A method for evaluating organ state based on laser speckle and an organ perfusion system

Abstract

This invention relates to the field of bio-optical imaging technology, specifically to an organ status assessment method and organ perfusion system based on laser speckle imaging. The method utilizes high-frame-rate laser speckle imaging technology combined with a physiological signal gating algorithm. It extracts the original speckle image only during the diastolic window when organ motion is minimal. The original speckle image is then preliminarily corrected using a vibration-speculiar distortion model. Subpixel-level registration is then performed on the preliminarily corrected speckle image using optical flow, generating a high signal-to-noise ratio microcirculation perfusion distribution map free of motion artifacts. Based on the microcirculation perfusion distribution map, perfusion heterogeneity indices for each target assessment region are calculated to determine whether local blood flow abnormalities exist in the transported organ, thus achieving status assessment and real-time monitoring of the transported organ. This assessment method is applicable not only to organ transport but also to organ quality evaluation and immediate post-transplant assessment scenarios.

Description

Technical Field

[0001] This invention relates to the field of bio-optical imaging technology, specifically to an organ status assessment method and organ perfusion system based on laser speckle. Background Technology

[0002] Currently, there are two main types of organ transport boxes used in organ transplantation. One type is a simple insulated box, where ice or other cryogenic substances are placed inside the organ during transport to maintain its activity for a period of time at a lower metabolic rate. The other type is a transport device with a perfusion system (such as the transport box developed by TransMedics). During organ transport, this device not only maintains the circulation of body fluids within the organ but is also equipped with a heater to maintain the organ's normal and continuous function at body temperature (e.g., maintaining normal heartbeat or normal lung ventilation), thereby maintaining organ activity over a longer period and enabling organ transport over longer distances. In addition, this type of transport box is equipped with various monitoring devices that can continuously monitor various physiological indicators of the organ during transport.

[0003] During organ transport, monitoring of organ function primarily relies on electrophysiological or biochemical parameters, such as the electrocardiogram (ECG) of the heart, or the vascular impedance or bioimpedance of the liver and kidneys, or the pressure, flow rate, pH value, and lactate level of the perfusion system. However, monitoring organ function using these parameters often only detects abnormalities after organ cell damage, failing to provide early warning and leading to missed opportunities for optimal intervention, severely impacting the entire organ transplant surgery. Furthermore, the pressure and flow rate of the perfusion system often only reflect the input and output of blood into the organ's main blood vessels, failing to indicate whether blood is effectively perfused into the capillaries at the organ's extremities, creating blind spots in organ function monitoring.

[0004] Laser speckle technology is a non-invasive medical detection method that utilizes the speckle pattern changes generated by the interaction of laser with living tissue to detect blood flow, metabolic state, and even structural abnormalities in organs. It is currently widely used in clinical and life science fields, commonly for examining cerebral blood flow and the healing status of wounds such as those on the skin. For example, existing technology CN103857335 B discloses a method for vascular imaging using laser speckle. This method calculates the direction of blood flow at the principal pixels in the original laser speckle image, then acquires anisotropic local neighborhoods arranged along the blood flow direction, calculates the local speckle contrast value at the principal pixels within these anisotropic neighborhoods, and obtains a laser speckle contrast image of that region based on these local speckle contrast values, thereby achieving anisotropic processing of the vascular region.

[0005] To address the impact of organ movement (artifacts) on laser speckle images during organ monitoring, prior art CN119136734 A discloses a motion-compensated laser speckle contrast imaging technique. This technique utilizes image registration to correct image misalignment caused by target tissue or camera movement, and then intelligently fuses multiple registered images to generate a clearer, more accurate speckle contrast image with significantly reduced motion artifacts. This prior art determines one or more registration parameters (e.g., vessel bifurcation, corners, etc.) based on the captured image sequence. These registration parameters describe the geometric transformation relationships (e.g., translation, rotation, scaling, etc.) between consecutive multi-frame speckle images. Motion analysis is used to align the multi-frame speckle images, achieving artifact removal. However, this prior art relies on image registration for artifact removal. Image registration (especially with dense optical flow and complex transformation models such as projective transformation), multi-frame weighted fusion, and optional multispectral correction are all computationally intensive tasks. This places high demands on processor performance (especially for high real-time requirements), potentially increasing system cost and power consumption, and making it difficult to meet the real-time needs of organ transport.

[0006] Meanwhile, while this scheme can effectively compensate for small to medium-amplitude movements, if the target undergoes very rapid, violent, or large-scale (relative to the scale of blood vessels) movement (which is almost unavoidable in organ transport), the overlapping area between consecutive frames may be too small, causing feature matching or region matching to fail, thus rendering the compensation algorithm ineffective. Furthermore, during transport, the organ may experience significant overall displacement, disrupting the continuous monitoring of the organ using existing technologies. Summary of the Invention

[0007] To address the technical problem that existing laser speckle technology is susceptible to vibration during organ transport, this application provides an organ condition assessment method and an organ perfusion system based on laser speckle, wherein the method includes the following steps:

[0008] Acquire raw speckle images of the transported organ in a stable phase and real-time vibration data of the organ storage device;

[0009] Using a vibration-speckle distortion model, rigid displacement correction is performed on the original speckle image based on real-time vibration data to obtain a rigidly corrected speckle image. The residual non-rigid deformation field in the rigidly corrected speckle image is calculated using the optical flow method, and pixel-level inverse compensation and alignment are performed to generate a quasi-static speckle image.

[0010] Calculate the microcirculation perfusion distribution map based on quasi-static speckle images;

[0011] The microcirculation perfusion distribution map is divided into multiple target assessment regions, and perfusion heterogeneity indexes are calculated based on the perfusion flux of each target assessment region.

[0012] A high-risk warning is issued when the infusion heterogeneity index exceeds the warning threshold.

[0013] The beneficial effects of the above method lie in that, based on high frame rate laser speckle imaging technology and combined with physiological signal gating algorithms, the original speckle image is extracted only during the diastolic window when organ motion is minimal. A vibration-speculiar distortion model is used to perform preliminary correction on the original speckle image, followed by sub-pixel registration using optical flow to generate a high signal-to-noise ratio microcirculation perfusion distribution map free of motion artifacts. Then, based on the microcirculation perfusion distribution map, perfusion heterogeneity indices for each target assessment region are calculated, thereby determining whether there are local blood flow abnormalities in the transported organ and achieving real-time assessment and monitoring of the transported organ's condition. This assessment method is not only applicable to organ transport but also to organ quality evaluation and immediate post-transplant assessment scenarios.

[0014] The organ perfusion system provided by this invention includes:

[0015] A perfusion subsystem for perfusing organ viability maintenance into a transport organ, the perfusion subsystem including an organ reservoir for maintaining the viability of the transport organ at body temperature;

[0016] The imaging module is used to acquire static speckle images and raw speckle images of the transported organ;

[0017] Vibration monitoring module is used to acquire real-time vibration data of the organ storage device;

[0018] The data processing module is used to perform rigid displacement correction on the original speckle image based on real-time vibration data using the vibration-speckle distortion model to obtain a rigidly corrected speckle image. It also uses the optical flow method to calculate the residual non-rigid deformation field in the rigidly corrected speckle image, performs pixel-level inverse compensation and alignment, generates a quasi-static speckle image, calculates the microcirculation perfusion distribution map based on the quasi-static speckle image, divides the microcirculation perfusion distribution map into multiple target evaluation regions, and calculates the perfusion heterogeneity index and the overall average perfusion flux based on the perfusion flux of each target evaluation region.

[0019] The perfusion system controller is used to control the operating status of the perfusion subsystem based on perfusion heterogeneity indicators and overall average perfusion flux, including perfusion pressure, pulsation waveform, and organ viability maintenance components.

[0020] The beneficial effects of the aforementioned organ perfusion system lie in establishing a microcirculation perfusion distribution map of the organ based on quasi-static speckle images corrected for motion artifacts. This serves as the foundation for constructing a closed-loop feedback regulation mechanism capable of real-time sensing changes in the organ's microscopic blood flow state. By dividing the microcirculation perfusion distribution map into several target assessment regions, areas of interest are observed in detail. Organ status is assessed based on overall perfusion flux and perfusion heterogeneity indices, thereby accurately identifying complex pathophysiological states where global ischemia and local microcirculation disorders coexist, and rapidly initiating synergistic intervention strategies. Simultaneously, this system establishes a precise feedback mechanism through real-time updated microcirculation perfusion distribution maps, using overall perfusion flux and perfusion heterogeneity indices to dynamically and individually adjust core operating parameters such as pressure and flow rate of the perfusion subsystem. This achieves a leap from "empirical, extensive regulation" to "refined, intelligent regulation based on real-time microscopic blood flow data," effectively improving the safety and effectiveness of organ perfusion and laying a solid foundation for improving the viability maintenance and transplant prognosis of transported organs. Attached Figure Description

[0021] Figure 1 This is an overall flowchart of the method provided by the present invention.

[0022] Figure 2 This is a flowchart for obtaining quasi-static speckle images in this invention.

[0023] Figure 3 This is a flowchart illustrating the training process of the vibration-speckle distortion model provided by the present invention.

[0024] Figure 4 This is a schematic diagram of the test system structure used in obtaining the vibration-speckle distortion model of this invention.

[0025] Figure 5 This is a schematic diagram of the overall structure of the organ perfusion system provided by the present invention. Detailed Implementation

[0026] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] Example 1

[0028] refer to Figure 1 This application provides a method for organ status assessment based on cardiac cycle gating, specifically including the following steps:

[0029] S1. Acquire raw speckle images of the transported organ in a stable phase and real-time vibration data of the organ storage device;

[0030] S2. Using the vibration-speckle distortion model, the original speckle image is rigidly displaced and corrected based on real-time vibration data to obtain a rigidly corrected speckle image. The residual non-rigid deformation field in the rigidly corrected speckle image is calculated using the optical flow method, and pixel-level inverse compensation and alignment are performed to generate a quasi-static speckle image.

[0031] S3. Calculate the microcirculation perfusion distribution map based on the quasi-static speckle image, and divide the microcirculation perfusion distribution map into multiple target evaluation areas. Calculate the perfusion heterogeneity index based on the perfusion flux of each target evaluation area.

[0032] S4. When the perfusion heterogeneity index is higher than the warning threshold, a high-risk warning is issued.

[0033] Specifically, when an isolated heart remains active, it undergoes most of its spontaneous movements, such as heartbeat and lung respiration. This results in motion artifacts in the acquired laser speckle images. Therefore, it is necessary to acquire laser speckle images when the organ is in a stable state, which can effectively reduce motion artifacts in the images.

[0034] In step S1, the method for determining the stable phase differs for different organs. For the heart, its circulatory cycle can be determined using electrophysiological signals or ECG (electrocardiogram) signals. For organs such as the lungs, the stable phase can be determined based on fluctuations in the perfusion pressure of the perfusion system.

[0035] The determination of the stable stage of an organ can also be based on the contrast or brightness fluctuations of the original speckle image.

[0036] When an organ is in a stable phase, the acquired laser speckle image has high quality. For example, when the heart is in the stable phase of diastole, its shape is relatively stable and its perfusion is high, resulting in a high-quality laser speckle image. Therefore, the original speckle image of the heart can be acquired at this stage, effectively reducing the influence of motion artifacts in the original speckle image.

[0037] Specifically, during the transport of an organ after it has been placed in an organ storage container (such as a transport box), the storage container inevitably vibrates, causing rigid displacement of the transported organ. Especially when significant vibrations occur, conventional motion artifact correction methods risk failure.

[0038] In step S2, refer to Figure 2 The present invention performs rigid displacement correction on the original speckle image through the following steps:

[0039] S21. Before transportation, after placing the transport organ into the transport box, obtain a static speckle image of the transport organ under static conditions as a reference speckle image, and calculate the global contrast of the reference speckle image as a reference global contrast.

[0040] When imaging blood vessels, laser speckle technology primarily relies on changes in the intensity of the laser speckle. Currently, the main parameter used to describe the dynamic changes in speckle is contrast. This is the ratio of the standard deviation of the light intensity fluctuation in the speckle image to the average light intensity:

[0041]

[0042] In the formula, This represents the standard deviation of the light intensity fluctuation in a speckle image. This represents the average light intensity.

[0043] Based on the relationship between the autocorrelation function of light intensity and the autocorrelation function of electric field, the relationship between contrast ratio and blood flow velocity is derived as follows:

[0044]

[0045] In the formula, For the decorrelation time of the speckle electric field, It is inversely proportional to the particle's velocity v. For the exposure time, This is the system correction factor, used to explain the polarization and differences between the detector and the speckle size; it is usually set to 1 by default.

[0046] For a given exposure time When the exposure time is much longer than the decorrelation time, the reciprocal of the square of the contrast can be approximated by the blood flow velocity. Proportional:

[0047]

[0048] There are two main ways to calculate contrast: spatial contrast and temporal contrast.

[0049] Spatial contrast is calculated by setting a sliding window on a single-frame speckle image to calculate the standard deviation and average value of the light intensity of pixels in the vicinity of each pixel. The contrast value of the local window centered on that pixel is then calculated according to the contrast calculation formula, and this value is used as the contrast value for that pixel, thus obtaining the contrast map of the single-frame speckle image. Although this method can process each frame of the speckle image individually, thus achieving real-time contrast map calculation, the sliding window method it uses results in a relatively low spatial resolution of the generated contrast.

[0050] Temporal contrast is calculated by taking a multi-frame speckle image sequence acquired over a period of time, and for each pixel, calculating the standard deviation and average value of its light intensity within the entire speckle image sequence, then calculating the contrast of that pixel, thus obtaining a contrast map of the organs within the entire camera's field of view during that sampling time period. The contrast map obtained in this way can spatially preserve the light intensity variation information of each pixel. However, because it requires calculation using multiple consecutive frames of speckle images, it has a certain time lag and needs to be calculated in a fixed period. However, with the increase in camera sampling frequency, the calculation period for temporal contrast has been effectively shortened. For example, some manufacturers' cameras can achieve a sampling frequency of 100Hz, enabling millisecond-level updates of speckle contrast images, which can basically meet the needs of most application scenarios.

[0051] Based on temporal contrast, global contrast can be obtained through the following steps:

[0052] S211. Sample multiple consecutive static speckle images of the transported organ under static conditions;

[0053] S212. For each pixel, extract the light intensity value of that pixel in all static speckle images, calculate the standard deviation and average value of the light intensity value of that pixel in all static speckle images, calculate the contrast of that pixel according to the contrast calculation formula, and establish a static contrast map.

[0054] S213. Calculate the average value of the contrast of each pixel in the static contrast image, which is the reference global contrast.

[0055] When the transport organ remains stationary, the contrast images acquired at different times should be consistent, and the calculated global contrast should not change significantly. However, when the transport organ vibrates, the movement of astigmatic particles on its surface causes an overall decrease in global contrast. Therefore, the change in global contrast can be used to determine the displacement correction effect of the speckle image.

[0056] S22. During the transport process, real-time vibration data of the transport box is acquired, and the original speckle images of the transported organs are acquired simultaneously. These original speckle images will be affected by the vibration during the transport process.

[0057] Specifically, the transport organ can be considered as a forced vibration under the action of external forces during transportation. Considering the organ as a forced vibration system, its forced vibration equation is:

[0058]

[0059] In the formula, Vibrational force applied externally to the organ, For the quality matrix of organs, For the damping matrix of the organ, Let be the stiffness matrix of the organ. Let be the acceleration vector of the organ. Since the laser speckle pattern primarily focuses on the upper surface of the organ, the acceleration of a reference point on the upper surface of the organ can be selected. This is the velocity vector of the organ; similarly, the velocity of a reference point on the upper surface of the organ can be chosen. Let be the displacement vector of a reference point on the upper surface of the organ, derived from the relationship between displacement, velocity, and acceleration:

[0060]

[0061]

[0062] The forced vibration equation can be modified as follows:

[0063]

[0064] Introducing state vectors :

[0065]

[0066] The original forced vibration equation can be simplified to:

[0067]

[0068] In the formula:

[0069]

[0070]

[0071]

[0072] The above equation establishes the relationship between the forcing force applied by the transport box and the displacement of the upper surface of the organ. The state matrix in this equation... and input matrix It is the mass matrix of organs Stiffness matrix and damping matrix These matrices are derived, but since they are unknown and difficult to measure accurately, they can be obtained by fitting using deep learning methods.

[0073] As can be seen from the above derivation, when establishing a vibration-speckle distortion model of an organ based on the aforementioned mathematical model, the required input is the forced driving force borne by the organ. However, since the vibration-speckle distortion model is not an exact analytical mathematical model, but rather an approximate solution obtained through fitting, the input vector in the above formula can be changed from force to acceleration, thereby reducing the difficulty of obtaining input data.

[0074] In summary, when obtaining real-time vibration data of the transport box using acceleration sensors (such as gyroscopes), what should be obtained is the acceleration of the transport box in each direction. In fact, it should be six values: displacement acceleration and rotational acceleration along the three axes.

[0075] S23. Use a pre-trained vibration-speckle distortion model to perform displacement correction on the original speckle image to obtain a rigidly corrected speckle image.

[0076] Specifically, training the vibration-speckle distortion model can be performed beforehand by vibrating organs. Since the organs need to remain viable during the experiment, and the preciousness of human organs makes it impossible to train the vibration-speckle distortion model using human organs, animal organs with similar anatomical structures and sizes to human organs can be used for vibration testing to train the model, such as pig livers and kidneys. (Reference) Figure 3 The specific training process of the vibration-speckle distortion model is as follows:

[0077] S231. Obtain a static speckle image of the test organ in a stationary state;

[0078] A typical test system consists of the following: Figure 4 As shown, the test organ is preserved in a temperature-controlled chamber and perfused with bodily fluids (and oxygen, such as in lungs, if necessary) through a perfusion system to maintain normal organ function. The temperature-controlled chamber is placed on a vibration table within the field of view of a laser speckle imager, which records static speckle images of the test organ in a stationary state and transmits them to the data processing module.

[0079] S232. Apply test vibration load to the test organ according to the test vibration load spectrum, and obtain test dynamic speckle image of the test organ under forced vibration.

[0080] During organ transport, there will be not only occasional large-amplitude vibrations (vibration amplitude relative to the pixel scale), such as vehicle bumps or aircraft airflow disturbances, but also certain high-frequency vibrations, such as the surge of a car engine under certain special operating conditions. Therefore, the test vibration loads applied to the organs should be as diverse as possible, including single-peak vibration loads, multi-cycle high-frequency vibrations, and combinations of vibration loads with different frequencies and amplitudes, so that the acquired test dynamic speckle images can include the organ's response to more vibration conditions as much as possible.

[0081] Preferably, the vibration load can be tested by collecting acceleration data of the transport box during multiple actual transports, and statistical analysis can be used to determine how to establish the test vibration load spectrum.

[0082] S233. Calculate the rigid body transformation vector of the test dynamic speckle image relative to the test static speckle image using the dense optical flow algorithm.

[0083] Dense optical flow calculates the displacement vector of each pixel between two adjacent frames (reflecting the apparent motion of the brightness pattern in the image plane), and uses polynomial fitting to calculate the rigid transformation between the two frames, thereby completing the registration between the two frames.

[0084] Based on the forced vibration model used in step S22, assuming the transport box (or vibration table) is an absolutely rigid object, the acceleration input to the surface of the test organ in contact with the transport box is the same everywhere on that surface. Since rigid body transformation is used to register the two frames of speckle images, the rigid body transformation between adjacent test dynamic speckle images calculated by the dense optical flow method is a global offset and rotation, where translation has two dimensions and rotation has one dimension. Therefore, the rigid body transformation vector should contain at least three dimensions of data. To maintain consistency between the mathematical forms of the input and output, the rigid body transformation vector can be expanded in dimension.

[0085] When calculating rigid body transformations, the dense optical flow method requires that the displacement between two adjacent speckle images be small. Therefore, a high-frequency speckle image sampling device is needed in the experiment to reduce the rigid body displacement between two adjacent speckle images.

[0086] S234. Continuously change the amplitude and direction of the test vibration load to obtain multiple rigid body transformation vectors;

[0087] During transportation, the vibration loads borne by the transported organ may come from different directions and have different amplitudes. In order for the vibration-speckle distortion model to correct rigid displacements in different directions, vibration loads with different directions and amplitudes should also be applied during the experiment to better train the vibration-speckle distortion model.

[0088] Because laser speckle imagers can perform high-frequency image sampling, they can sample the entire response process of the test organ to the vibration load during the vibration of the vibration table, thus forming multiple frames of dynamic speckle images. By using dense optical flow, multiple rigid body transformation vectors in the entire vibration response process can be obtained, and the time of each rigid body transformation vector can be marked, thereby preserving the temporal characteristics of the test organ's response to the vibration load.

[0089] S235. The initial vibration-speckle distortion model is trained using a deep learning algorithm with multiple rigid body transformation vectors and corresponding test vibration loads.

[0090] Specifically, by combining the time-varying characteristics of vibration loads with the dynamic response of the test organ, the training effect of the model can be further optimized. During the experiment, it is necessary to ensure that the output frequency and amplitude of the vibration table can cover various situations that may be encountered in actual transportation, thereby improving the model's generalization ability. Furthermore, by decomposing and recombining vibration loads in different directions, a more comprehensive rigid body transformation dataset can be obtained, providing richer training samples for subsequent deep learning algorithms. In the data preprocessing stage, denoising and enhancement operations should be performed on the acquired speckle images to ensure the accuracy of the optical flow calculation results. Simultaneously, the timestamp information should be synchronized with the changes in vibration loads to ensure the consistency of temporal characteristics.

[0091] Furthermore, after initial training of the initial vibration-speckle distortion model, the trained vibration-speckle distortion model can be used to perform displacement correction on the acquired test dynamic speckle images to verify the accuracy of the vibration-speckle distortion model. This specifically includes the following steps:

[0092] S236. Use the pre-trained vibration-speckle distortion model to perform displacement correction on the test dynamic speckle image, and calculate the SSIM (structural similarity) between the corrected test dynamic speckle image and the test static speckle image.

[0093] S237. Establish a loss function using SSIM, iteratively train the vibration-speckle distortion model, and terminate the training when SSIM is greater than the qualified threshold, outputting the vibration-speckle distortion model.

[0094] SSIM (Structural Similarity Simulation) is a process designed to quantify the similarity between two images from a human visual perception perspective. It measures the similarity between two speckle images using three independent comparison dimensions: brightness, contrast, and structure. It calculates the mean, variance, and covariance (brightness or contrast) of pixels in the local region surrounding each pixel in the corrected test dynamic speckle image and the static speckle image using a sliding window approach. After traversing the entire image, an SSIM map with a size similar to the original image (slightly smaller due to boundary effects) is obtained. Each pixel value in the map represents the similarity of the region surrounding that pixel. The global SSIM (the average value of all pixels in the SSIM map) is then calculated to evaluate the similarity between the test dynamic speckle image and the test static speckle image, thus assessing the correction effect of the vibration-speckle distortion model. A higher global SSIM indicates a greater similarity between the test dynamic and test static speckle images, signifying a better correction effect of the vibration-speckle distortion model. Typically, a global SSIM greater than 0.9 indicates that the test dynamic speckle image is almost identical to the test static speckle image, while a global SSIM less than 0.7 indicates severely poor correction, affecting subsequent contrast calculations. Therefore, using SSIM as an image quality metric provides a clear optimization target for the model training process, enabling retraining of the pre-trained vibration-speckle distortion model and further improving its accuracy.

[0095] During organ transport, real-time vibration data of the transport container is sampled, including displacement acceleration in three directions and rotational acceleration in three directions, to construct the input vector for the vibration-speckle distortion model. A sampling timestamp is then appended to this input vector. The vibration-speckle distortion model outputs a speckle distortion correction vector at the corresponding time based on the input vector, and offsets and / or rotates the dynamic speckle image sampled at the same time to align it with a reference speckle image.

[0096] S24. Calculate the residual non-rigid deformation field in the rigidly corrected speckle image using the optical flow method, perform pixel-level inverse compensation and alignment on the rigidly corrected speckle image, and generate a quasi-static speckle image.

[0097] Specifically, optical flow is a method for calculating the motion trajectory of pixels in an image sequence. It estimates the displacement vector of pixels, i.e., the optical flow field, by analyzing the changes in gray-level distribution in consecutive frames of images. In this step, the rigidly corrected speckle image and a reference speckle image are first used as inputs to the optical flow method. An image pyramid structure is constructed using multi-scale strategies such as image pyramids to address the challenges of optical flow estimation under large displacement conditions. Then, based on the gray-level conservation assumption and spatial smoothness constraints, the optical flow vector of each pixel is solved through iterative optimization. This optical flow vector represents the residual non-rigid deformation field of the rigidly corrected speckle image relative to the reference speckle image. Finally, based on the solved residual non-rigid deformation field, inverse displacement compensation is performed on each pixel in the rigidly corrected speckle image, thereby further eliminating the subtle deformations caused by non-rigid tissue movement in the original speckle image and generating a quasi-static speckle image that is closer to a static state, providing a more stable speckle image data foundation for subsequent organ state assessment.

[0098] S25. Synthesize a real-time contrast image based on the quasi-static speckle image, and calculate the real-time global contrast of the real-time contrast image. If the absolute difference between the real-time global contrast and the reference global contrast is greater than the allowable contrast deviation, then retrain the vibration-speckle distortion model using the acquired original speckle image; otherwise, output the quasi-static speckle image.

[0099] Specifically, since a high-frequency sampling laser speckle imager can be used, the time-contrast method is used to calculate the real-time contrast image, which not only preserves the high spatial resolution of the contrast but also has better temporal resolution.

[0100] Based on the above analysis, the process of synthesizing real-time contrast images is as follows:

[0101] S241. Based on the multi-frame corrected speckle images obtained during the dynamic monitoring period, calculate the standard deviation and average value of the light intensity value of each pixel in the multi-frame corrected speckle images, calculate the contrast of the pixel, and use it as the pixel value corresponding to the real-time contrast image.

[0102] For example, if the laser speckle imager used has an image sampling frequency of 100Hz, in order to maintain the millisecond-level update rate of the contrast image, a real-time contrast image can be synthesized every 10 frames of dynamic speckle images sampled.

[0103] After acquiring the real-time contrast image, calculate the real-time global contrast and the absolute difference between the real-time global contrast and the reference global contrast.

[0104] For transport organs, when they are functioning well, the brightness and darkness of laser speckle at different locations should remain relatively consistent at different times. This is reflected in real-time contrast images, where the value of each pixel (i.e., the contrast value) should not differ significantly from the reference image in real-time contrast images acquired at different times, and the contrast value of the stationary area of ​​the transport organ should be lower than that of the moving area (e.g., blood vessels). Even if the transport organ experiences contrast changes due to lesions in some areas (e.g., local thrombosis or tissue necrosis leading to impaired blood flow and a decrease in contrast in that area), the impact on the overall global contrast of the image is small. Therefore, the absolute difference between the real-time global contrast and the reference global contrast should be within a reasonable deviation range (allowing for contrast deviation). However, when there is a large displacement deviation between the original speckle images of different frames, it leads to the fusion of speckles of different brightness and darkness, causing the contrast values ​​of different areas to average out. This manifests in the contrast image as blurred boundaries between the static and dynamic areas of the transport organ, resulting in a decrease in global contrast. Therefore, comparing the global contrast can determine whether the vibration-speckle distortion model effectively eliminates artifacts in the original speckle image. If the distortion is not eliminated, the vibration-speckle distortion model is retrained using the original speckle images that have been acquired.

[0105] During this retraining process, the loss function can be constructed using the absolute difference between the real-time global contrast and the reference global contrast. The training objective for the vibration-speckle distortion model is that the absolute difference between the real-time global contrast and the reference global contrast is less than the allowable contrast deviation. Since the real-time global contrast of each real-time contrast image has already been calculated in the aforementioned calculation process, the computational load of the retraining process can be reduced, thereby improving the model's improvement speed and ensuring the real-time monitoring of transported organs.

[0106] The aforementioned method samples vibration data from the organ storage device during transport and uses a pre-trained vibration-speckle distortion model to correct the displacement of dynamic speckle images, thereby eliminating vibration-induced speckle image distortion and improving the accuracy of dynamic speckle images. Simultaneously, global contrast is calculated based on a real-time contrast image synthesized from the corrected speckle image. The displacement correction effect of the vibration-speckle distortion model is judged by comparing it with a reference global contrast. When the correction effect is unsatisfactory, a loss function is established based on the deviation of the global contrast to retrain the vibration-speckle distortion model, thereby optimizing the displacement correction effect, further improving the model's artifact removal performance, and increasing the accuracy of displacement correction. Furthermore, since displacement correction is performed using a pre-trained vibration-speckle distortion model, which is a mathematical model based on the forced vibration model of an elastic system, the vibration-speckle distortion model has a faster computation speed and better real-time performance. The above method achieves alignment of the laser speckle image through a dual correction step of "large displacement + small displacement", thereby avoiding the failure of motion artifact correction of the original speckle image caused by the vibration of the organ storage device, and enabling a wider range of applications for monitoring organ status using laser speckle imaging technology.

[0107] In step S3, each pixel value of the microcirculation perfusion distribution map represents the fluid circulation velocity at the corresponding location on the transport organ. For transport organs preserved in vitro and maintained by a perfusion system, it reflects the perfusion flux at the corresponding location of the transport organ. Each pixel value of the microcirculation perfusion distribution map can be obtained from a real-time contrast map synthesized from a quasi-static speckle map, or it can be directly calculated from the quasi-static speckle map using the speckle contrast calculation formula. When calculating the pixel values ​​of the microcirculation perfusion distribution map, either the time contrast method (as described in steps S211 to S213 above) or the spatial contrast method can be used.

[0108] For organs preserved in vitro, the blood flow velocity is directly proportional to the perfusion flux, and the blood flow velocity is inversely proportional to the square of the laser speckle contrast. Therefore, the pixel value of each pixel in the microcirculation perfusion distribution map can be calculated by the following formula:

[0109]

[0110] In the formula, For pixels The time speckle contrast is given by C, which is a calibration constant obtained experimentally.

[0111] Specifically, in step S3, the division of the target assessment area can be done manually by medical staff to focus on monitoring the state changes of some areas, or it can be done automatically by a program. For example, the minimum blood vessel length can be used as the diagonal length of the target assessment area, and image recognition technology can be used to divide several non-overlapping target assessment areas on the microcirculation perfusion distribution map. The standard deviation and coefficient of variation of the pixel values ​​of each target assessment area in the microcirculation perfusion distribution map are calculated as perfusion heterogeneity indicators.

[0112] The coefficient of variation (CV) is preferred as an indicator of perfusion heterogeneity. The CV is the ratio of the standard deviation to the mean of pixel values ​​within the target assessment region. The CV provides a purely "non-uniform" measure independent of the overall perfusion level. A high CV indicates significant differences in perfusion flux at different points within the organ, meaning some areas have abundant blood flow while others are severely ischemic. This suggests microcirculatory dysfunction, disordered blood flow distribution, or capillary thinning, commonly seen in vascular occlusion or endothelial injury. Conversely, a low CV indicates spatially uniform blood flow, a well-functioning microvascular network, and coordinated perfusion. Therefore, when the perfusion heterogeneity index exceeds the warning threshold, there is a potential risk of organ damage, triggering a high-risk warning.

[0113] The organ status assessment method provided in this application can be applied not only to organ transportation, but also to the assessment of the quality of transported organs and the assessment after transplantation.

[0114] Specifically, the following steps are performed when conducting a quality assessment of transported organs:

[0115] S5. After obtaining the transport organ, perform a short "load test" to temporarily increase the perfusion flow of the transport organ and obtain the change in the perfusion heterogeneity index of the target assessment area. If the change in the perfusion heterogeneity index of the target assessment area is greater than the risk threshold, it indicates that the increase in perfusion flow in the target assessment area is uneven, and there is micro-damage or no-reflow phenomenon. A high-risk warning is output. Otherwise, it indicates that the perfusion flux is increased relatively evenly and the condition of the transport organ is good.

[0116] The change in the irrigation heterogeneity index is the absolute difference between the irrigation heterogeneity index at the current irrigation flow rate and the irrigation heterogeneity index before the change in irrigation flow rate.

[0117] Specifically, when applied to post-transplant assessment scenarios, the following steps can also be performed:

[0118] S6. After the vascular anastomosis is completed, the microcirculation perfusion distribution map is acquired in real time;

[0119] S7. Calculate the perfusion heterogeneity index and average perfusion flux change rate for each target assessment area based on the microcirculation perfusion distribution map, and mark the target assessment areas where the perfusion heterogeneity index is greater than the warning threshold, or the target assessment areas where the average perfusion flux change rate is less than the reperfusion slope threshold.

[0120] For example, after a transplant, once the blood vessels are anastomosed, the blood flow occlusion clamp is released, the transported organ is reperfused with blood, and the perfusion heterogeneity index and the average perfusion flux change rate of the target assessment area are calculated based on the real-time microcirculation perfusion distribution map. The areas with excessively large perfusion heterogeneity index or small average perfusion flux change rate are displayed on the monitor in pseudo-color form to indicate anastomotic stenosis, vascular tortuosity, or distal thrombosis, assisting the surgeon in making real-time corrections during the operation.

[0121] In the post-transplantation evaluation scenario, since the organ is no longer stored in the organ storage device, the real-time vibration data can be defined as 0 in step S2. That is, there is no rigid displacement of the transported organ caused by the vibration of the organ storage device, so there is no need to correct this part of the rigid displacement.

[0122] In summary, the assessment method provided in this application, based on high frame rate laser speckle imaging technology and combined with a physiological signal gating algorithm, extracts the original speckle image only during the diastolic window when organ motion is minimal. A vibration-speculiar distortion model is used to perform preliminary correction on the original speckle image, followed by sub-pixel registration using optical flow to generate a high signal-to-noise ratio microcirculation perfusion distribution map free of motion artifacts. The perfusion heterogeneity index for each target assessment region is then calculated based on the microcirculation perfusion distribution map to determine whether there are local blood flow abnormalities in the transported organ, thus achieving assessment and real-time monitoring of the transported organ's condition. This assessment method is applicable not only to organ transport but also to organ quality evaluation and immediate post-transplant assessment scenarios.

[0123] Example 2

[0124] Based on the method provided in Embodiment 1, the present invention also provides an organ perfusion system, referencing... Figure 5 ,include:

[0125] The perfusion subsystem is used to infuse the transport organ with organ activity maintainers, which can be liquids such as blood and drug solutions, and oxygen (e.g., lungs) when necessary. The perfusion subsystem includes an organ reservoir for storing the transport organ. Generally, the organ reservoir can maintain the internal temperature close to the human body temperature, so that the transport organ can maintain its activity at body temperature. The organ reservoir can also send stimulating electrical signals (pulsatile electrical signals) to the transport organ through built-in electrodes to induce regular pulsation of the transport organ (e.g., heart).

[0126] The imaging module is used to acquire static speckle images and raw speckle images of the transported organ;

[0127] Vibration monitoring module is used to acquire real-time vibration data of the organ storage device;

[0128] The data processing module is used to perform rigid displacement correction on the original speckle image based on real-time vibration data using the vibration-speckle distortion model to obtain a rigidly corrected speckle image. It also uses the optical flow method to calculate the residual non-rigid deformation field in the rigidly corrected speckle image, performs pixel-level inverse compensation and alignment, generates a quasi-static speckle image, calculates the microcirculation perfusion distribution map based on the quasi-static speckle image, divides the microcirculation perfusion distribution map into multiple target evaluation regions, and calculates the perfusion heterogeneity index based on the perfusion flux of each target evaluation region.

[0129] The perfusion system controller is used to control the operating status of the perfusion subsystem based on perfusion heterogeneity indicators and the overall average perfusion flux of the entire transport organ, including perfusion pressure, pulsation waveform, and organ viability maintenance components.

[0130] The overall average perfusion flux can be calculated based on the value of each pixel in the microcirculation perfusion distribution map, i.e., the average value of all pixel values.

[0131] Specifically, when the perfusion heterogeneity index of the target assessment area does not change significantly (less than or equal to the warning threshold) and the overall average perfusion flux is low (lower than the theoretical perfusion flux threshold), it indicates that the entire organ is at a low perfusion level, and the perfusion pressure of the perfusion subsystem should be increased.

[0132] Specifically, when the perfusion heterogeneity index of a target assessment area exceeds the warning threshold, while the overall average perfusion flux is normal, it indicates that the total flux of the perfusion system is sufficient, but there is insufficient blood flow distribution within the target assessment area, resulting in ischemia in some areas. In this case, the perfusion strategy can be adjusted. For example, adding vasodilators (prostaglandins, papaverine, etc.) to organ activity maintainers can relax constricted microvessels and improve blood inflow to ischemic areas. Alternatively, optimizing the pulsation waveform to better match the physiological state of the organ may help open microcirculation, or adjusting the composition of organ activity maintainers, such as pH and calcium ion concentration, can create an environment more conducive to microvascular dilation.

[0133] Specifically, when the overall average perfusion flux is low (below the theoretical perfusion flux threshold) and the perfusion heterogeneity index of one or more target assessment areas is higher than the warning threshold, it indicates that the transport organ is in a state of overall ischemia accompanied by severe microcirculatory dysfunction. In this case, the perfusion pressure of the perfusion subsystem is increased first, and vasodilator injection is triggered simultaneously. The microcirculatory perfusion distribution map of the transport organ is continuously acquired, and subsequent changes in the microcirculatory status are monitored. The working status of the perfusion subsystem is then dynamically adjusted.

[0134] The organ perfusion system provided in this embodiment establishes a microcirculation perfusion distribution map of the organ based on quasi-static speckle images after motion artifact correction. Based on this, a closed-loop feedback regulation mechanism capable of real-time sensing changes in the organ's microcirculation blood flow state is constructed. By dividing the microcirculation perfusion distribution map into several target assessment regions, areas of interest are observed in detail. The organ state is assessed based on the total perfusion flux and perfusion heterogeneity index, thereby accurately identifying the complex pathophysiological state of coexisting global ischemia and local microcirculation disorders, and rapidly initiating synergistic intervention strategies. Simultaneously, this system establishes a precise feedback mechanism through a real-time updated microcirculation perfusion distribution map, using the total perfusion flux and perfusion heterogeneity index to dynamically and individually adjust core operating parameters of the perfusion subsystem, such as pressure and flow rate. This achieves a leap from "empirical, extensive regulation" to "refined, intelligent regulation based on real-time microcirculation data," effectively improving the safety and effectiveness of organ perfusion and laying a solid foundation for improving the viability maintenance and transplant prognosis of transported organs.

[0135] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for assessing organ condition based on laser speckle, characterized in that, Includes the following steps: Acquire raw speckle images of the transported organ in a stable phase and real-time vibration data of the organ storage device; Using a vibration-speckle distortion model, rigid displacement correction is performed on the original speckle image based on real-time vibration data to obtain a rigidly corrected speckle image. The residual non-rigid deformation field in the rigidly corrected speckle image is calculated using the optical flow method, and pixel-level inverse compensation and alignment are performed to generate a quasi-static speckle image. Calculate the microcirculation perfusion distribution map based on quasi-static speckle images; The microcirculation perfusion distribution map is divided into multiple target assessment regions, and perfusion heterogeneity indexes are calculated based on the perfusion flux of each target assessment region. A high-risk warning is issued when the infusion heterogeneity index exceeds the warning threshold.

2. The method according to claim 1, characterized in that, The quasi-static speckle image is generated through the following steps: Acquire static speckle images of transport organs under static conditions, and calculate the reference global contrast of the static speckle images; A real-time contrast image is synthesized based on the quasi-static speckle image. The real-time global contrast of the real-time contrast image is calculated. If the absolute difference between the real-time global contrast and the reference global contrast is greater than the allowable contrast deviation, the vibration-speckle distortion model is retrained using the original speckle image to make the absolute difference between the real-time global contrast and the reference global contrast less than the allowable contrast deviation. Otherwise, a quasi-static speckle image is output.

3. The method according to claim 1, characterized in that, The acquisition of the vibration-speckle distortion model includes the following steps: Acquire static speckle images of the test organ in a resting state; According to the test vibration load spectrum, a test vibration load is applied to the test organ to obtain a test dynamic speckle image; Calculate the rigid body transformation vector of the test dynamic speckle image relative to the test static speckle image; By continuously changing the amplitude and direction of the test vibration load, multiple rigid body transformation vectors are obtained. The initial vibration-speckle distortion model is trained using rigid body transformation vectors to obtain the vibration-speckle distortion model.

4. The method according to claim 3, characterized in that, The acquisition of the vibration-speckle distortion model also includes the following steps: The vibration-speckle distortion model was used to correct the displacement of the dynamic speckle image and to calculate the SSIM of the corrected dynamic speckle image and the static speckle image. The loss function is established using SSIM, and the vibration-speckle distortion model is iteratively trained. When SSIM is greater than the qualified threshold, the training ends and the vibration-speckle distortion model is output.

5. The method according to claim 1, characterized in that, The method further includes: When the perfusion flow rate of the transport organ increases, the change in perfusion heterogeneity index of the target assessment area is obtained. If the change in perfusion heterogeneity index of the target assessment area is greater than the risk threshold, a high-risk warning is output.

6. The method according to claim 1, characterized in that, The method further includes: Based on the microcirculation perfusion distribution map, the perfusion heterogeneity index and the average perfusion flux change rate of each target assessment area are calculated. Target assessment areas with perfusion heterogeneity index greater than the warning threshold or target assessment areas with average perfusion flux change rate less than the reperfusion slope threshold are marked.

7. An organ perfusion system, characterized in that, include: A perfusion subsystem for perfusing organ viability maintenance into a transport organ, the perfusion subsystem including an organ reservoir for maintaining the viability of the transport organ at body temperature; The imaging module is used to acquire static speckle images and raw speckle images of the transported organ; Vibration monitoring module is used to acquire real-time vibration data of the organ storage device; The data processing module is used to perform rigid displacement correction on the original speckle image based on real-time vibration data using the vibration-speckle distortion model to obtain a rigidly corrected speckle image. It also uses the optical flow method to calculate the residual non-rigid deformation field in the rigidly corrected speckle image, performs pixel-level inverse compensation and alignment, generates a quasi-static speckle image, calculates the microcirculation perfusion distribution map based on the quasi-static speckle image, divides the microcirculation perfusion distribution map into multiple target evaluation regions, and calculates the perfusion heterogeneity index and the overall average perfusion flux based on the perfusion flux of each target evaluation region. The perfusion system controller is used to control the operating status of the perfusion subsystem based on perfusion heterogeneity indicators and overall average perfusion flux, including perfusion pressure, pulsation waveform, and organ viability maintenance components.

8. The system according to claim 7, characterized in that, The steps for the infusion system controller to adjust the operating state of the infusion subsystem include: When the perfusion heterogeneity index of the target assessment area is less than or equal to the warning threshold and the overall average perfusion flux is lower than the theoretical perfusion flux threshold, the perfusion pressure of the perfusion subsystem shall be increased. When the perfusion heterogeneity index of a target assessment area is higher than the warning threshold, and the overall average perfusion flux is greater than or equal to the theoretical perfusion flux threshold, vasodilators are added to the organ activity maintainer, or the pulsation waveform is optimized, or the composition of the organ activity maintainer is adjusted. When the overall average perfusion flux is lower than the theoretical perfusion flux threshold and the perfusion heterogeneity index of the target assessment area is higher than the warning threshold, the perfusion pressure of the perfusion subsystem is increased, and the injection of vasodilating drugs is triggered simultaneously.

9. The system according to claim 7, characterized in that, The acquisition of the vibration-speckle distortion model includes the following steps: Acquire static speckle images of the test organ in a resting state; According to the test vibration load spectrum, a test vibration load is applied to the test organ to obtain a test dynamic speckle image; Calculate the rigid body transformation vector of the test dynamic speckle image relative to the test static speckle image; By continuously changing the amplitude and direction of the test vibration load, multiple rigid body transformation vectors are obtained. The initial vibration-speckle distortion model is trained using rigid body transformation vectors to obtain the vibration-speckle distortion model.

10. The system according to claim 7, characterized in that, The acquisition of the vibration-speckle distortion model also includes the following steps: The vibration-speckle distortion model was used to correct the displacement of the dynamic speckle image and to calculate the SSIM of the corrected dynamic speckle image and the static speckle image. The loss function is established using SSIM, and the vibration-speckle distortion model is iteratively trained. When SSIM is greater than the qualified threshold, the training ends and the vibration-speckle distortion model is output.

Images