A method for automatic reconstruction of tissue perfusion in microsurgery based on real-time blood flow imaging

Abstract

The application discloses a kind of based on real-time blood flow imaging's microsurgery in-situ tissue perfusion automatic reconstruction method, through the two light guide beam channels of intraoperative microscope and RGB-IR camera, using near-infrared light and white light alternately illuminates, acquires original laser speckle image (camera near-infrared channel) and RGB color image under white light irradiation.In the original laser speckle image is constructed multi-angle scattering random matrix, and the second order central moment of single and multiple scattering light intensity is calculated respectively, and the blood flow information of surface and deep tissue in-situ is obtained by Monte Carlo simulation, to produce optimization coefficient c;Utilize RGB color image to calculate tissue surface blood oxygen saturation information, combine blood flow velocity and blood oxygen saturation to calculate tissue area perfusion quantity in real time.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of optical imaging, in particular to a method for automatic reconstruction of tissue perfusion in microsurgery based on real-time blood flow imaging. BACKGROUND

[0002] In modern medicine, the surgical microscope is essential for surgeons. By selecting different magnification, it can provide a magnified view of the surgical field, so that the surgeon can even operate on structures that are almost invisible to the naked eye, especially for neurosurgery, to handle fine structures such as capillaries or nerves. At the same time, since the light source is directly incorporated into the surgical microscope, the surgical microscope can also provide good illumination of the surgical field.

[0003] The existing technology and commonly used surgical microscope equipment mainly provide morphological observation assistance for surgeons, and cannot reflect the functional information such as blood flow velocity in blood vessels and blood flow perfusion of tissues in real time. These additional information has very important significance for neurosurgery type operations. For example, in the intracranial aneurysm clipping and cerebral vascular malformation resection, real-time monitoring of cerebral blood flow is beneficial to the surgeon to dynamically grasp the degree of abnormal blood flow blockage, while avoiding the misclipping of normal blood vessels; during the cerebral vascular bypass surgery, real-time monitoring of cerebral cortical blood flow can also help to evaluate the patency of the bridge blood vessels and whether the cerebral blood flow has returned to normal level.

[0004] In the aspect of blood flow imaging, laser speckle imaging (LSI) technology, as a new blood vessel and blood flow optical imaging method, can obtain visualized and quantitative real-time tissue blood flow perfusion images by analyzing the "speckle" caused by the coherent superposition of diffuse reflected laser light through different propagation paths. It is a non-invasive, non-contact, contrast agent-free, high temporal and spatial resolution two-dimensional full-field blood flow imaging method. In biomedical applications, LSI technology has been used to study the surface blood flow characteristics of skin, retina, optic nerve and mesentery, etc. In addition, since the real-time two-dimensional distribution map of cerebral cortical blood flow can be easily obtained, this technology is also very suitable for studying the vascular network and blood flow distribution of the cerebral cortex under different physiological and pathological conditions.

[0005] However, the existing system only considers the blood flow velocity in the observation area, and ignores the change information of blood oxygen saturation, so that the change of biological information in the observation area cannot be more comprehensively and specifically detected. SUMMARY

[0006] Invention purposes: The purpose of the present application is to provide a real-time blood flow imaging-based automatic reconstruction method for microsurgery tissue perfusion, which uses near-infrared light and white light to illuminate alternately through two light guide channels of an intraoperative microscope and an RGB-IR camera, and collects original laser speckle images (camera near-infrared channel) and RGB color images under white light illumination. A multi-angle scattering random matrix is constructed using the original laser speckle images, and the second-order central moments of single and multiple scattering light intensities are calculated respectively, and the blood flow information of the surface and deep tissues during the operation is obtained through Monte Carlo simulation to generate an optimization coefficient c; the tissue surface blood oxygen saturation information is calculated using the RGB color image, and the tissue region perfusion is calculated in real time by combining the blood flow velocity and the blood oxygen saturation.

[0007] Technical scheme: The real-time blood flow imaging-based automatic reconstruction method for microsurgery tissue perfusion comprises the following steps:

[0008] Step S1, the illumination and image acquisition method is: the laser light source generated by the 780-850nm laser and the white light are connected to the two light guide beams of the intraoperative microscope, and each light guide beam is controlled to be illuminated alternately, and the frequency of the illumination is set to be able to analyze the blood sample change and the blood flow velocity in a continuous small time window, and a certain frequency is maintained to meet the real-time requirement. The near-infrared (IR) channel of the camera (RGB-IR camera) is used to synchronously and continuously collect the original speckle images, and the (RGB) channel of the camera is used to collect the color RGB images under white light.

[0009] Step S2, the color RGB image is separated and linearly combined in three channels to calculate the starting time blood oxygen saturation:

[0010] Firstly, the RGB color image is decomposed into three channels of gray scale images, and the difference between the absorption coefficients of the specific wavelengths of light between the oxygenated hemoglobin and the non-oxygenated hemoglobin in the blood is used to linearly combine the gray scale values of the RGB three channels to obtain the PPG signal;

[0011] Secondly, the blood oxygen saturation at the starting stage in a certain time period is obtained by using the PPG signal.

[0012] Step S3, the covariance matrix M H in the step S3 is constructed as follows:

[0013]

[0014] In the above formula, The matrix S H is of the same size, The element value in S H is the mean value of the row where the corresponding position element in the matrix is located, is The transpose of the matrix.

[0015] Step S4, calculate the contrast K of the spatial neighborhood of the pixel gray value of the region of interest in the collected N=n original speckle images st As follows

[0016]

[0017] Where N is the number of collected original speckle image frames, Ns represents the width of the spatial neighborhood, p represents the serial number of the pixel, I p represents the gray value of the pth pixel in the spatial neighborhood of the same position in N frames of original speckle images, is the average value of the gray values of the pixels.

[0018] Step S5, calculate the blood flow velocity of the corresponding biological tissue at the position using the contrast in the spatial neighborhood:

[0019]

[0020] Step S6, combine the time contrast algorithm and the spatial contrast algorithm in a fixed time period using the RPE algorithm to improve the time resolution and reliability, the RPE theory considers the light scattering phenomenon in a sufficiently short camera exposure time [t0, t1] as a continuous random process, and Y(t) represents the random variable of the light intensity integral at time t, so it can be considered that the movement speed of the scattering particles in the imaging area remains unchanged in this sufficiently short time. For N consecutive speckle images, the light intensity standard deviation at a certain pixel point is calculated using a 3x3 time window, and the contrast value can be represented as:

[0021]

[0022] Where t0 represents the starting time of the exposure time [t0, t1], σ rpe is the estimated value of the standard deviation of the random process, μ rpe is the estimated value of the mean of the random process, and are the theoretical values of the mean and standard deviation of the random process, k y is the theoretical value of the contrast value.

[0023] Step S7, use simulation to obtain the correction parameter c of the blood flow velocity:

[0024] Use Monte Carlo simulation to simulate the component size of laser speckle at different depths, and obtain the proportion of surface blood flow velocity to the whole, i.e. the correction parameter c of the blood flow velocity, and calculate the corrected blood flow velocity v' by the correction parameter c, which is the surface blood flow velocity

[0025] ​Since the penetration depth of infrared laser in tissue is greater than that of white light in tissue, the blood flow velocity and blood oxygen saturation changes obtained are not from the same depth position. In order to correct this error, the component size of laser speckle at different depths is simulated by using Monte Carlo simulation, and the proportion of surface blood flow velocity to the whole is obtained, and the correction parameter c of blood flow velocity is obtained.

[0026] Step S8, repeating step S2, calculating the end time blood oxygen saturation.

[0027] Step S9, combining the blood oxygen saturation change value and blood flow velocity in the time period to calculate the perfusion quantity, since the change of blood oxygen saturation in the time period is considered, the calculation of perfusion quantity is optimized. BRIEF DESCRIPTION OF DRAWINGS

[0028] Figure 1 is a flowchart of an endoscopic intraoperative blood flow imaging method based on a multi-angle scattering random matrix of the present application; DETAILED DESCRIPTION

[0029] The specific embodiments of the present application will be further described in detail below in combination with the drawings and examples, and the following examples are used to illustrate the present application, but not to limit the scope of the present application.

[0030] Example 1:

[0031] As shown in the drawings, Figure 1 The present application proposes an endoscopic intraoperative blood flow imaging method based on a multi-angle scattering random matrix, which specifically includes the following steps:

[0032] Step S1, the illumination and image acquisition method is: the laser light source generated by the 830nm laser is connected to the two light guides of the intraoperative microscope, and each light guide is controlled to be turned on in turn, and the frequency of turning on is set to be able to analyze the blood sample change and blood flow velocity in a continuous small time window, and to maintain a certain frequency to meet the real-time requirement. The near-infrared (IR) channel of the camera (RGB-IR camera) is used to synchronously and continuously acquire the original speckle image, and the (RGB) channel of the camera is used to acquire the color RGB image under white light.

[0033] Step S2, separating and linearly combining the three channels of the color RGB image to calculate the start time blood oxygen saturation:

[0034] First, the RGB color image is decomposed into three channels of gray scale image, and the difference between the absorption coefficients of specific wavelength light between oxygenated hemoglobin and non-oxygenated hemoglobin in blood is used to linearly combine the gray scale values of the RGB three channels to obtain the PPG signal;

[0035] Secondly, the blood oxygen saturation at the beginning stage in a certain time period is obtained by using the PPG signal.

[0036] Step S3, in which a covariance matrix M is constructed H The formula is as follows:

[0037]

[0038] In the above formula, The matrix S H is of the same size, The element value in S H is the mean value of the row in which the element in the corresponding position of the matrix S is The transpose of the matrix.

[0039] Step S4, the contrast K in the spatial neighborhood is calculated for the pixel gray value of the region of interest in the collected N=n original speckle images st As follows

[0040]

[0041] Where N is the number of frames of the collected original speckle images, Ns represents the width of the spatial neighborhood, p represents the serial number of the pixel, I p represents the average gray value of the pth pixel in the spatial neighborhood of the same position in N frames of original speckle images, is the average of the gray values of the pixels.

[0042] Step S5, the blood flow velocity of the corresponding biological tissue at the place is calculated by using the contrast in the spatial neighborhood:

[0043]

[0044] Step S6, the time contrast algorithm and the spatial contrast algorithm in a fixed time period are combined by using the RPE algorithm to improve the time resolution and reliability, the RPE theory regards the light scattering phenomenon in a sufficiently short camera exposure time [t0, t1] as a continuous random process, Y(t) represents the random variable of the light intensity integral at time t, then it can be considered that the movement speed of the scattering particles in the imaging area remains unchanged in this sufficiently short time. For N consecutive speckle images, the light intensity standard deviation at a certain pixel point is calculated by using a 3x3 time window, and the contrast value can be represented as:

[0045]

[0046] Where t0 represents the starting time of the exposure time [t0, t1], σ rpeis an estimate of the standard deviation of the random process, μ rpe is an estimate of the mean of the random process, and are the theoretical values of the mean and standard deviation of the random process, k y is the theoretical value of the contrast value.

[0047] Step S7, using simulation to obtain the parameter c of the corrected blood flow velocity:

[0048] Since the penetration depth of infrared laser in the tissue is greater than the penetration depth of white light in the tissue, the obtained blood flow velocity and blood oxygen saturation change do not come from the same depth position. In order to correct this error, Monte Carlo simulation is used to simulate the component size of laser speckle at different depths, and the proportion of surface blood flow velocity in the whole is obtained, and the correction parameter c of the blood flow velocity is obtained.

[0049] Step S8, repeat step S2 to calculate the end time blood oxygen saturation.

[0050] Step S9, combine the blood oxygen saturation change value and the blood flow velocity in the time period to calculate the perfusion quantity. Since the change of the blood oxygen saturation in the time period is considered, the calculation of the perfusion quantity is optimized.

[0051] Finally, the method of the present application is only a preferred embodiment, and is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for automatic reconstruction of tissue perfusion during microsurgery based on real-time blood flow imaging, characterized in that, It comprises the following steps: Step S1, the laser light source generated by the 780-850nm laser is connected to the two light beams of the intraoperative microscope, and the white light is connected to the two light beams of the intraoperative microscope, and the light beams are controlled to pass through the light beams in turn, the near-infrared channel of the RGB-IR camera is used to synchronously and continuously collect the original speckle images, and the RGB channel of the RGB-IR camera is used to collect the color RGB images under the white light; Step S2, the three channels of the color RGB image are separated and linearly combined to calculate the starting time blood oxygen saturation; Step S3, for each pixel position of the original speckle image, an initial scattering light intensity matrix is constructed; Step S4, the contrast in the spatial neighborhood is calculated for the pixel gray value of the region of interest in the N=n collected original speckle images; Step S5, the blood flow velocity of the corresponding biological tissue at the position is calculated by using the contrast in the spatial neighborhood; Step S6, the RPE algorithm is used to combine the time contrast algorithm and the spatial contrast algorithm in a fixed time period to improve the time resolution and reliability; Step S7, the result obtained by simulation is used to correct the blood flow velocity; Step S8, step S2 is repeated to calculate the ending time blood oxygen saturation; Step S9, the blood perfusion is calculated by combining the blood oxygen saturation change value and the blood flow velocity in the time period.

2. The method of claim 1, wherein the method further comprises: The frequency of the light passing in step S1 is set to be able to analyze the blood sample change and blood flow velocity in a continuous small time window, and to maintain a certain frequency to meet the real-time requirement; the frequency range of the light passing is set to be 5-20Hz.

3. The method of claim 1, wherein: In step S2, First, the RGB color image is decomposed into three channels of gray scale images, and the difference between the absorption coefficients of the specific wavelength of light between the oxygenated hemoglobin and the non-oxygenated hemoglobin in the blood is used to solve the value T of the blood oxygen saturation SpO2 of the RGB three-channel gray value.

4. The method of claim 1, wherein: The initial scattered light intensity matrix M is constructed in step S3 H The formula is as follows: In the above formula, The matrix is the mean matrix, which is S H The same size, The element value in S H The mean of the row in which the element in the corresponding position of the matrix is located, The matrix is the centralized data matrix, is The transpose of the matrix.

5. The method for real-time blood flow imaging based microsurgery intraoperative tissue perfusion automatic reconstruction according to claim 1, characterized in that: The step S4 calculates the contrast K in the spatial neighborhood of the pixel gray value of the region of interest in the collected N=n original speckle images st As follows where N is the number of original speckle image frames collected, Ns represents the spatial neighborhood width, p denotes the serial number of the pixel, I p represents the average value of the gray scale of the pth pixel in the spatial neighborhood of the same position in the N original speckle image frames, the average value of the gray scale of the pth pixel in the spatial neighborhood of the same position in the N original speckle image frames,​​ 6. The method of claim 1, wherein: In step S5, the blood flow velocity v of the corresponding biological tissue at the position is calculated by using the contrast in the spatial neighborhood. where (i,j) is the ith row and jth column, K st is the contrast.

7. The method of claim 1, wherein: In step S6, the RPE algorithm is used to combine the time contrast algorithm and the spatial contrast algorithm in a fixed time period to improve the time resolution and reliability. The RPE theory regards the light scattering phenomenon in a sufficiently short camera exposure time [t0, t1] as a continuous random process, and Y(t) represents the random variable of the light intensity integral at time t. It is believed that the movement speed of the scattering particles in the imaging area remains unchanged within this sufficiently short camera exposure time. For continuous N frames of original speckle images, the light intensity standard deviation at a certain pixel point is calculated using a 3x3 time window, and the contrast value is represented as: where t0represents the start time of the exposure time [t0, t1], σ rpe is an estimate of the standard deviation of the random process, μ rpe is an estimate of the mean of the random process, and are the theoretical values of the mean and standard deviation of the random process, k y is the theoretical value of the contrast value.

8. The method for real-time blood flow imaging based microsurgery intraoperative tissue perfusion automatic reconstruction according to claim 1, characterized in that: Step S7 Monte Carlo simulation is used to simulate the component size of laser speckle at different depths, and the proportion of surface blood flow velocity to the whole is obtained, that is, the correction parameter c of blood flow velocity is obtained. The corrected blood flow velocity is calculated by using the correction parameter c, that is, the surface blood flow velocity v' = cv.

9. The method for real-time blood flow imaging based microsurgery intraoperative tissue perfusion automatic reconstruction according to claim 1, characterized in that: In step S9, the blood perfusion P is calculated by combining the blood oxygen saturation change value and the blood flow velocity in the time period, and the formula is as follows: Where D1 and D2 are proportional coefficients, the range is 0-1, v' is the blood flow velocity of the nearest blood vessel from the distance (i, j), and r is the pipe diameter radius of the blood vessel.

Images