Flap transplantation imaging method and system based on flap blood supply evaluation

Abstract

The application relates to the field of medical imaging technology and discloses a flap grafting imaging method and system based on flap blood supply evaluation. The method comprises the following steps: collecting an original spectral image, separating channels to obtain different wavelength images, constructing a blood flow dynamics feature map, extracting microcirculation perfusion and static tissue regions, identifying a blood flow signal point set, establishing a dynamic perfusion trajectory model according to space-time division, calculating a blood flow direction vector and a flow velocity estimation value of each pixel, fusing to generate a two-dimensional blood supply vector field map, superimposing the two-dimensional blood supply vector field map and a specific wavelength image into a composite image, calculating a blood supply perfusion index in real time, adjusting a scanning frequency and a light source intensity according to the blood supply perfusion index, and optimizing imaging through closed-loop feedback. Pixel-level blood flow parameters can be obtained, and acquisition conditions can be dynamically adapted, so that the fineness of flap blood supply evaluation and the imaging continuity are improved.

Description

Technical Field

[0001] This invention relates to the field of medical imaging technology, specifically to an imaging method and system for flap transplantation based on flap blood supply assessment. Background Technology

[0002] In current flap transplantation procedures, spectral imaging is often used to obtain information about the flap region to assess blood supply. Conventional methods involve acquiring spectral images, performing basic separation and display, and relying on manual methods or simple algorithms to identify blood flow regions. This lack of continuous tracking and modeling of spatiotemporal changes in blood flow signals makes it difficult to obtain the blood flow direction and velocity for each pixel. Imaging equipment often operates at fixed scanning frequencies and light source intensities, failing to sense changes in blood supply status during acquisition and make corresponding adjustments. This results in limited image quality when there are significant differences in blood supply or dynamic changes, and some perfusion information may be missed or distorted.

[0003] Conventional techniques, lacking a dynamic perfusion trajectory model based on the spatiotemporal distribution of blood flow signal points, cannot calculate pixel-level blood flow direction vectors and velocity estimates, yielding only general regional perfusion conclusions and failing to characterize the fine blood flow patterns within the flap. Furthermore, fixed parameter acquisition cannot be linked to real-time blood flow assessment, and the imaging process cannot adaptively adjust scanning frequency and light source intensity according to changes in the overall and regional blood flow perfusion index of the flap, limiting the continuous and accurate capture and presentation of blood flow status. This invention addresses the problems of needing to establish a dynamic perfusion trajectory model in the flap region to calculate pixel-level blood flow parameters and needing to optimize imaging by adjusting acquisition parameters in a closed loop based on real-time changes in the blood flow perfusion index. Summary of the Invention

[0004] The purpose of this invention is to provide an imaging method and system for flap transplantation based on flap blood supply assessment, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides an imaging method for flap transplantation based on flap blood supply assessment, the method comprising: The original spectral image data of the flap region was acquired, and the spectral channels of the original spectral image data were separated to obtain separated image data containing different characteristic wavelengths; Hemodynamic feature maps were constructed based on the separated image data, and the microcirculation perfusion area and static tissue area were extracted from the feature maps; Identify blood flow signal point sets within the microcirculation perfusion area and construct a dynamic perfusion trajectory model based on the spatiotemporal distribution of the blood flow signal point sets; The blood flow direction vector and velocity estimate for each pixel are calculated based on the dynamic perfusion trajectory model; By fusing blood flow direction vectors and velocity estimates, a two-dimensional blood flow vector field map of the flap region is generated. The two-dimensional blood flow vector field image is superimposed with a specific wavelength image from the separated image data at the pixel level to form an enhanced blood flow assessment composite image. The perfusion index of the entire flap and its zones is calculated in real time using composite images of blood supply assessment. Based on the changing trend of the blood perfusion index, the scanning frequency and light source intensity of the spectral image acquisition device are adaptively adjusted; The adjusted scanning frequency and light source intensity parameters are fed back to the image acquisition stage to form a closed-loop control and continuously optimize the flap blood supply imaging process.

[0006] Preferably, the raw spectral image data of the flap region is acquired, and spectral channel separation is performed on the raw spectral image data, including: The flap area is continuously scanned using a multispectral imaging device to acquire raw spectral image data streams containing visible and near-infrared bands. A spectral response correction matrix is ​​established to correct the spectral response non-uniformity of each pixel in the original spectral image data stream. The spectral unmixing algorithm is applied to decompose the corrected original spectral image data into independent spectral component images corresponding to oxyhemoglobin, deoxyhemoglobin and background tissue. Temporal filtering is performed on each independent spectral component image to suppress low-frequency background noise and high-frequency random noise that are unrelated to blood flow signals; The images of each independent spectral component after time-domain filtering are arranged in wavelength order to form a separated image data cube with a time dimension.

[0007] Preferably, a hemodynamic feature map is constructed based on the separated image data, and the microcirculation perfusion region and static tissue region are extracted from the feature map, including: Select specific wavelength image sequences that reflect hemodynamic changes from the separated image data cube; Calculate the rate of change of grayscale values ​​of corresponding pixels in consecutive frames of an image sequence at a specific wavelength, and generate an initial hemodynamic rate of change map; A region growing algorithm is applied to the initial hemodynamic rate of change map, and connected pixel regions with a rate of change exceeding a set threshold are marked as candidate microcirculation perfusion regions; Within the candidate microcirculation perfusion area, the periodicity and directional consistency of pixel gray value changes are further analyzed to eliminate false blood flow signal areas and confirm the final microcirculation perfusion area. The portion of the separated image data that is not marked as a microcirculation perfusion region and has a gradual change in grayscale temporal sequence is identified as a static tissue region, and a static tissue region mask is created.

[0008] Preferably, a set of blood flow signal points is identified within the microcirculation perfusion region, and a dynamic perfusion trajectory model is constructed based on the spatiotemporal distribution of the blood flow signal point set, including: In a continuous sequence of images at specific wavelengths, feature points are detected in the finally confirmed microcirculation perfusion area, and pixels with significant gray-scale changes are extracted as initial blood flow signal points. Track the position movement of the initial blood flow signal point in the image sequence to form multiple discrete blood flow signal point motion trajectory segments; The trajectory clustering algorithm is applied to merge and connect the motion trajectory segments of blood flow signal points with similar motion direction and velocity to form a complete perfusion trajectory; Each complete perfusion trajectory is smoothed and its spatiotemporal motion equation is fitted. The spatiotemporal motion equation describes the change of the position of the blood flow signal point in the imaging plane over time. By summarizing all complete perfusion trajectories and their spatiotemporal motion equations, a dynamic perfusion trajectory model describing the microcirculatory blood flow pattern of the flap is constructed.

[0009] Preferably, the blood flow direction vector and velocity estimate for each pixel are calculated based on the dynamic perfusion trajectory model, including: Based on the spatiotemporal motion equations in the dynamic perfusion trajectory model, calculate all perfusion trajectories that pass through the position of each pixel at a specified time point; For multiple perfusion trajectories passing through the same pixel, calculate the weighted average vector of their motion direction. The weighted average vector is used as the blood flow direction vector of the pixel, and the weight is determined by the signal-to-noise ratio and continuous length of the corresponding trajectory. Based on the spatial displacement and time interval between adjacent time points on the infusion trajectory, the local flow velocity represented by the infusion trajectory is calculated. The local flow velocities of all perfusion trajectories covering the same pixel are statistically fused, and the median value is taken as the flow velocity estimate of the pixel. Each pixel in the image plane is assigned a corresponding blood flow direction vector and a flow velocity estimate. If no trajectory passes through a pixel, its vector and estimate are obtained by interpolation of neighboring pixels.

[0010] Preferably, a two-dimensional blood flow vector field map of the flap region is generated by fusing blood flow direction vector and flow velocity estimation, including: Create a blank vector field layer with the same spatial resolution as the original image; The blood flow direction vector calculated for each pixel is converted into a vector graphic element with a directional arrow. The length of the vector graphic is proportional to the estimated flow velocity of the pixel. Vector graphic elements are drawn onto the corresponding pixel positions of a blank vector field layer, and color mapping is used to convert the flow rate estimate into the corresponding color intensity, forming a two-dimensional blood flow vector field map that uses hue and arrows to express blood flow information. Morphological closing operations are performed on the two-dimensional blood circulation vector field to fill in any small breaks and voids that may be caused by interpolation, ensuring the visual continuity of the vector field. The processed 2D blood circulation vector field map is Gaussian smoothed to reduce the visual abruptness of vector graphic elements, generating the final 2D blood circulation vector field map for display.

[0011] Preferably, the two-dimensional blood flow vector field image is superimposed pixel-level with a specific wavelength image from the separated image data to form an enhanced blood flow assessment composite image, including: The specific wavelength image that best reflects the tissue structure is selected from the separated image data cube as the background image; Adjust the transparency parameter of the two-dimensional blood circulation vector field map so that its vector graphic elements and color information can be displayed in a semi-transparent overlay; The two-dimensional blood flow vector field map with adjusted transparency is then alpha-blended pixel by pixel with the selected background image. In the blended image, a highlighted outline is added to the edge of the microcirculation perfusion area to enhance its visual distinction from the static tissue area; The complete image, which overlays blood supply vector information, background tissue structure, and regional contours, is output as a composite image for blood supply assessment.

[0012] Preferably, the perfusion index of the entire flap and its regions is calculated in real time using composite images of blood supply assessment, including: In the data layer corresponding to the blood flow assessment composite image, read the flow velocity estimation data of each pixel contained in the two-dimensional blood flow vector field map; Within the entire flap region of interest, the average value of the estimated flow velocity of all pixels is calculated, and this average value is used as the overall blood perfusion index of the flap. The flap area is divided into partitions according to a preset anatomical grid. Within each anatomical grid partition, the average value of the estimated flow velocity of the corresponding pixel is independently calculated and used as the blood perfusion index of the corresponding anatomical grid partition. Real-time tracking of the overall blood perfusion index of the flap and the change curves of the blood perfusion index of each zone over time; A database of blood perfusion index variation curves was established to record and analyze the long-term dynamic changes in flap blood supply.

[0013] Preferably, the scanning frequency and light source intensity of the spectral image acquisition device are adaptively adjusted according to the changing trend of the blood perfusion index, including: The system monitors the rate of change of the overall blood perfusion index of the flap in real time. When the rate of change exceeds the preset sensitivity threshold, the acquisition parameter adjustment mechanism is triggered. If the blood perfusion index shows a rapid upward trend, the scanning frequency of the multispectral imaging device is gradually increased by a preset step size to capture faster blood flow dynamics details; if it shows a rapid downward trend, the light source intensity in the near-infrared band is simultaneously enhanced to improve the detection capability of deep or weak blood flow. While adjusting the scanning frequency and light source intensity, the changes in the image signal-to-noise ratio are monitored, and feedback adjustment is used to ensure that the image quality is not lower than the set standard. The adjusted scanning frequency and light source intensity parameters are used as new acquisition parameters and configured in real time to the multispectral imaging device. The image data obtained under the new acquisition parameters is continuously used in subsequent processing, forming a closed-loop feedback from parameter adjustment to image acquisition to analysis and evaluation.

[0014] Preferably, when the processor executes the computer program, it implements the steps of the flap transplantation imaging method based on flap blood supply assessment as described in any of the above-mentioned methods.

[0015] Compared with the prior art, the beneficial effects of the present invention are: Based on the separated image data, a hemodynamic feature map is constructed, and the microcirculation perfusion region and static tissue region are extracted. Within the microcirculation perfusion region, a set of blood flow signal points is identified, and a dynamic perfusion trajectory model is built based on their spatiotemporal distribution. The blood flow direction vector and velocity estimate for each pixel are calculated. Pixel-by-pixel information on blood flow direction and velocity can be obtained within the flap region, revealing the intricate blood flow paths and intensity variations within the flap. This advances the manifestation of blood supply status from the regional level to the pixel level, enabling the identification of local perfusion differences and flow direction characteristics in complex blood supply distributions, allowing for assessment coverage of a more complete spatial structure.

[0016] A two-dimensional blood flow vector field map is generated by fusing blood flow direction vectors and velocity estimates. This map is then superimposed pixel-wise with images of a specific wavelength to form an enhanced blood flow assessment composite image. The perfusion index of the entire flap and its sub-regions is calculated in real time. The scanning frequency and light source intensity of the spectral image acquisition device are adaptively adjusted based on the index's changing trend, and the adjusted parameters are fed back to the acquisition stage to form a closed-loop control. During imaging, the acquisition conditions can be dynamically optimized according to changes in blood flow status, ensuring the device maintains effective signal acquisition capabilities at different perfusion levels. This allows the imaging to continuously match the actual blood flow dynamics of the flap, reducing signal weakening or redundant acquisition caused by fixed parameters, and ensuring the blood flow assessment composite image maintains detailed presentation synchronized with changes in blood flow. Attached Figure Description

[0017] Figure 1 This is a schematic diagram illustrating the working principle of the flap transplantation imaging method based on flap blood supply assessment described in this invention. Figure 2 This is a flowchart of spectral image data acquisition and separation; Figure 3 A flowchart for hemodynamic feature map construction and region extraction; Figure 4 Composite images for assessing flap blood supply; Figure 5 This is a diagram showing the vector field and velocity distribution characteristics of blood flow in the skin flap microcirculation. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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.

[0019] Please see Figure 1 This invention provides an imaging method for flap transplantation based on flap blood supply assessment. The method includes: acquiring raw spectral image data of the flap region and performing spectral channel separation on these raw spectral image data to obtain separated image data containing different characteristic wavelengths. A hemodynamic feature map is constructed using the separated image data, and the microcirculation perfusion region and static tissue region are extracted from the feature map. A set of blood flow signal points is identified within the microcirculation perfusion region, and a dynamic perfusion trajectory model is constructed based on the spatiotemporal distribution of this set of points. The blood flow direction vector and velocity estimate for each pixel are calculated according to the dynamic perfusion trajectory model, and these vectors and estimates are fused to generate a two-dimensional blood supply vector field map of the flap region. The two-dimensional blood supply vector field map is pixel-level superimposed with images of specific wavelengths from the separated image data to form an enhanced blood supply assessment composite image. The blood supply perfusion index of the entire flap and its sub-regions is calculated in real time using this composite image. The scanning frequency and light source intensity of the spectral image acquisition device are adaptively adjusted according to the changing trend of the blood supply perfusion index, and the adjusted parameters are fed back to the image acquisition stage to form a closed-loop control for continuous optimization of the imaging process.

[0020] In one embodiment of the present invention, see [reference] Figure 2A multispectral imaging device was used to continuously scan the flap area, acquiring a raw spectral image data stream containing visible and near-infrared bands. A spectral response correction matrix was established to correct the spectral response non-uniformity of each pixel in the raw spectral image data stream. A spectral demixing algorithm was applied to decompose the corrected raw spectral image data into independent spectral component images corresponding to oxyhemoglobin, deoxyhemoglobin, and background tissue. Temporal filtering was performed on each independent spectral component image to suppress low-frequency background noise and high-frequency random noise unrelated to blood flow signals. The temporally filtered independent spectral component images were arranged in wavelength order to form a separated image data cube with a time dimension.

[0021] In specific implementation, when acquiring raw spectral image data of the flap area and performing spectral channel separation, a multispectral imaging device is used to continuously scan the flap area to obtain a raw spectral image data stream containing visible and near-infrared bands. The multispectral imaging device acquires images at a preset frame rate and wavelength interval. The raw spectral image data stream contains multi-band images in a time series. In some embodiments, the multispectral imaging device is equipped with a tunable filter or a liquid crystal tunable filter to achieve selective transmission of specific wavelengths. Optionally, each pixel of the raw spectral image data stream contains spectral information in the range of 400 nanometers to 1000 nanometers, covering the characteristic absorption bands of oxyhemoglobin and deoxyhemoglobin.

[0022] A spectral response correction matrix is ​​established to correct the spectral response non-uniformity of each pixel in the original spectral image data stream. The spectral response correction matrix is ​​obtained by pre-calibrating the response differences of the multispectral imaging device to light of different wavelengths. The correction process is performed independently for each wavelength channel of each pixel. It can be understood that the spectral response non-uniformity correction compensates for the inconsistency in response between sensor pixels and the vignetting effect of the optical system. The application of the spectral response correction matrix is ​​achieved through the following formula: in: Indicates the corrected pixel points At wavelength The intensity value at that location, Represents the original pixel. At wavelength The intensity value at that location, Indicates wavelength The response gain coefficient at that point Indicates wavelength The offset at that point, the spectral response correction matrix contains all wavelengths. and parameter.

[0023] A spectral unmixing algorithm is applied to decompose the corrected original spectral image data into independent spectral component images corresponding to oxyhemoglobin, deoxyhemoglobin, and background tissue. The algorithm is based on a linear mixture model, assuming that the spectrum of each pixel is a linear combination of the oxyhemoglobin, deoxyhemoglobin, and background tissue spectra. In practice, the linear mixture model is implemented by first obtaining the spectral characteristic curves of oxyhemoglobin and deoxyhemoglobin through experimental measurements as known endmember spectra, and simultaneously estimating the spectral characteristic curve of the background tissue from the corrected original spectral image data through principal component analysis as a third endmember. The algorithm decomposes the spectral data of each pixel, calculates the contribution ratio of each endmember in the pixel, and thus generates the corresponding oxyhemoglobin component image, deoxyhemoglobin component image, and background tissue component image. In some embodiments, the spectral unmixing algorithm uses nonnegative matrix factorization or least squares fitting to extract the oxyhemoglobin component image, the deoxyhemoglobin component image, and the background tissue component image from the corrected original spectral image data. Optionally, the spectral characteristic curves of oxyhemoglobin and deoxyhemoglobin are obtained in advance through experimental measurements, and the background tissue spectrum is estimated from the image data through principal component analysis.

[0024] Temporal filtering is performed on each independent spectral component image to suppress low-frequency background noise and high-frequency random noise unrelated to blood flow signals. Temporal filtering is performed on the time-series data of each pixel, and a bandpass filter is used to retain signal components related to blood flow pulse frequency. It can be understood that low-frequency background noise comes from slow tissue movement or changes in ambient light, and high-frequency random noise comes from sensor thermal noise. In specific implementation, the passband range of the bandpass filter is set to 0.5 Hz to 5 Hz, corresponding to the typical microcirculatory blood flow fluctuation frequency. After temporal filtering, the noise components in each independent spectral component image are weakened, and the blood flow-related signals are enhanced. The images of each independent spectral component after time-domain filtering are arranged in wavelength order to form a separated image data cube with a time dimension. The separated image data cube contains rows and columns in the spatial dimension, wavelength channels in the spectral dimension, and frame sequences in the time dimension. In specific implementation, the separated image data cube is organized as a three-dimensional array, where the first and second dimensions represent the image height and width, the third dimension represents the wavelength channel index, and the fourth dimension represents the time point index. The separated image data cube serves as input data for subsequent steps to construct hemodynamic feature maps.

[0025] In one embodiment of the present invention, see [reference] Figure 3From the separated image data cube, a specific wavelength image sequence reflecting hemodynamic changes is selected. The grayscale value change rate of corresponding pixels in consecutive frames of the sequence is calculated to generate an initial hemodynamic change rate map. A region growing algorithm is applied to the initial hemodynamic change rate map, and connected pixel regions with change rates exceeding a set threshold are marked as candidate microcirculation perfusion regions. Within the candidate microcirculation perfusion regions, the periodicity and directional consistency of pixel grayscale value changes are further analyzed to eliminate false blood flow signal regions and confirm the final microcirculation perfusion regions. Unmarked portions of the separated image data with gradual grayscale temporal changes are identified as static tissue regions, and their masks are established. In the continuous specific wavelength image sequence, feature point detection is performed on the finally confirmed microcirculation perfusion regions, and pixels with significant grayscale changes are extracted as initial blood flow signal points. The positional movement of the initial blood flow signal points in the image sequence is tracked to form discrete motion trajectory segments. A trajectory clustering algorithm is applied to merge and connect segments with similar motion directions and velocities to form a complete perfusion trajectory. Each complete infusion trajectory is smoothed and its spatiotemporal motion equation is fitted. All trajectories and equations are then combined to form a dynamic infusion trajectory model.

[0026] In specific implementation, when constructing the hemodynamic feature map and extracting regions, a specific wavelength image sequence reflecting hemodynamic changes is selected from the separated image data cube. The selection of the specific wavelength image sequence is based on the characteristic absorption peaks of oxyhemoglobin and deoxyhemoglobin in the visible and near-infrared bands. In some embodiments, the selected wavelengths include image sequences near 660 nm and 850 nm. The grayscale value change rate of corresponding pixels in consecutive frames of the specific wavelength image sequence is calculated to generate an initial hemodynamic change rate map. The calculation uses the following formula: in: Indicates the position located in the image. Line number The pixels in the column at time points grayscale change rate This indicates that the pixel is at time point The grayscale intensity value and the initial hemodynamic rate of change map reflect how fast the relative intensity of each pixel changes over time.

[0027] A region growing algorithm is applied to the initial hemodynamic rate of change map to mark connected pixel regions whose rate of change exceeds a set threshold as candidate microcirculation perfusion regions. The seed point of the region growing algorithm is selected as the pixel point in the initial hemodynamic rate of change map whose rate of change exceeds a first set threshold. The growth criterion is that the rate of change of adjacent pixels exceeds a second set threshold and the difference between the rate of change of the adjacent pixel point and the seed point is within the tolerance range. In some embodiments, the first set threshold is a positive value and the second set threshold is lower than the first set threshold. Within the candidate microcirculation perfusion region, the periodicity and directional consistency of pixel grayscale value changes are further analyzed to eliminate pseudo-blood flow signal regions and confirm the final microcirculation perfusion region. Periodicity is analyzed by performing a Fast Fourier Transform on the time-series grayscale data of each pixel within the candidate region to find the dominant frequency peak that matches the heart rate. Directional consistency is analyzed by calculating the temporal correlation of grayscale changes between adjacent pixels. Regions that do not have significant periodicity or whose change direction is inconsistent with that of neighboring pixels are identified as pseudo-signals and removed. It can be understood that pseudo-blood flow signal regions originate from slight local tissue movement or light reflection. The parts of the separated image data that are not marked as microcirculation perfusion regions and have smooth grayscale temporal changes are identified as static tissue regions, and a static tissue region mask is established. The criterion for smooth grayscale temporal changes is that the standard deviation of the pixel over the entire time series is lower than the preset static tissue standard deviation threshold.

[0028] When identifying blood flow signal point sets and constructing a dynamic perfusion trajectory model within the microcirculation perfusion region, feature point detection is performed on the finally confirmed microcirculation perfusion region in a continuous sequence of images at specific wavelengths. Pixels with significant grayscale changes are extracted as initial blood flow signal points. The feature point detection adopts a spatiotemporal interest point detection method, which locates points with significant grayscale changes by calculating the second-order matrices of the image sequence in spatial and temporal gradients. In specific implementation, an extended form of the Harris corner detector is used in the three-dimensional spatiotemporal domain to identify initial blood flow signal points. The movement of the initial blood flow signal points in the image sequence is tracked to form multiple discrete blood flow signal point motion trajectory segments. The tracking method is based on optical flow or a matching algorithm based on feature point description, and the corresponding position of each detected initial blood flow signal point is found in subsequent frames.

[0029] A trajectory clustering algorithm is applied to merge and connect blood flow signal point motion trajectory segments with similar motion directions and velocities to form a complete perfusion trajectory. The trajectory clustering algorithm first calculates the average motion direction vector and average velocity scalar of each trajectory segment, and then uses a density-based clustering method in the direction-velocity feature space to group trajectory segments with similar features into one class. Optionally, for trajectory segments belonging to the same class, segments that are temporally adjacent and have a spatial distance less than the connection threshold will be connected into a longer trajectory. Each complete perfusion trajectory is smoothed and its spatiotemporal motion equation is fitted. The smoothing process uses moving average filtering or spline interpolation to reduce the position noise of trajectory points. The spatiotemporal motion equation uses a parameterized form to describe the law of change of the position of blood flow signal points in the imaging plane over time.

[0030] In one embodiment of the present invention, based on the spatiotemporal motion equations in the dynamic perfusion trajectory model, all perfusion trajectories passing through each pixel at a specified time point are calculated. For multiple perfusion trajectories passing through the same pixel, a weighted average vector of their motion directions is calculated as the blood flow direction vector of that pixel, with the weights determined by the signal-to-noise ratio and continuous length of the corresponding trajectory. Local flow velocities are calculated based on the spatial displacement and time interval between adjacent time points on the perfusion trajectory. The local flow velocities of all trajectories covering the same pixel are statistically fused, and the median is taken as the flow velocity estimate for that pixel. Each pixel in the image plane is assigned a corresponding blood flow direction vector and flow velocity estimate. If no trajectory passes through a pixel, its vector and estimate are obtained by interpolation from neighboring pixels. A blank vector field layer with the same spatial resolution as the original image is created, and the blood flow direction vector calculated for each pixel is converted into a vector graphic element with a directional arrow whose length is proportional to the flow velocity estimate. These vector graphic elements are drawn onto the corresponding pixel positions of the blank vector field layer, and color mapping is used to convert the flow velocity estimate into color intensity, thereby forming a two-dimensional blood flow vector field map that expresses blood flow information using hue and arrows. The vector field map is subjected to morphological closing operations to fill in small breaks and voids, and then Gaussian smoothing is performed to generate the final two-dimensional blood flow vector field map for display.

[0031] In practical implementation, when calculating the blood flow direction vector and velocity estimate based on the dynamic perfusion trajectory model, according to the spatiotemporal motion equation in the dynamic perfusion trajectory model, all perfusion trajectories passing through each pixel position at a specified time point are calculated. The spatiotemporal motion equation is given in parametric form. By solving the equation, it can be determined which perfusion trajectories pass through the given pixel coordinates at a specific time. In some embodiments, the specified time point is usually selected as the middle time of the image sequence or a specific time required for evaluation. For multiple perfusion trajectories passing through the same pixel, the weighted average vector of their motion direction is calculated as the blood flow direction vector of that pixel. The weight is determined by the signal-to-noise ratio and continuous length of the corresponding trajectory. The calculation formula is: in: This represents the weighted average direction vector. This indicates the total number of infusion trajectories passing through this pixel. Indicates the first The unit direction vector of the infusion trajectory when it passes through the pixel at a specified time point. Indicates the first The weights of the perfusion trajectories, weights It is defined as the product of the signal-to-noise ratio of the infusion trajectory and the length of continuous tracking on its time axis.

[0032] The local velocity represented by the perfusion trajectory is calculated based on the spatial displacement and time interval between adjacent time points on the perfusion trajectory. For a perfusion trajectory, the two closest trajectory points before and after it passes through the target pixel are selected. The instantaneous velocity is calculated based on the spatial coordinate difference and time difference between these two points. The local velocities of all perfusion trajectories covering the same pixel are statistically fused and the median value is taken as the velocity estimate of the pixel. The statistical fusion process collects the local velocity values ​​calculated from all relevant perfusion trajectories. After arranging these values ​​in ascending order, the value in the middle position is selected as the final velocity estimate of the pixel. Each pixel in the image plane is assigned a corresponding blood flow direction vector and velocity estimate. If no trajectory passes through the pixel, its vector and estimate are obtained by interpolation from neighboring pixels. The interpolation method adopts the inverse distance weighting method based on distance. The blood flow direction vector and velocity estimate are obtained from pixels with trajectories passing through a certain neighborhood around the pixel and weighted for calculation.

[0033] When generating a two-dimensional blood circulation vector field map by fusing blood flow direction vectors and velocity estimates, a blank vector field layer with the same spatial resolution as the original image is created. The blank vector field layer is a data matrix with the same height and width as the input image, used to store the visualization elements corresponding to the blood flow direction vector and velocity estimate at each location. The blood flow direction vector calculated for each pixel is converted into a vector graphic element with a directional arrow whose length is proportional to the velocity estimate. The conversion process maps the velocity estimate to the physical display length of the arrow graphic according to a preset mapping scale, while the arrow pointing in the same direction angle as the blood flow direction vector. These vector graphic elements are drawn to the corresponding pixel positions of the blank vector field layer. At the same time, color mapping is used to convert the velocity estimate into the corresponding color intensity. The color mapping scheme linearly maps the velocity estimate range to a color gradient bar, such as from blue to red, forming a two-dimensional blood circulation vector field map that expresses blood circulation information through hue and arrows. Morphological closing operations are performed on the two-dimensional blood circulation vector field image to fill small breaks and holes. The morphological closing operation uses a structuring element to first dilate and then erode the corresponding binary skeleton image of the vector field image. It can be understood that small breaks and holes are caused by sparse trajectory points or discontinuous interpolation. The closing operation can connect adjacent breakpoints and fill small holes. Then, Gaussian smoothing is performed to generate the final two-dimensional blood circulation vector field image for display. Gaussian smoothing applies low-pass filtering to the coordinate position and color intensity of the arrow graphics in the vector field image to reduce the visual abruptness of vector graphic elements caused by single trajectory noise or interpolation abrupt changes. The size and standard deviation of the Gaussian smoothing kernel are set according to the image resolution and display requirements.

[0034] In one embodiment of the present invention, a specific wavelength image that best reflects the tissue structure is selected from the separated image data cube as a background image. The transparency parameter of the two-dimensional blood circulation vector field image is adjusted to allow it to be superimposed semi-transparently. The two-dimensional blood circulation vector field image with adjusted transparency is then alpha-blended with the selected background image at the pixel level. Highlighted outlines are added to the edges of the microcirculation perfusion area in the blended image to enhance visual differentiation. A complete image superimposed with blood circulation vector information, background tissue structure, and area outlines is output as a blood circulation assessment composite image. The flow velocity estimation data of each pixel in the two-dimensional blood circulation vector field image is read from the data layer corresponding to the blood circulation assessment composite image. The average flow velocity estimation of all pixels within the entire flap region of interest is calculated as the blood circulation perfusion index of the flap as a whole. The flap region is divided into partitions according to a preset anatomical grid. The average flow velocity estimation of the corresponding pixels in each partition is independently calculated as the blood circulation perfusion index of that partition. The blood circulation perfusion index of the flap as a whole and each partition is tracked in real time over time, and a blood circulation perfusion index change curve database is established to record and analyze long-term dynamic changes.

[0035] In specific implementation, when forming an enhanced blood supply assessment composite image, a specific wavelength image that best reflects the tissue structure is selected from the separated image data cube as the background image. The selection criteria are based on image contrast and tissue detail clarity. In some embodiments, wavelength images with weak absorption (such as hemoglobin) and significant tissue scattering characteristics, such as images around 540 nm or 570 nm, are selected as the background image. The transparency parameter of the two-dimensional blood supply vector field image is adjusted so that its vector graphic elements and color information can be displayed semi-transparently. The alpha value of the transparency parameter is set between 0.5 and 0.7 to achieve clear fusion of vector information and background image. The two-dimensional blood supply vector field image with adjusted transparency is then subjected to pixel-level alpha mixing with the selected background image. The alpha mixing operation synthesizes the color value of each pixel according to the formula: in: This indicates the color value of that pixel in the output composite image. This represents the color value of the corresponding pixel in the two-dimensional blood circulation vector field image. This represents the color value of the corresponding pixel in the background image. This indicates the preset transparency parameter.

[0036] In the blended image, a highlighted outline is added to the edge of the microcirculation perfusion area to enhance its visual distinction from the static tissue area. The highlighted outline is drawn by extracting the boundary of the microcirculation perfusion area mask and using lines with a set color and width. The output is a complete image superimposed with blood supply vector information, background tissue structure, and area outline as a blood supply assessment composite image. The blood supply assessment composite image is stored in a standard image format and displayed in real time on the monitoring interface. When calculating the blood supply perfusion index of the flap as a whole and its sub-regions in real time using the blood supply assessment composite image, the velocity estimation data of each pixel in the two-dimensional blood supply vector field map contained in the data layer corresponding to the blood supply assessment composite image is read. The velocity estimation data corresponds one-to-one with the pixel position of the composite image and is directly obtained from the underlying data array of the generated vector field map. The average value of the velocity estimation of all pixels in the entire flap region within the area of ​​interest is calculated and used as the blood supply perfusion index of the flap as a whole. The area of ​​interest is defined by the flap outline region delineated in advance on the image. Pixels in the static tissue area are excluded during the calculation process.

[0037] The flap region is divided into sub-regions according to a predefined anatomical grid. Within each sub-region, the average velocity estimate of pixels is independently calculated as the perfusion index for that sub-region. The predefined anatomical grid is based on the vascular anatomy of the flap, such as dividing the flap into multiple rectangular or arbitrarily shaped regions like proximal, mid-segment, and distal ends. The perfusion index of each sub-region is calculated and recorded independently. The overall perfusion index of the flap and the perfusion index of each sub-region are tracked in real time over time. The tracking process involves sampling the perfusion index value at fixed time intervals, such as every 10 seconds or every minute, and recording the timestamp along with the index value. A perfusion index change curve database is established to record and analyze the long-term dynamic changes in flap blood flow. The database stores the time point of each calculation, the overall perfusion index, and the perfusion index values ​​of each sub-region in tabular form. See Table 1, which shows the structure and content of a fragment of the perfusion index change curve database.

[0038] Table 1: Database of Blood Perfusion Index Variation Curves See Figure 4 The image uses a specific wavelength image of the flap region (such as an image around 540nm or 570nm that shows the tissue structure) as the background image, and pixel-level overlays with images that have been transparently adjusted. A two-dimensional blood flow vector field map (adjusted to a value of 0.5-0.7) is generated, and alpha mixing is used to fuse vector information with the background. Simultaneously, the boundaries of the microcirculation perfusion region are marked with highlighted outlines in the image. These boundaries are extracted and drawn from the boundaries of the microcirculation perfusion region mask, clearly distinguishing the perfusion region from the static tissue region. The color distribution of the image (e.g., pink-yellow gradient) corresponds to the spatial differences in flow velocity estimation. Combined with the pixel coordinate system, this visually presents the blood flow distribution pattern and perfusion range of the flap microcirculation, providing a visual basis for subsequent calculation of the blood flow perfusion index.

[0039] In one embodiment of the invention, the rate of change of the overall blood perfusion index of the flap is monitored in real time. When the rate of change exceeds a preset sensitivity threshold, a parameter adjustment mechanism is triggered. If the blood perfusion index shows a rapid upward trend, the scanning frequency of the multispectral imaging device is gradually increased by a preset step size to capture faster blood flow dynamics. If it shows a rapid downward trend, the intensity of the near-infrared light source is simultaneously enhanced to improve the detection capability of deep or weak blood flow. While adjusting the scanning frequency and light source intensity, the change in image signal-to-noise ratio is monitored, and feedback adjustment is used to ensure that the image quality is not lower than the set standard. The adjusted scanning frequency and light source intensity parameters are configured as new acquisition parameters in real time into the multispectral imaging device. The image data obtained under the new acquisition parameters is continuously used in subsequent processing, forming a closed-loop feedback from parameter adjustment to image acquisition to analysis and evaluation.

[0040] In practice, when adaptively adjusting parameters based on the changing trend of the perfusion index, the rate of change of the overall perfusion index of the flap is monitored in real time. When the rate of change exceeds a preset sensitivity threshold, the parameter adjustment mechanism is triggered. The rate of change of the overall perfusion index of the flap is obtained by calculating the relative change between the current perfusion index value and the previous perfusion index value. The calculation formula is: in: Indicates relative change. This represents the overall blood perfusion index of the flap calculated at the current moment. This represents the overall blood perfusion index of the flap at the previous monitoring time. The sensitivity threshold is set to an absolute value according to clinical monitoring needs. For example, an increase or decrease of more than 5% is defined as a rapid change and triggers an adjustment.

[0041] If the blood perfusion index shows a rapid upward trend, the scanning frequency of the multispectral imaging equipment is gradually increased by a preset step size to capture faster blood flow dynamics. The rapid upward trend is determined by the rate of change. If the value remains positive and its absolute value exceeds the sensitivity threshold, the scanning frequency is increased in Hertz, for example, gradually increasing from an initial 10 frames per second to 15 or 20 frames per second. If the perfusion index shows a rapid downward trend, the intensity of the near-infrared light source is simultaneously increased to improve the detection capability of deep or weak blood flow. The rapid downward trend is judged by the rate of change. If the signal intensity remains negative and its absolute value exceeds the sensitivity threshold, the enhancement of the near-infrared light source intensity is achieved by controlling the driving current or voltage of the illumination unit within the multispectral imaging device. While adjusting the scanning frequency and light source intensity, the changes in the image signal-to-noise ratio (SNR) are monitored. Feedback adjustment ensures that the image quality is not lower than the set standard. The image SNR is evaluated by calculating the ratio of the mean to the standard deviation of the signal intensity in the microcirculation perfusion region of the image sequence. The feedback adjustment mechanism pauses further increases in the scanning frequency or moderately adjusts the magnitude of the light source enhancement when the image SNR falls below the set standard. It is understood that an excessively high scanning frequency leads to insufficient exposure time for a single frame, thus reducing the SNR, while an excessively strong light source introduces additional thermal noise or causes light saturation on the tissue surface. The adjusted scanning frequency and light source intensity parameters are configured as new acquisition parameters in real time into the multispectral imaging device. The configuration process uses software instructions to control the hardware driver of the imaging device to update its operating mode and illumination output. In some embodiments, the closed-loop feedback system has a minimum stable time window, which pauses the judgment of the rate of change for a fixed period of time after each parameter adjustment to prevent the system from oscillating repeatedly near the critical point. Optionally, the adjustment history of all collected parameters, together with the corresponding blood perfusion index value, is recorded in the log for traceability analysis.

[0042] See Figure 5 This image illustrates the vector field and velocity distribution characteristics of blood flow in the flap's microcirculation during rapid ascent. Specifically, a two-dimensional coordinate system is constructed using pixels as units. Color mapping (from light to dark corresponding to relative velocities of 0 to 6) visually represents the differences in blood flow velocity in different regions, while arrow vectors indicate the direction of blood flow at each pixel. The perfusion index rises rapidly; red and dark purplish-red areas correspond to higher velocities (relative values ​​mostly in the 4-6 range), while blue areas correspond to lower velocities (relative values ​​mostly in the 1-2 range). The distribution of arrow directions reflects the dynamic direction of blood flow within the flap's microcirculation, indicating differences in perfusion status in different areas of the flap during this stage. This visualization of vector field and velocity distribution is generated by fusing blood flow direction vectors and velocity estimates calculated based on a dynamic perfusion trajectory model, and can be used to assist in assessing the perfusion status of the flap during the rapid ascent phase.

[0043] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0044] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An imaging method for flap transplantation based on flap blood supply assessment, characterized in that, Includes the following steps: The original spectral image data of the flap region was acquired, and the spectral channels of the original spectral image data were separated to obtain separated image data containing different characteristic wavelengths; Hemodynamic feature maps were constructed based on the separated image data, and the microcirculation perfusion area and static tissue area were extracted from the feature maps; Identify blood flow signal point sets within the microcirculation perfusion area and construct a dynamic perfusion trajectory model based on the spatiotemporal distribution of the blood flow signal point sets; The blood flow direction vector and velocity estimate for each pixel are calculated based on the dynamic perfusion trajectory model; By fusing blood flow direction vectors and velocity estimates, a two-dimensional blood flow vector field map of the flap region is generated. The two-dimensional blood flow vector field image is superimposed with a specific wavelength image from the separated image data at the pixel level to form an enhanced blood flow assessment composite image. The perfusion index of the entire flap and its zones is calculated in real time using composite images of blood supply assessment. Based on the changing trend of the blood perfusion index, the scanning frequency and light source intensity of the spectral image acquisition device are adaptively adjusted; The adjusted scanning frequency and light source intensity parameters are fed back to the image acquisition stage to form a closed-loop control and continuously optimize the flap blood supply imaging process.

2. The imaging method for flap transplantation based on flap blood supply assessment according to claim 1, characterized in that, Raw spectral image data of the flap region was acquired, and spectral channel separation was performed on the raw spectral image data, including: The flap area is continuously scanned using a multispectral imaging device to acquire raw spectral image data streams containing visible and near-infrared bands. A spectral response correction matrix is ​​established to correct the spectral response non-uniformity of each pixel in the original spectral image data stream. The spectral unmixing algorithm is applied to decompose the corrected original spectral image data into independent spectral component images corresponding to oxyhemoglobin, deoxyhemoglobin and background tissue. Temporal filtering is performed on each independent spectral component image to suppress low-frequency background noise and high-frequency random noise that are unrelated to blood flow signals; The images of each independent spectral component after time-domain filtering are arranged in wavelength order to form a separated image data cube with a time dimension.

3. The imaging method for flap transplantation based on flap blood supply assessment according to claim 2, characterized in that, Hemodynamic feature maps were constructed based on the separated image data, and the microcirculation perfusion region and static tissue region were extracted from the feature maps, including: Select specific wavelength image sequences that reflect hemodynamic changes from the separated image data cube; Calculate the rate of change of grayscale values ​​of corresponding pixels in consecutive frames of an image sequence at a specific wavelength, and generate an initial hemodynamic rate of change map; A region growing algorithm is applied to the initial hemodynamic rate of change map, and connected pixel regions with a rate of change exceeding a set threshold are marked as candidate microcirculation perfusion regions; Within the candidate microcirculation perfusion area, the periodicity and directional consistency of pixel gray value changes are further analyzed to eliminate false blood flow signal areas and confirm the final microcirculation perfusion area. The portion of the separated image data that is not marked as a microcirculation perfusion region and has a gradual change in grayscale temporal sequence is identified as a static tissue region, and a static tissue region mask is created.

4. The imaging method for flap transplantation based on flap blood supply assessment according to claim 3, characterized in that, A set of blood flow signal points is identified within the microcirculation perfusion region, and a dynamic perfusion trajectory model is constructed based on the spatiotemporal distribution of the blood flow signal point set, including: In a continuous sequence of images at specific wavelengths, feature points are detected in the finally confirmed microcirculation perfusion area, and pixels with significant gray-scale changes are extracted as initial blood flow signal points. Track the position movement of the initial blood flow signal point in the image sequence to form multiple discrete blood flow signal point motion trajectory segments; The trajectory clustering algorithm is applied to merge and connect the motion trajectory segments of blood flow signal points with similar motion direction and velocity to form a complete perfusion trajectory; Each complete perfusion trajectory is smoothed and its spatiotemporal motion equation is fitted. The spatiotemporal motion equation describes the change of the position of the blood flow signal point in the imaging plane over time. By summarizing all complete perfusion trajectories and their spatiotemporal motion equations, a dynamic perfusion trajectory model describing the microcirculatory blood flow pattern of the flap is constructed.

5. The imaging method for flap transplantation based on flap blood supply assessment according to claim 4, characterized in that, The blood flow direction vector and velocity estimate for each pixel are calculated based on the dynamic perfusion trajectory model, including: Based on the spatiotemporal motion equations in the dynamic infusion trajectory model, calculate all infusion trajectories that pass through the position of each pixel at a specified time point; For multiple perfusion trajectories passing through the same pixel, calculate the weighted average vector of their motion direction. The weighted average vector is used as the blood flow direction vector of the pixel, and the weight is determined by the signal-to-noise ratio and continuous length of the corresponding trajectory. Based on the spatial displacement and time interval between adjacent time points on the infusion trajectory, the local flow velocity represented by the infusion trajectory is calculated. The local flow velocities of all perfusion trajectories covering the same pixel are statistically fused, and the median value is taken as the flow velocity estimate of the pixel. Each pixel in the image plane is assigned a corresponding blood flow direction vector and a flow velocity estimate. If no trajectory passes through a pixel, its vector and estimate are obtained by interpolation of neighboring pixels.

6. The imaging method for flap transplantation based on flap blood supply assessment according to claim 5, characterized in that, By fusing blood flow direction vectors and velocity estimates, a two-dimensional blood flow vector field map of the flap region is generated, including: Create a blank vector field layer with the same spatial resolution as the original image; The blood flow direction vector calculated for each pixel is converted into a vector graphic element with a directional arrow. The length of the vector graphic is proportional to the estimated flow velocity of the pixel. Vector graphic elements are drawn onto the corresponding pixel positions of a blank vector field layer, and color mapping is used to convert the flow rate estimate into the corresponding color intensity, forming a two-dimensional blood flow vector field map that uses hue and arrows to express blood flow information. Morphological closing operations are performed on the two-dimensional blood circulation vector field to fill in any small breaks and voids that may be caused by interpolation, ensuring the visual continuity of the vector field. The processed 2D blood circulation vector field map is Gaussian smoothed to reduce the visual abruptness of vector graphic elements, generating the final 2D blood circulation vector field map for display.

7. The imaging method for flap transplantation based on flap blood supply assessment according to claim 6, characterized in that, The two-dimensional blood flow vector field image is superimposed pixel-level with a specific wavelength image from the separated image data to form an enhanced blood flow assessment composite image, including: The specific wavelength image that best reflects the tissue structure is selected from the separated image data cube as the background image; Adjust the transparency parameter of the two-dimensional blood circulation vector field map so that its vector graphic elements and color information can be displayed in a semi-transparent overlay; The two-dimensional blood flow vector field map with adjusted transparency is then alpha-blended pixel by pixel with the selected background image. In the blended image, a highlighted outline is added to the edge of the microcirculation perfusion area to enhance its visual distinction from the static tissue area; The complete image, which overlays blood supply vector information, background tissue structure, and regional contours, is output as a composite image for blood supply assessment.

8. The imaging method for flap transplantation based on flap blood supply assessment according to claim 7, characterized in that, The perfusion index of the entire flap and its sub-flap is calculated in real time using composite images of blood supply assessment, including: In the data layer corresponding to the blood flow assessment composite image, read the flow velocity estimation data of each pixel contained in the two-dimensional blood flow vector field map; Within the entire flap region of interest, the average value of the estimated flow velocity of all pixels is calculated, and this average value is used as the overall blood perfusion index of the flap. The flap area is divided into partitions according to a preset anatomical grid. Within each anatomical grid partition, the average value of the estimated flow velocity of the corresponding pixel is independently calculated and used as the blood perfusion index of the corresponding anatomical grid partition. Real-time tracking of the overall blood perfusion index of the flap and the change curves of the blood perfusion index of each zone over time; A database of blood perfusion index variation curves was established to record and analyze the long-term dynamic changes in flap blood supply.

9. The imaging method for flap transplantation based on flap blood supply assessment according to claim 8, characterized in that, Based on the changing trend of the blood perfusion index, the scanning frequency and light source intensity of the spectral image acquisition equipment are adaptively adjusted, including: The system monitors the rate of change of the overall blood perfusion index of the flap in real time. When the rate of change exceeds the preset sensitivity threshold, the acquisition parameter adjustment mechanism is triggered. If the blood perfusion index shows a rapid upward trend, the scanning frequency of the multispectral imaging device is gradually increased by a preset step size to capture faster blood flow dynamics details; if it shows a rapid downward trend, the light source intensity in the near-infrared band is simultaneously enhanced to improve the detection capability of deep or weak blood flow. While adjusting the scanning frequency and light source intensity, the changes in the image signal-to-noise ratio are monitored, and feedback adjustment is used to ensure that the image quality is not lower than the set standard. The adjusted scanning frequency and light source intensity parameters are used as new acquisition parameters and configured in real time to the multispectral imaging device. The image data obtained under the new acquisition parameters is continuously used in subsequent processing, forming a closed-loop feedback from parameter adjustment to image acquisition to analysis and evaluation.

10. An imaging system for flap transplantation based on flap blood supply assessment, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the flap transplantation imaging method based on flap blood supply assessment as described in any one of claims 1 to 9.

Images