Method and system for generating a path for a cranial window surgery of a mouse and storage medium

Abstract

The application relates to the technical field of surgical equipment, and particularly discloses a path generation method and system for mouse cranial window surgery and a storage medium, wherein light is split by a fiber coupler to a reflecting mirror and focused on the surface of a mouse skull, the light reflected by the two is interfered in the fiber coupler, the interference light signal is output to a spectrometer for capture, and the interference light signal is collected by a high-speed data acquisition card; the collected data is processed by Python programming to form a tomographic signal, an algorithm is designed for the tomographic signal to separate the upper and lower surfaces, and a craniotomy path for automatic craniotomy surgery of a micrometer precision small carving machine is generated.

Description

Technical Field

[0001] The present invention relates to the technical field of surgical equipment, and in particular to a path generation method, system and storage medium for mouse cranial window surgery. Background Art

[0002] Mice are currently widely used in brain science and neuroscience experiments. However, both the introduction of neural probes and high-resolution optical imaging of the brain require partial skull removal to expose relevant brain regions, which is where cranial windows come in. The thickness of the mouse skull ranges from 100 to 400 microns, and craniotomy is currently performed manually. This is associated with high training costs, lengthy procedures, and low success rates.

[0003] Existing research on automated mouse craniotomy techniques primarily involves foreign-based impedance-feedback craniotomies and contact-feedback surface-modeling craniotomies. The former utilizes the impedance difference between the skull and dura mater for feedback control of incision depth, limiting its ability to penetrate the skull for implant windows and susceptible to false positives caused by small blood vessels within the skull, leading to premature termination. The latter utilizes micro-force contact sensors to contact the skull surface, modeling the upper surface. A consistent incision depth is then set, iteratively deepened, and manual inspection is required after each step to ensure the desired depth is achieved. A significant difference between the thickest and thinnest areas of the cranial window can result in suboptimal results. Both impedance and contact sensing are slow due to the feedback loop.

[0004] Currently, there are no commercial mouse surgical robots / platforms on the market, and existing studies have been unable to obtain sufficient cranial window path data, making it impossible to perform fast, high-precision, fully automatic craniotomy. Therefore, how to optimize the existing system to obtain a high-precision cranial window surgical path and execute it quickly and accurately has become a technical problem that needs to be solved urgently.

[0005] Based on the above problems, the present invention proposes a path generation method, system and storage medium for mouse cranial window surgery. Summary of the Invention

[0006] The purpose of the present invention is to provide a path generation method, system and storage medium for mouse cranial window surgery, which solves the problem that cranial window surgery is prone to failure in the prior art.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] The path generation method for mouse cranial window surgery includes the following steps:

[0009] Step 1: Fix the mouse at the target position and then use optical interferometry to obtain the mouse's spectral data;

[0010] Step 2: Perform Python data processing on the mouse spectral data obtained in step 1 to obtain the Aline signal of the current point on the skull, that is, the preliminary tomographic data to be saved;

[0011] Step 3: Move the probe to scan the skull area according to the corresponding tomographic data, so that the tomographic data of the entire skull range can be obtained and saved;

[0012] Step 4: Based on the tomographic data from step 3, a two-dimensional craniotomy path is selected. If it is an implant window, the two-dimensional path is a closed loop line connected end to end; if it is a thinned window, the two-dimensional path is a covered surface, and then the corresponding tomographic data is extracted;

[0013] Step 5: Separate the upper and lower surfaces of the tomographic data in step 4 to generate a three-dimensional craniotomy path.

[0014] On the basis of the above technical solutions, the present invention also provides the following optional technical solutions:

[0015] In the optional solution: the specific operations in step 1 are:

[0016] An ultra-wideband light source emits low-coherence light with a central wavelength of 1550nm, which is split into two beams by a fiber coupler. The two beams hit the reflector and collimated by a lens, and then focused on the mouse skull by a lens.

[0017] The mouse is fixed in a mouse holder, and the light reflected from the mouse skull and the light reflected from the reflector interfere with each other in the fiber optic coupler.

[0018] The interference light propagates back from the fiber coupler to the spectrometer, and the spectral data is collected and transmitted to the computer by a high-speed linear array camera and a high-speed acquisition card.

[0019] In the optional scheme: Step three is specifically: controlling the X-axis motor, Y-axis motor, and Z-axis motor on the engraving machine through computer serial port communication to realize the movement of the main axis in the three-dimensional workspace, and then carrying the probe to scan the skull area, so that the tomographic data of the entire skull range can be obtained and the data can be saved. The probe consists of a collimating lens and a focusing lens.

[0020] In the optional solution: Step 5 specifically comprises: using a segmentation algorithm to separate the upper and lower surfaces to generate a three-dimensional craniotomy path for craniotomy;

[0021] The segmentation algorithm includes a threshold segmentation algorithm, which separates the upper surface of the tomographic data extracted in step 4, then performs linear interpolation on some discrete points and undetected points, and then generates a smooth upper surface contour through mean filtering.

[0022] In an optional solution: the segmentation algorithm includes a semi-automatic segmentation algorithm, which processes the tomographic data in step 4 to generate a visual tomographic map, manually marks the corresponding points in the map, and then performs polynomial fitting. The fitted curve is observed in real time in the map until it fully overlaps with the lower surface. When it overlaps, the punctuation is terminated, and a smooth lower surface contour is generated using the fitted polynomial.

[0023] A path generation system for mouse cranial window surgery comprises a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above method are implemented.

[0024] A computer-readable storage medium stores a computer program, which implements the steps of the above method when executed by a processor.

[0025] Compared with the prior art, the present invention has the following beneficial effects:

[0026] 1. The present invention uses SDOCT imaging technology, which has a micron-level longitudinal resolution and is non-invasive, avoiding problems such as skull deformation caused by contact. It also uses near-infrared light, which is safer than currently commonly used X-rays and avoids radiation risks.

[0027] 2. This invention aims to use OCT tomographic data to obtain skull depth information before craniotomy. By presetting the drilling depth, the corresponding craniotomy path can be generated to complete cranial window surgeries for different purposes. Compared with existing impedance sensing and contact sensing automatic craniotomy platforms, this system has a higher success rate, higher precision, faster speed, and wider application.

[0028] 3. The OCT technology used in the present invention is also a commonly used technology for high-resolution optical imaging. The overall device is easy to use under the interactive design of the GUI. It is also conducive to direct on-site observation in conjunction with imaging technology after craniotomy. It has great value and application potential in biological and scientific research. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 This is a principle block diagram of the present invention.

[0030] Figure 2 Detailed system diagram of the present invention.

[0031] Figure 3 It is the workflow diagram of the present invention.

[0032] Figure 4 Schematic diagram of the segmentation method of the present invention.

[0033] Figure 5Schematic diagram of the method for achieving drill bit and scanning point position calibration according to the present invention.

[0034] Notes on the accompanying drawings: ultra-wideband light source 1, fiber coupler 2-1, reflector 2-2, fiber collimator 2-3, lens 2-4, engraving machine 3-1, mouse holder 3-2, X-axis motor 4-1, Y-axis motor 4-2, Z-axis motor 4-3, high-speed drill 4-4, spectrometer 5-1, high-speed linear array camera 5-2, high-speed acquisition card 5-3, computer 6. DETAILED DESCRIPTION

[0035] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without creative work are within the scope of protection of the present invention.

[0036] like Figure 1-Figure 5 As shown, an embodiment of the present invention provides a path generation method for mouse cranial window surgery, comprising:

[0037] In step (1), an ultra-wideband light source 1 emits low-coherence light with a central wavelength of 1550 nm and a bandwidth of 210 nm, which is split into two beams by a fiber coupler 2-1 at a ratio of 25% and 75%, respectively, hitting a reflector 2-2 and collimating through a lens 2-3, and then focusing on the mouse skull through a lens 2-4;

[0038] Step (2) The mouse is fixed on the mouse holder 3-2, and the light reflected from the mouse skull and the light reflected from the reflector 2-2 interfere with each other in the optical fiber coupler 2-1;

[0039] Step (3) The interference light propagates back from the fiber coupler 2-1 to the spectrometer 5-1 (Cobra1600, Wasatch Photonics can be used), and the spectral data is collected and transmitted to the computer 6 by the high-speed linear array camera 5-2 (GL2048R-10A-ENC-STD-210, Sensors Unlimited can be used) and the high-speed acquisition card 5-3 (NI-1433 can be used, which can be triggered by the IO card model NI-6323). At this time, the spectral data is a function of wavelength and light intensity, including DC and AC interference. The AC interference is the data we need.

[0040] Step (4) Data processing is performed using Python in computer 6. First, the original spectrum is averaged to obtain the spectral DC amount. After subtracting the DC amount, only the AC interference amount remains in the original spectrum;

[0041] Step (5) The AC interference quantity is a function of wavelength and light intensity, and is non-uniform. Using the correspondence between wavelength and wave number, a uniform linear interpolation is performed on it to convert the wavelength space (λ) into the wave number space (k), and the result is a function of wave number and light intensity;

[0042] Step (6) finally converts this function into the time domain through inverse Fourier transform to obtain the function of depth and light intensity, that is, the Aline signal of the current point of the skull, which is also the tomographic data to be saved;

[0043] Step (7) controls the X-axis motor 4-1, the Y-axis motor 4-2, and the Z-axis motor 4-3 on the engraving machine 3-1 through the serial communication of the computer 6 to realize the movement of the main shaft in the three-dimensional workspace, and then uses the probe (composed of a collimating lens 2-3 and a focusing lens 2-4) to perform a large-scale scan of the skull area (such as 10mm*10mm, which can cover the skull craniotomy range), so as to obtain tomographic data of the entire skull range and save the data;

[0044] Step (8) in computer 6, a two-dimensional path of craniotomy is selected from the data saved in step (7). If it is an implanted window (only the edge of the cranial window area needs to be polished), the two-dimensional path is a closed loop line connected end to end; if it is a thinned window (the cranial window area surface needs to be thinned), the two-dimensional path is a covered surface, and then the corresponding tomographic data on the two-dimensional path is extracted;

[0045] After step (9), a segmentation algorithm is designed to separate the upper and lower surfaces based on the extracted tomographic structure data, and a three-dimensional craniotomy path for craniotomy is generated to guide the engraving machine 3-1 to carry the high-speed drill 4-4 to perform craniotomy.

[0046] Segmentation algorithm scheme:

[0047] ① Design a threshold segmentation algorithm to separate the upper surface of the tomographic data extracted in step (8), then perform linear interpolation on some discrete points and undetected points, and then generate a smooth upper surface contour through mean filtering; then design a semi-automatic segmentation algorithm to separate the lower surface, generate a visual tomographic map by processing the tomographic data, manually mark the corresponding points in the map, and then perform polynomial fitting. It can be observed in real time in the map whether the fitted curve fully overlaps with the lower surface. When it overlaps enough, end the punctuation and use the fitted polynomial to generate a smooth lower surface contour;

[0048] ②Use a semi-automatic segmentation algorithm to segment the upper and lower surfaces as the standard. After segmenting enough data, design a semantic segmentation network for training and learning. Then, use the trained semantic segmentation network model to automatically segment the upper and lower surface contours for the tomographic structure data.

[0049] The embodiment adopts open-source Pycharm software to process OCT data in Python programming environment, to obtain OCT tomographic data of the whole skull surface, then program to select a point of the surface projection map, set the diameter, obtain the two-dimensional plane path range of the desired craniotomy, then program to extract all signals on the path, write segmentation algorithm according to the segmentation scheme to segment the upper and lower surfaces, then set the drill bit pitch (0.2mm) according to the width of the drill bit used (0.3mm), and discretize the upper and lower surface points of the craniotomy from the continuous upper and lower surfaces as the guide of the three-dimensional craniotomy path. Finally, the high-precision carving machine is calibrated to the starting point of the skull surface of the selected two-dimensional path, and follows the three-dimensional craniotomy path to gradually deepen from 20% to 90% of the craniotomy surgery.

[0050] Our robot platform provides a solution for non-professional trained researchers to perform craniotomy surgery, and OCT itself is an imaging technology, which is our exploration of image-guided surgical robots, and the integration of the OCT probe on the main shaft breaks through the limitation of small OCT imaging area, greatly increasing the degree of freedom, and theoretically can scan and operate on any large area in the platform. Therefore, we believe that this technology has great potential in biomedical research, can perform three-dimensional imaging on a large area, can replace the optical probe and drill bit tool with minimal error, and then customize the area to perform minimally invasive surgery at a customized depth;

[0051] Figure 3 The figure shows the flow scheme of the whole system, and the specific embodiment is:

[0052] (1) Scan the mouse skull with a large range of Cscan to obtain tomographic data of the whole skull surface;

[0053] (2) Select the corresponding two-dimensional path (for example, a circular glass skull window) in the GUI (user interface);

[0054] (3) Extract the tomographic data on the path and separate the upper and lower surfaces according to the tomographic data to obtain the corresponding three-dimensional path;

[0055] (4) Calibrate the drill bit to the starting point of the surgical path and control the carving machine to perform craniotomy surgery along the three-dimensional path;

[0056] Segmentation method and calibration method:

[0057] Figure 4 is a schematic diagram of the segmentation method, which is a row of Bscan (OCT image, composed of multiple Aline, which is a cross-sectional view) of the mouse skull.

[0058] Figure 4: For the segmentation of the upper surface, we first select the adaptive threshold by Otsu's method to binarize the image, then connect some gaps by morphological closing operation, and then use Canny edge detection algorithm to detect the edge of the black and white image to get the edge profile (top left), and then filter out irrelevant edges by screening the edges to get the upper surface (top right). For the segmentation of the lower surface, we manually mark some key points on the image (bottom left), and use high-order polynomial fitting to interpolate these key points to get the lower surface (bottom right). Finally, we get the segmented upper and lower surfaces (right).

[0059] Figure 5 is a schematic diagram of the method for calibrating the drill bit and the scanning point position, in order to realize the switching of the tool from imaging to actual operation.

[0060] Figure 5 : Our light path and drill bit are not co-path, and need to be calibrated to successfully align the drill bit to the starting point after imaging and segmentation. The lateral distance [Δx, Δy] between the scanning point and the drill bit tip is fixed, and the longitudinal distance has the following relationship:

[0061] Δz = Δz1 + Δz2 (1)

[0062] In the formula, Δz is the total longitudinal distance, Δz1 is the distance between the drill bit tip and the imaging zero point, which is fixed. Δz2 is the actual distance from the imaging zero point to the upper surface, which is obtained by the following formula:

[0063] Δz2 = k air ·z up (2)

[0064] In the formula, z up is the pixel distance of the upper surface from the imaging zero point, and k air is the pixel resolution in air. Since uniform interpolation was performed during previous OCT data processing, the pixel resolution here is a constant. In this embodiment, the pixel resolution of the OCT signal in air is 5 μm, and the upper surface is located at the 100 pixel point, so Δz2 = 100 * 5 = 500 μm.

[0065] Thus, the relative distance between the drill bit and the scanning point can be obtained, and the replacement of the OCT probe and the drill bit is realized, which is also a key step from imaging to craniotomy.

[0066] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A path generation method for mouse cranial window surgery, characterized by: The following steps are involved: Step 1: Fix the mouse at the target position and then use optical interferometry to obtain the mouse's spectral data; Step 2: Perform Python data processing on the mouse spectral data obtained in step 1 to obtain the Aline signal of the current point on the skull, that is, the preliminary tomographic data to be saved; Step 3: Move the probe to scan the skull area according to the corresponding tomographic data, obtain tomographic data of the entire skull range, and save the data; Step 4: Based on the tomographic data obtained in step 3, a two-dimensional craniotomy path is selected. If it is an implant window, the two-dimensional path is a closed loop line connected end to end; If it is a thinned window, the two-dimensional path is a coverage surface, and then the corresponding tomographic data are extracted; Step 5: Separate the upper and lower surfaces of the tomographic data extracted in step 4 to generate a three-dimensional craniotomy path; Step 5 is specifically as follows: using a segmentation algorithm to separate the upper and lower surfaces to generate a three-dimensional craniotomy path for craniotomy; The segmentation algorithm includes a threshold segmentation algorithm, which separates the upper surface of the tomographic data extracted in step 4, then performs linear interpolation on some discrete points and undetected points, and then generates a smooth upper surface contour through mean filtering; The segmentation algorithm includes a semi-automatic segmentation algorithm, which generates a visual tomographic map using the tomographic data extracted in step 4, manually marks the corresponding points in the map, and then performs polynomial fitting. The fitted curve is observed in real time in the map until it fully overlaps with the lower surface. When it overlaps, the punctuation is terminated and a smooth lower surface contour is generated using the fitted polynomial.

2. The path generation method for mouse cranial window surgery according to claim 1, characterized in that: The specific operations of step one are: An ultra-wideband light source (1) emits low-coherence light with a central wavelength of 1550 nm, which is split into two beams by a fiber coupler (2-1). The two beams hit the reflector (2-2) and the lens (2-3) for collimation, and then are focused on the mouse skull by the lens (2-4). The mouse is fixed on the mouse holder (3-2), and the light reflected from the mouse skull and the light reflected from the reflector (2-2) interfere with each other in the fiber coupler (2-1); The interference light propagates back from the fiber coupler (2-1) to the spectrometer (5-1), and the high-speed linear array camera (5-2) and the high-speed acquisition card (5-3) collect the spectral data and transmit them to the computer (6).

3. The path generation method for mouse cranial window surgery according to claim 2, characterized in that: Step three is specifically as follows: controlling the X-axis motor (4-1), Y-axis motor (4-2), and Z-axis motor (4-3) on the engraving machine (3-1) through the serial communication of the computer (6) to realize the movement of the main axis in the three-dimensional workspace, and then scanning the skull area with the probe to obtain the tomographic data of the entire skull range and save the data.

4. A path generation system for mouse cranial window surgery, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 3 are implemented.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 3 are implemented.

Images