not been accurately positioned within the range of the currently collected image sequence, and the probe may be in a normal tissue or blood vessel region that is not a lesion. At this time, the puncture probe does not stop, but continues to flexibly adjust the direction and position within the initial coverage area of the lesion, avoids the existing low probability area, and continues to acquire new image sequences for analysis, in order to gradually narrow the range of lesion positioning through the continuously updated texture features and fluctuation information. When the lesion existence probability of one or more sub-regions reaches or exceeds the threshold, it is determined that the possibility of the lesion existing in these regions is high, and the possibility of blood vessels or other normal tissues is low. In this case, the sub-region with the highest lesion existence probability is selected as the target sub-region, and the puncture probe is guided to move in this direction to achieve accurate positioning of the lesion.

[0089] It should be noted that through the above dynamic judgment and adjustment, combined with continuous real-time image acquisition and analysis, it can effectively avoid mispenetration of normal tissues such as blood vessels, improve the safety and success rate of puncture, and at the same time improve the accuracy of lesion positioning and the efficiency of operation, and improve the precision and safety of lesion positioning.

[0090] Based on the same inventive concept, the embodiments of the present application also provide a lesion puncture positioning device based on multi-source data analysis. Referring to Figure 2 , Figure 2 A framework diagram of a lesion puncture positioning device based on multi-source data analysis is provided for the embodiments of the present application, and the device comprises:

[0091] Image sequence module: obtain the initial coverage area of the lesion through CT, MRI or ultrasonic image data, and record the image obtained by the puncture needle at a certain position in the initial coverage area of the lesion as the acquisition image; and continuously acquire several frames of images at a certain position to obtain an acquisition image sequence;

[0092] Analysis module: divide the images in the acquisition image sequence into several sub-regions, analyze the texture direction consistency information and texture fluctuation information in the acquisition image, and calculate the texture direction consistency index and texture fluctuation index of each sub-region;

[0093] Probability module: calculate the lesion existence probability of each sub-region according to the texture direction consistency index and the texture fluctuation index;

[0094] Positioning module: analyze the probability of each sub-region existing a lesion, determine the sub-region of the direction of the lesion as the target sub-region, and determine the final position of the lesion.

[0095] Based on the lesion puncture positioning device based on multi-source data analysis provided by the embodiment of the application, when the puncture needle approaches the initial coverage area of the lesion, the position of the true lesion can be judged according to the image actually obtained by the puncture needle, the influence of the false lesion is reduced, and the accuracy of lesion puncture positioning is ensured.

[0096] In one embodiment, the analysis module comprises:

[0097] The division module: each frame of image in the collected image sequence is subjected to grayscale and normalization processing; each frame of image is divided into a fixed size sub-region;

[0098] The extraction module: the principal texture direction angle θ i,t of each sub-region is extracted using the structure tensor method i,t , wherein θ i,t ∈ [0, π).

[0099] The calculation module: for each sub-region, the standard deviation σθ i of the principal texture direction angle is calculated.

[0100] The texture direction consistency module: the standard deviation is normalized and mapped into a texture direction consistency index, and the calculation formula is: In the formula, TOC i is the texture direction consistency index of the sub-region R i .

[0101] In one embodiment, the analysis module further comprises:

[0102] The division module: each frame of image in the collected image sequence is subjected to grayscale and normalization processing; each frame of image is divided into a fixed size sub-region;

[0103] The transformation module: two-dimensional Fourier transform is performed on each sub-region to extract the frequency domain image of the region in the t-th frame, and the frequency spectrum energy distribution, i.e., the power spectrum P , of each frame is calculated, wherein P i represents the power spectrum of the sub-region R

[0104] The probability distribution module: the normalized frequency spectrum probability distribution P is calculated, wherein P i represents the normalized power spectrum value of the sub-region R

[0105] The spectrum entropy module: the spectrum entropy H of the sub-region in the t-th frame is calculated, wherein H i is the spectrum entropy of the sub-region R

[0106] texture fluctuation index module: calculate the spectral entropy of the sub-region the standard deviation of the spectral entropy, and take the standard deviation as the texture fluctuation index.

[0107] In one embodiment, the probability module is applied to:

[0108] normalizing the texture direction consistency index and the texture fluctuation index of each sub-region, removing the unit, mapping to the interval of 0-1, and performing weighted summation on the normalized texture direction consistency index and the texture fluctuation index to obtain the lesion existence probability of each sub-region.

[0109] In one embodiment, the positioning module comprises:

[0110] The first positioning module: compare the lesion existence probability of each sub-region with a preset threshold value. If the lesion existence probability is less than the preset threshold value, it indicates that the probability of the sub-region having a lesion is low. If the lesion existence probability of all sub-regions is low, the puncture probe continues to detect in the initial coverage area of the lesion, avoids the current position, and determines the position of the lesion.

[0111] The second positioning module: if the lesion existence probability is not less than the preset threshold value, it indicates that the probability of the sub-region having a lesion is high. The sub-region with the maximum lesion existence probability is recorded as a target sub-region, and the puncture probe moves in the direction of the target sub-region to determine the position of the lesion.

[0112] The above describes one embodiment of the present application in detail, but the content is only the preferred embodiment of the present application and cannot be considered as limiting the scope of the present application. Any equivalent changes and improvements made according to the scope of the present application should still belong to the scope of the present application.

Claims

1. A lesion puncture positioning method based on multi-source data analysis, characterized in that, The method comprises the following steps: Obtaining an initial coverage area of the lesion through CT, MRI or ultrasound image data, and taking an image obtained by the puncture needle at a certain position in the initial coverage area of the lesion as an acquisition image; and continuously obtaining a plurality of frames of images at the certain position to obtain an acquisition image sequence; Dividing the images in the acquisition image sequence into a plurality of sub-regions, analyzing the texture direction consistency information and texture fluctuation information in the acquisition image, and calculating the texture direction consistency index and the texture fluctuation index of each sub-region; Calculating the lesion existence probability of each sub-region according to the texture direction consistency index and the texture fluctuation index; Analyzing the probability of each sub-region existing the lesion, determining the sub-region of the direction of the lesion as a target sub-region, and determining the final position of the lesion. 2.The lesion puncture positioning method based on multi-source data analysis according to claim 1, characterized in that, The calculation step of the texture direction consistency index is: For each frame of image in the acquisition image sequence, perform grayscale and normalization processing; and divide each frame of image into sub-regions of a fixed size; Using structure tensor method for each sub-region, the main texture direction angle θ i,t , θ i,t represents the main texture direction angle of the i-th sub-region in the t-th frame image, θ i,t ∈ [0, π); For each sub-region, calculate the standard deviation σθ of the main texture direction angles i ; The standard deviation is normalized and mapped to the texture direction consistency index, and the formula is: In the formula, TOC i is the texture direction consistency index of the sub-region R i . 3.The lesion puncture positioning method based on multi-source data analysis according to claim 2, characterized in that, The calculation step of the texture fluctuation index is: A two-dimensional Fourier transform is performed on each sub-region to extract the frequency domain image of the region in the t-th frame, and the frequency spectrum energy distribution, i.e. the power spectrum of each frame is calculated denotes the power spectrum of the sub-region R i in the t-th frame. Normalized spectral probability distribution representing a sub-region R in the t-th frame i normalized power spectrum value at frequency coordinates (u, v); calculating a spectral entropy of the sub-region in the tth frame for the sub-region R in the tth frame i spectral entropy Computing a spectral entropy of the sub-region of the standard deviation, the standard deviation being used as a texture fluctuation index. 4.The lesion puncture positioning method based on multi-source data analysis according to claim 1, characterized in that, The step of calculating the lesion existence probability of each sub-region according to the texture direction consistency index and the texture fluctuation index is: Normalizing the texture direction consistency index and the texture fluctuation index of each sub-region, removing the unit, mapping to the interval of 0-1, and performing weighted summation on the normalized texture direction consistency index and the texture fluctuation index to obtain the lesion existence probability of each sub-region.

5. The method of claim 1, wherein, The step of analyzing the probability of each sub-region existing the lesion, determining the sub-region of the direction of the lesion as a target sub-region, and determining the final position of the lesion is: Comparing the lesion existence probability of each sub-region with a preset threshold value, if the lesion existence probability is not less than the preset threshold value, it indicates that the probability of the sub-region existing the lesion is relatively high; then the sub-region with the maximum lesion existence probability is recorded as the target sub-region, and the puncture probe is moved in the direction of the target sub-region to determine the position of the lesion; If the lesion existence probability is less than the preset threshold value, it indicates that the probability of the sub-region existing the lesion is relatively low; if the lesion existence probability of all sub-regions is relatively low, the puncture probe continues to detect in the initial coverage area of the lesion, avoids the current position, and determines the position of the lesion.

6. A lesion puncture positioning device based on multi-source data analysis, characterized in that, The device comprises: An image sequence module: obtaining an initial coverage area of the lesion through CT, MRI or ultrasound image data, taking an image obtained by the puncture needle at a certain position in the initial coverage area of the lesion as an acquisition image, and continuously obtaining a plurality of frames of images at the certain position to obtain an acquisition image sequence; An analysis module: dividing the images in the acquisition image sequence into a plurality of sub-regions, analyzing the texture direction consistency information and texture fluctuation information in the acquisition image, and calculating the texture direction consistency index and the texture fluctuation index of each sub-region; A probability module: calculating the lesion existence probability of each sub-region according to the texture direction consistency index and the texture fluctuation index; A positioning module: analyzing the probability of each sub-region existing the lesion, determining the sub-region of the direction of the lesion as a target sub-region, and determining the final position of the lesion.

7. The lesion puncture positioning device based on multi-source data analysis according to claim 6, characterized in that, The analysis module comprises: A division module: for each frame of image in the acquisition image sequence, performing grayscale and normalization processing; and dividing each frame of image into sub-regions of a fixed size; Extraction module: use structure tensor method to extract the main texture direction angle θ i,t , θ i,t represents the main texture direction angle of the i-th sub-region in the t-th frame image, θ i,t ∈ [0, π). Calculation module: For each sub-region, calculate the standard deviation of the main texture direction angle σθ i ; Texture direction consistency module: normalize the standard deviation and map it to a texture direction consistency index, the formula is: In the formula, TOC i is the texture direction consistency index of the sub-region R i .

8. The lesion puncture positioning device based on multi-source data analysis according to claim 7, characterized in that, The analysis module further comprises: The dividing module: each frame of image in the collected image sequence is subjected to grayscale and normalization processing; each frame of image is divided into sub-regions of fixed size; Transform module: two-dimensional Fourier transform is performed on each sub-region to extract the frequency domain image of the region in the tth frame, and the frequency spectrum energy distribution, i.e. power spectrum of each frame is calculated represents the power spectrum of the sub-region R i in the tth frame; Probability distribution module: Normalized spectral probability distribution denotes a sub-region R in the t-th frame i normalized power spectrum value at frequency coordinates (u, v); Spectrum entropy module: calculate the spectrum entropy of the sub-region in the tth frame For the spectrum entropy of the sub-region R i in the tth frame Texture fluctuation index module: calculates the spectral entropy of a sub-region. The standard deviation is used as the texture fluctuation index.

9. The lesion puncture positioning device based on multi-source data analysis according to claim 6, characterized in that, The probability module is applied to: The texture direction consistency index and the texture fluctuation index of each sub-region are subjected to normalization processing, the units are removed, are mapped to the interval of 0-1, and the normalized texture direction consistency index and the texture fluctuation index are subjected to weighted summation to obtain the lesion existence probability of each sub-region.

10. The lesion puncture positioning device based on multi-source data analysis according to claim 6, characterized in that, The positioning module comprises: The first positioning module: the lesion existence probability of each sub-region is compared with a preset threshold value; if the lesion existence probability is less than the preset threshold value, it indicates that the probability of the sub-region existing the lesion is low; if the lesion existence probability of all sub-regions is low, the puncture probe continues to detect in the initial coverage area of the lesion, avoids the current position, and determines the position of the lesion until the position of the lesion is determined; The second positioning module: if the lesion existence probability is not less than the preset threshold value, it indicates that the probability of the sub-region existing the lesion is high; the sub-region with the maximum lesion existence probability is recorded as a target sub-region, the puncture probe moves in the direction of the target sub-region, and the position of the lesion is determined.

Images