A method for spatial segmentation lattice target delineation based on functional imaging bio-guidance

Abstract

The present invention provides a method for spatial segmentation lattice target area delineation based on functional imaging biological guidance, which belongs to the spatial segmentation technology and comprises the following steps: setting the scanning parameters of DWI and imaging to obtain the DWI image; using the two-dimensional rigid sub-pixel image registration algorithm to register the DWI image; performing two-step weighted linear least squares fitting on the registered image, and performing image transformation and bilinear interpolation to generate an ADC parameter image; reducing the GTV volume by three-dimensional filtering to obtain the GTV decreased ; Obtain the lattice point coordinate set in the ADC parameter map through three-dimensional filtering; Calculate the lattice spacing threshold of different voxels through dynamic programming; Screen the lattice point coordinate set; Delineate the lattice target area for the screened lattice point coordinates, and perform biologically guided spatial segmentation. The present invention realizes accurate identification of tumor areas with low ADC values ​​based on DWI, and can better control the risk of local recurrence in tumor areas with low ADC values.

Description

Technical Field

[0001] The invention belongs to the technical field of space segmentation, and in particular relates to a method for delineating a space segmentation lattice target area based on functional imaging biological guidance. Background Art

[0002] Spatial Fractionated Radiotherapy (SFRT) proposes the concept of highly non-uniform dose irradiation of tumors. This strategy has achieved remarkable results in the treatment of giant metastatic and primary tumors, and effectively reduced the high toxicity to surrounding organs. The successful implementation of SFRT depends greatly on the development of physics and radiotherapy technology, especially the enhancement of modulation capability, which has promoted the transformation of SFRT from two-dimensional grid radiotherapy to three-dimensional lattice radiation therapy (Lattice Radiation Therapy, LRT). Lattice radiation therapy (LRT) has a higher degree of conformity with the target area and can better protect organs near the target area. It is the main technical means to achieve SFRT at this stage.

[0003] The parameters of the lattice (diameter, spacing and position) have a great influence on the LRT dose distribution. How to calculate the optimal lattice arrangement structure in the target area while ensuring the gradient of the peak and valley doses is the key to optimizing LRT treatment. Some researchers have proposed using the densest packing model to arrange and outline the lattice in the target area, generating a peak dose around the lattice target area and a valley dose in the tumor area where the lattice is not arranged. Although this method has a high spatial utilization rate and can produce a certain peak-to-valley ratio, since the location information of the biological target area is not taken into account, the high dose generated by the lattice peak dose may miss the solid tumor area in the target area. The solid tumor area may persist after radiotherapy or recur early after treatment, causing unnecessary burden on patients.

[0004] The diffusion-weighted imaging (DWI) sequence in magnetic resonance imaging (MRI) generates MR contrast based on the diffusion of water molecules and quantifies the in vivo diffusion rate through the parameter of apparent diffusion coefficient (ADC). The presence of cell membranes limits the diffusion of water molecules, resulting in low ADC values. Therefore, tumor areas with low ADC values ​​may reflect highly dense tumor cells. During radiotherapy, a decrease in low ADC volume may mean a reduction in solid tumors, while an increase in low ADC volume may indicate that the tumor is not responding to treatment. Tumor areas with low ADC values ​​are at a higher risk of local recurrence. As shown in the figure, accurate identification of tumor areas with low ADC based on the nuclear magnetic imaging DWI sequence can effectively guide the position optimization, arrangement and delineation of SFRT high-dose lattice points, thereby improving the efficacy of SFRT treatment. Summary of the invention

[0005] In view of the above situation, the present invention provides a method for spatial segmentation lattice target area delineation based on functional imaging biological guidance, which standardizes the measurement environment and measurement process, as well as the reverse modeling process of pipeline point cloud data, to achieve rapid extraction of pipeline assembly features from pipeline point cloud data.

[0006] In order to achieve one or more of the above purposes, the technical solution adopted by the present invention is:

[0007] Step S1: setting the scanning parameters of diffusion-weighted imaging (DWI), using a 3T magnetic resonance scanner to perform diffusion-weighted magnetic imaging, and obtaining a DWI image;

[0008] Step S2: using a two-dimensional rigid sub-pixel image registration algorithm to register the DWI images and obtain transformation parameters (translation, rotation and scaling) between the DWI images;

[0009] Step S3: The registered image is fitted by two-step weighted linear least squares (LLS) to generate an apparent diffusion coefficient (ADC) parameter image, and image transformation and bilinear interpolation are performed to generate an ADC parameter image with the same matrix size as that of computed tomography (CT);

[0010] Step S4: Reduce the volume of the gross tumor volume (GTV) through three-dimensional filtering to obtain the reduced tumor target volume GTV decreased ;

[0011] Step S5: obtaining a lattice point coordinate set lattice_locations in the ADC parameter map through three-dimensional filtering;

[0012] Step S6: Calculate the lattice spacing thresholds of different voxels through dynamic programming to screen lattice points;

[0013] Step S7: screening each lattice point coordinate in the lattice point coordinate set lattice_locations to determine whether it meets the lattice setting requirements;

[0014] Step S8: Delineate the lattice target area using the screened lattice point coordinates to perform biologically guided spatial segmentation.

[0015] In one embodiment, the step S1 comprises the following steps:

[0016] Step S101: Set the DWI scanning parameters, the ratio of repetition time to echo time is 6443 / 77msec, the flip angle is 90°, and the field of view is 400×324mm 2 , the acquisition matrix size is 64×94, the reconstruction matrix size is 256×216, 30 slices, the ratio of layer thickness to gap is 5 / 1mm, and the spatial resolution is 0.92×0.92×5mm 3 , the diffusion sensitivity gradient factor b is 0, 10, 25, 50, 75, 100 sec / mm 2 , the number of repetitions is 111;

[0017] Step S102: Use a 3T magnetic resonance scanner to perform diffusion weighted magnetic imaging to obtain a DWI image.

[0018] In one embodiment, step S2 comprises the following steps:

[0019] Step S201: taking the DWI image when b is 0 as the reference image, using a two-dimensional rigid sub-pixel image registration algorithm to achieve registration between DWI images with different b values, and obtaining transformation parameters (translation, rotation and scaling) between the DWI images;

[0020] Step S202: Assume that there are two images I and J, where J is the image of I after translation, rotation and scaling. The coordinates on image I are (x, y), the coordinates on image J are (x', y'), and the translation parameter is (t x ,t y ), the rotation parameter is θ, the scaling parameter is s, and the transformation relationship is as follows:

[0021] x′=s*(x*cos(θ)-y*sin(θ))+t x

[0022] y′=s*(x*sin(θ)+y*cos(θ))+t γ ;

[0023] Step S203: Evaluate the similarity between image I and image J by using similarity measurement functions (cross-correlation, mean square error, normalized cross-correlation coefficient);

[0024] Step S204: Minimize the similarity metric function through an optimization algorithm (gradient descent, least squares method) to obtain translation, rotation and scaling parameters.

[0025] In one embodiment, step S3 comprises the following steps:

[0026] Step S301: For the registered DWI image, a two-step weighted linear least squares fit is performed using the formula Log(S b)=Log(S0)-b*ADC Calculate the ADC parameter image at the voxel level, and the weight of the two-step weighted linear least squares is N(b)*S LLS (b) 2 , b is the diffusion sensitive gradient factor, S b and S0 are the signal intensities at a given b value and b = 0, respectively, and N(b) is the number of repeated acquisitions at a given b value;

[0027] Step S302: After the ADC parameter image is generated, the CT image is used as the reference image, and the DWI image with b=500 is registered by a two-dimensional rigid sub-pixel image registration algorithm. The same image transformation is applied to the ADC parameter image, and an ADC parameter image with the same matrix size as the CT is generated by bilinear interpolation.

[0028] In one embodiment, step S4 comprises the following steps:

[0029] Step S401: according to the set retraction distance L scale and spatial dimension resolution X space , Y space , Z space , generate spheroid filter spheroid scale , spheroid scale The semi-major axis a spheroid_scale =L scale / X space , b spheroid_scale =L scale / Y space , c spheroid_scale =L scale / Z space ;

[0030] Step S402: When filtering the GTV binary matrix, the stride is 1 pixel in each of the three dimensions, and the reduced tumor target area GTV is generated by judging whether the current center is equal to the minimum filtering value. decreased .

[0031] In one embodiment, step S5 comprises the following steps:

[0032] Step S501: According to the set search distance L search and spatial dimension resolution X space , Y space , Z space Construct spheroid filter search , spheroid search The semi-major axis a spheroid_search =L search / X space , bspheroid_search =L search / Y space , c spheroid_search =L search / Z space ;

[0033] Step S502: By decreased The ADC parameter map in the image is filtered with a step of 1 in three dimensions, and it is determined whether the center is the minimum value of the filter, and the lattice point coordinate set lattice_locations in the ADC parameter map is retrieved.

[0034] In one embodiment, step S6 comprises the following steps:

[0035] Step S601: Setting the minimum lattice spacing L lattice_min and the maximum spacing L lattice_max , the maximum lattice spacing L lattice_max By L search Decision, L. lattice_max =2*L search ;

[0036] Step S602: construct a linear regression equation between ADC and lattice spacing threshold, and solve the parameters k and c of the linear equation, as shown below: ADC GTV For GTV decreased The collection of ADC values ​​of the voxels within,

[0037] L lattice_limit =k*ADC+c

[0038] k=(L lattice_max -L lattice_min ) / (max(ADC GTV )-min(ADC GTV ))

[0039] c=L lattice_min -min(ADC GTV )*k.

[0040] In one embodiment, step S7 comprises the following steps:

[0041] Step S701: lattice_locations each lattice point coordinate lattice_location j , through the linear regression equation L j_limit =k*ADC j +cCalculate the lattice spacing threshold, ADC j lattice_location j The corresponding ADC value;

[0042] Step S702: for lattice_location j and the coordinates of the five nearest lattice points lattice_location1, lattice_location2…lattice_location n (n=5), calculate lattice_location j The sum of the distance limit ratios L n lattice_location j to lattice_location n The geometric distance of

[0043] Step S703: For each lattice point coordinate Lattice_Location in lattice_locations i , satisfying any of the following conditions, any other Lattice_Location j To Lattice_Location i The distance is greater than L i_limit , or for distances less than L i_limit Any Lattice_Location j , Lattice_Location i The distance limit ratio is Ratio i_sum5 Greater than Lattice_Location j The distance limit ratio is Ratio j_sum5 , the Lattice_Location i It can be regarded as the geometric center of the lattice lattice_location i ;

[0044] Step S704: For each lattice_location i , with a radius of 5mm, draw a lattice sphere i .

[0045] In one embodiment, in step S8, the lattice target area is delineated by using the selected lattice point coordinates to perform biologically guided spatial segmentation.

[0046] The beneficial effects of the present invention include at least one of the following:

[0047] 1) The present invention proposes a method for spatial segmentation lattice target delineation based on functional imaging biological guidance. Through the SFRT biological target guidance method of diffusion-weighted imaging, the position, diameter and spacing of the spatial segmentation LRT lattice points are optimized, so that a higher peak dose can be directly applied to the biological target area (low ADC, high tumor cell density proliferation area), resulting in a higher ablation ratio, without pursuing an excessively high peak-to-valley ratio, and pursuing a higher peak-to-valley ratio for the lattice points generated in the area outside the biological target area.

[0048] 2) Compared with the densest packing method, the method proposed in the present invention can ensure the peak-to-valley ratio of spatial segmentation treatment while obtaining a better LRT lattice arrangement structure, better control the risk of local recurrence in tumor areas with low ADC, and thus optimize the LRT treatment effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 is a flow chart of the method of the present invention;

[0050] Figure 2 Schematic diagram of SFRT lattice arrangement based on biological functional target area;

[0051] Figure 3 GTV in the 3D view decreased Generate renderings;

[0052] Figure 4 is the GTV and GTV of a single layer in the ADC parameter diagram decreased Range display diagram;

[0053] Figure 5 34 lattice coordinate diagrams generated;

[0054] Figure 6 The lattice coordinate diagram is screened according to the dynamic programming of lattice spacing;

[0055] Figure 7 The effect diagram of SFRT lattice arrangement based on biological target area is outlined. DETAILED DESCRIPTION

[0056] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0057] The present invention proposes a method for spatial segmentation lattice target area delineation based on functional imaging bio-guidance, which can accurately identify tumor areas with low ADC values ​​based on DWI, guide the position optimization, arrangement and delineation of SFRT high-dose lattice points, and calculate the optimal lattice arrangement structure in the target area while ensuring the gradient of peak and valley doses, so as to better control the risk of local recurrence in tumor areas with low ADC, thereby optimizing the LRT treatment effect.

[0058] The method for spatial segmentation lattice target area delineation based on functional imaging biological guidance proposed in the present invention comprises the following steps:

[0059] Step S1: Setting the scanning parameters of diffusion-weighted imaging (DWI), using a 3T magnetic resonance scanner to perform diffusion-weighted magnetic imaging, and obtaining a DWI image. Preferably, the specific implementation method of this step is as follows:

[0060] Step S101: Set the DWI scanning parameters, the ratio of repetition time to echo time is 6443 / 77msec, the flip angle is 90°, and the field of view is 400×324mm 2 , the acquisition matrix size is 64×94, the reconstruction matrix size is 256×216, 30 slices, the ratio of layer thickness to gap is 5 / 1mm, and the spatial resolution is 0.92×0.92×5mm 3 , the diffusion sensitivity gradient factor b is 0, 10, 25, 50, 75, 100 sec / mm 2 , the number of repetitions is 111;

[0061] Step S102: Use a 3T magnetic resonance scanner to perform diffusion weighted magnetic imaging to obtain a DWI image.

[0062] Step S2: Use a two-dimensional rigid sub-pixel image registration algorithm to register the DWI images and obtain the transformation parameters (translation, rotation and scaling) between the DWI images. Preferably, the specific implementation method of this step is as follows:

[0063] Step S201: taking the DWI image when b is 0 as the reference image, using a two-dimensional rigid sub-pixel image registration algorithm to achieve registration between DWI images with different b values, and obtaining transformation parameters (translation, rotation and scaling) between the DWI images;

[0064] Step S202: Assume that there are two images I and J, where J is the image of I after translation, rotation and scaling. The coordinates on image I are (x, y), the coordinates on image J are (x', y'), and the translation parameter is (t x ,t y), the rotation parameter is θ, the scaling parameter is s, and the transformation relationship is as follows:

[0065] x′=s*(x*cos(θ)-y*sin(θ))+t x

[0066] y′=s*(x*sin(θ)+y*cos(θ))+t γ ;

[0067] Step S203: Evaluate the similarity between image I and image J by using similarity measurement functions (cross-correlation, mean square error, normalized cross-correlation coefficient);

[0068] Step S204: Minimize the similarity metric function through an optimization algorithm (gradient descent, least squares method) to obtain translation, rotation and scaling parameters.

[0069] Step S3: The registered image is fitted by two-step weighted linear least squares (LLS) to generate an apparent diffusion coefficient (ADC) parameter image, and image transformation and bilinear interpolation are performed to generate an ADC parameter image with the same matrix size as computed tomography (CT). Preferably, the specific implementation method of this step is as follows:

[0070] Step S301: For the registered DWI image, a two-step weighted linear least squares fit is performed using the formula Log(S b )=Log(S0)-b*ADC Calculate the ADC parameter image at the voxel level, and the weight of the two-step weighted linear least squares is N(b)*S LLS (b) 2 , b is the diffusion sensitive gradient factor, S b and S0 are the signal intensities at a given b value and b = 0, respectively, and N(b) is the number of repeated acquisitions at a given b value;

[0071] Step S302: After the ADC parameter image is generated, the CT image is used as the reference image, and the DWI image with b=500 is registered by a two-dimensional rigid sub-pixel image registration algorithm. The same image transformation is applied to the ADC parameter image, and an ADC parameter image with the same matrix size as the CT is generated by bilinear interpolation.

[0072] Step S4: Reduce the volume of the gross tumor volume (GTV) through three-dimensional filtering to obtain the reduced tumor target volume GTV decreased Preferably, the specific implementation method of this step is as follows:

[0073] Step S401: according to the set retraction distance L scale and spatial dimension resolution X space , Y space , Z space , generate spheroid filter spheroid scale , spheroid scale The semi-major axis a spheroid_scale =L scale / X space , b spheroid_scale =L scale / Y space , c spheroid_scale =L scale / Z space ;

[0074] Step S402: When filtering the GTV binary matrix, the stride is 1 pixel in each of the three dimensions, and the reduced tumor target area GTV is generated by judging whether the current center is equal to the minimum filtering value. decreased , GTV in 3D view decreased The resulting effect is as follows Figure 3 As shown in the ADC parameter diagram, the GTV and GTV of a single layer decreased Range such as Figure 4 As shown, green is GTV and red is GTV decreased .

[0075] Step S5: Obtain the lattice point coordinate set lattice_locations in the ADC parameter map through three-dimensional filtering, such as Figure 5 Preferably, the specific implementation method of this step is as follows:

[0076] Step S501: According to the set search distance L search and spatial dimension resolution X space , Y space , Z space Construct spheroid filter search , spheroid search The semi-major axis a spheroid_search =L search / X space , b spheroid_search =L search / Y space , c spheroid_search =L search / Z space ;

[0077] Step S502: By decreasedThe ADC parameter map in the ADC parameter map is filtered with a step of 1 in three dimensions, and it is determined whether the center is the minimum value of the filter, and the lattice point coordinate set lattice_locations in the ADC parameter map is retrieved.

[0078] Step S6: Calculate the lattice spacing thresholds of different voxels through dynamic programming for screening lattice points. Preferably, the specific implementation method of this step is as follows:

[0079] Step S601: Setting the minimum lattice spacing L lattice_min and the maximum spacing L lattice_max , the maximum lattice spacing L lattice_max By L search Decision, L. lattice_max =2*L search ;

[0080] Step S602: construct a linear regression equation between ADC and lattice spacing threshold, and solve the parameters k and c of the linear equation, as shown below: ADC GTV For GTV decreased The collection of ADC values ​​of the voxels within,

[0081] L lattice_limit =k*ADC+c

[0082] k=(L lattice_max -L lattice_min ) / (max(ADC GTV )-min(ADC GTV ))

[0083] c=L lattice_min -min(ADC GTV )*k.

[0084] Step S7: Screen each lattice point coordinate in the lattice point coordinate set lattice_locations to determine whether it meets the lattice setting requirements. Preferably, the specific implementation method of this step is as follows:

[0085] Step S701: lattice_locations each lattice point coordinate lattice_location j , through the linear regression equation L j_limit =k*ADC j +cCalculate the lattice spacing threshold, ADC j lattice_location j The corresponding ADC value;

[0086] Step S702: for lattice_location jand the coordinates of the five nearest lattice points lattice_location1, lattice_location2…lattice_location n (n=5), calculate lattice_location j The sum of the distance limit ratios L n lattice_location j to lattice_location n The geometric distance of

[0087] Step S703: For each lattice point coordinate Lattice_Location in lattice_locations i , satisfying any of the following conditions, any other Lattice_Location j To Lattice_Location i The distance is greater than L i_limit , or for distances less than L i_limit Any Lattice_Location j , Lattice_Location i The distance limit ratio is Ratio i_sum5 Greater than Lattice_Location j The distance limit ratio is Ratio j_sum5 , the Lattice_Location i It can be regarded as the geometric center of the lattice lattice_location i ;

[0088] Step S704: For each lattice_location i , with a radius of 5mm, draw a lattice sphere i The results of screening based on dynamic programming of lattice spacing are as follows Figure 6 As shown, green represents the 16 retained lattices, and yellow represents the 18 eliminated lattices.

[0089] Step S8: Delineate the lattice target area using the selected lattice point coordinates. The SFRT lattice arrangement based on the biological target area is shown in the following figure: Figure 7 As shown, pink is the biological target area, and then biologically guided spatial segmentation is performed.

Claims

1. A method for spatial segmentation lattice target delineation based on functional imaging biological guidance, characterized in that: The following steps are involved: Step S1: setting the scanning parameters of diffusion-weighted imaging (DWI), using a 3T magnetic resonance scanner to perform diffusion-weighted magnetic imaging, and obtaining a DWI image; Step S2: using a two-dimensional rigid sub-pixel image registration algorithm to register the DWI images and obtain transformation parameters between the DWI images; Step S3: The registered image is fitted by two-step weighted linear least squares (LLS) to generate an apparent diffusion coefficient (ADC) parameter image, and image transformation and bilinear interpolation are performed to generate an ADC parameter image with the same matrix size as that of computed tomography (CT); Step S4: Reduce the volume of the gross tumor volume (GTV) through three-dimensional filtering to obtain the reduced tumor target volume GTV decreased ; Step S5: obtaining a lattice point coordinate set lattice_locations in the ADC parameter map through three-dimensional filtering; Step S6: Calculate the lattice spacing thresholds of different voxels through dynamic programming to screen lattice points; Step S601: Setting the minimum lattice spacing L lattice_min and the maximum spacing L lattice_max , the maximum lattice spacing L lattice_max By L search Decision, L. lattice_max =2*L search ; L search The search distance is set; Step S602: construct a linear regression equation between ADC and lattice spacing threshold, and solve the parameters k and c of the linear equation, as shown below: ADC GTV For GTV decreased The collection of ADC values ​​of the voxels within, L lattice_limit =k*ADC+c k=(L lattice_max -L lattice_min ) / (max(ADC GTV )-min(ADC GTV )) c=L lattice_min -min(ADC GTV )*k Step S7: Screen each lattice point coordinate in the lattice point coordinate set lattice_locations to determine whether it meets the lattice setting requirements; Step S701: Screen each lattice point coordinate lattice_location in lattice_locations j , through the linear regression equation L j_limit =k*ADC j +cCalculate the lattice spacing threshold, ADC j lattice_location j The corresponding ADC value; Step S702: for lattice_location j and the coordinates of the five nearest lattice points lattice_location1, lattice_location2…lattice_location5, calculate lattice_location j The sum of the distance limit ratios L n lattice_location j to lattice_location n The geometric distance, n = 1, 2, ..., 5; Step S703: For each lattice point coordinate Lattice_Location in lattice_locations i , satisfying any of the following conditions, any other Lattice_Location j To Lattice_Location i The distance is greater than L i_limit , or for distances less than L i_limit Any Lattice_Location j , Lattice_Location i The distance limit ratio is Ratio i_sum5 Greater than Lattice_Location j The distance limit ratio is Ratio j_sum5 , the Lattice_Location i It can be regarded as the geometric center of the lattice lattice_location i ; Step S704: For each lattice_location i , with a radius of 5mm, draw a lattice sphere i ; Step S8: Delineate the lattice target area using the screened lattice point coordinates to perform biologically guided spatial segmentation.

2. The method for spatial segmentation lattice target delineation based on functional imaging bio-guidance according to claim 1, characterized in that: The step S1 comprises the following steps: Step S101: Set the DWI scanning parameters, the ratio of repetition time to echo time is 6443 / 77msec, the flip angle is 90°, and the field of view is 400×324mm 2 , the acquisition matrix size is 64×94, the reconstruction matrix size is 256×216, 30 slices, the ratio of layer thickness to gap is 5 / 1mm, and the spatial resolution is 0.92×0.92×5mm 3 , the diffusion sensitivity gradient factor b is 0, 10, 25, 50, 75, 100 sec / mm 2 , the number of repetitions is 111; Step S102: Use a 3T magnetic resonance scanner to perform diffusion weighted magnetic imaging to obtain a DWI image.

3. The method for spatial segmentation lattice target delineation based on functional imaging bio-guidance according to claim 2, characterized in that: The step S2 comprises the following steps: Step S201: taking the DWI image when b is 0 as the reference image, using a two-dimensional rigid sub-pixel image registration algorithm to achieve registration between DWI images with different b values, and obtaining transformation parameters between DWI images; Step S202: Assume that there are two images I and J, where J is the image of I after translation, rotation and scaling. The coordinates on image I are (x, y), the coordinates on image J are (x', y'), and the translation parameter is (t x ,t y ), the rotation parameter is θ, the scaling parameter is s, and the transformation relationship is as follows: x′=s*(x*cos(θ)-y*sin(θ))+t x y′=s*(x*sin(θ)+y*cos(θ))+tγ; Step S203: Evaluate the similarity between image I and image J by using a similarity measurement function; Step S204: Minimize the similarity metric function by using a gradient descent optimization algorithm or a least squares method to obtain translation, rotation and scaling parameters.

4. The method for spatial segmentation lattice target delineation based on functional imaging bio-guidance according to claim 3, characterized in that: The step S3 comprises the following steps: Step S301: For the registered DWI image, a two-step weighted linear least squares fit is performed using the formula Log(S b )=Log(S0)-b*ADC Calculate the ADC parameter image at the voxel level, and the weight of the two-step weighted linear least squares is N(b)*S LLS (b) 2 , b is the diffusion sensitive gradient factor, S b and S0 are the signal intensities at a given b value and b = 0, respectively, and N(b) is the number of repeated acquisitions at a given b value; Step S302: After the ADC parameter image is generated, the CT image is used as the reference image, and the DWI image with b=500 is registered by a two-dimensional rigid sub-pixel image registration algorithm. The same image transformation is applied to the ADC parameter image, and an ADC parameter image with the same matrix size as the CT is generated by bilinear interpolation.

5. The method for spatial segmentation lattice target delineation based on functional imaging bio-guidance according to claim 4, characterized in that: The step S4 comprises the following steps: Step S401: according to the set retraction distance L scale and spatial dimension resolution X space , Y space , Z space , generate spheroid filter spheroid scale , spheroid scale The semi-major axis a spheroid_scale =L scale / X space , b spheroid_scale =L scale / Y space , c spheroid_scale =L scale / Z space ; Step S402: When filtering the GTV binary matrix, the stride is 1 pixel in each of the three dimensions, and the reduced tumor target area GTV is generated by judging whether the current center is equal to the minimum filtering value. decreased .

6. The method for spatial segmentation lattice target delineation based on functional imaging bio-guidance according to claim 5, characterized in that: The step S5 comprises the following steps: Step S501: According to the set search distance L search and spatial dimension resolution X space , Y space , Z space Construct spheroid filter search , spheroid search The semi-major axis a spheroid_search =L search / X space , b spheroid_search =L search / Y space , c spheroid_search =L search / Z space ; Step S502: By decreased The ADC parameter map in the ADC parameter map is filtered with a step of 1 in three dimensions. When the center is the minimum value of the filter, the lattice point coordinate set lattice_locations in the ADC parameter map is retrieved.

Images