Boron neutron capture therapy blood boron concentration detection method
Through adaptive registration and boron concentration calculation of PET and CT images, the problem of difficult monitoring of the boron concentration distribution of blood and tumor tissue in boron neutron capture treatment was solved, and the uniformity and accuracy of the treatment effect were achieved.
Patent Information
- Application Number
- CN202411496685.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2024-01-08
- Filing Date
- 2024-10-25
- Publication Date
- 2024-11-26
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In boron neutron capture treatment, it is difficult for the prior art to accurately detect the distribution of boron concentration in blood and tumor tissue, resulting in uneven treatment effects, especially the changes in boron concentration at the edge of the organ or region of interest are difficult to monitor, affecting the treatment effect.
Boron concentration is calculated by adaptive high-precision registration of PET and CT images, combined with standard density and material composition information reported by ICRU, and segment the tumor areas by pixel gradient or geometric adaptive scaling to evaluate boron concentration and dose in different regions. .
High-precision and accurate boron concentration detection is achieved, ensuring uniformity in the treatment effect of tumor margins and organ areas, and reducing the risk of tumor recurrence.
Smart Images

Figure CN119027415A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the field of medical image processing, and in particular to a method for detecting blood boron concentration in boron neutron capture therapy. Background Art
[0002] Boron Neutron Capture Therapy (BNCT) is a radiotherapy technique commonly used to treat certain types of cancer (melanoma, glioblastoma). In BNCT treatment, the patient is injected with a drug containing B-10 that can be specifically absorbed by tumor cells (such as BPA or BSH), and then the affected area is irradiated with a neutron beam. Boron-10 absorbs neutrons and produces short-range α and Li ions. The range of these ions in tumor cells is in the order of um (the same order of magnitude as the cell diameter), which only harms nearby cancer cells, thereby reducing damage to surrounding normal tissues. Compared with traditional radiotherapy, BNCT technology has higher accuracy and fewer side effects.
[0003] In BNCT, blood boron concentration is an important concept. Blood boron concentration refers to the ratio of boron concentration in the blood to boron concentration in other organs. In BNCT treatment, patients need to receive injections of drugs containing boron-10 (BPA, BSH, etc.), which will be absorbed by the patient's tumor cells, thereby enhancing the therapeutic effect during neutron beam irradiation. However, if the boron concentration in the blood is too high, boron-10 will interact with boron in the blood during neutron beam irradiation, thereby weakening the therapeutic effect. In clinical treatment, the greater the ratio of boron concentration in the tumor to boron concentration in the blood, the better. Therefore, in BNCT, the detection of blood boron concentration is particularly important.
[0004] In the known technology, when simulating the dose of boron neutron capture therapy, the boron concentration in the human body is mostly set based on biological biopsy or theoretical ideal boron concentration reference value. However, in clinical treatment, biological biopsy cannot accurately obtain the distribution of boron concentration in the human body, and it is difficult to implement biological biopsy of tumor tissue (biological biopsy is a method of taking a slice of the tissue to be tested in the human body for chemical testing); in addition, when setting the boron concentration of a specific organ or interest in clinical treatment, if the ideal boron concentration reference value is used, it may be very different from the actual situation of the patient.
[0005] It is worth noting that in boron neutron capture therapy, according to boron pharmacokinetics, the dynamic exchange constants between different organs and tissues or blood are different, which will cause a step-by-step change in boron concentration at the edge of the organ or tissue. The drastic change in boron concentration will cause the adjacent areas belonging to different organs to have very different treatment effects when receiving neutron beam irradiation. If the boron concentration at the edge of the tumor decreases in a step-by-step manner, the killing effect on cancer cells at the edge of the tumor will be greatly reduced, which will lead to secondary recurrence of the tumor. Therefore, it is also particularly important to detect the distribution of boron concentration at the edge of the organ or region of interest and its changing trend. Summary of the invention
[0006] In order to overcome the shortcomings of the prior art, the present invention proposes a method for detecting blood boron concentration in boron neutron capture therapy. The technical solution of the present invention is implemented as follows: A method for detecting blood boron concentration in boron neutron capture therapy comprises the following steps:
[0007] S1: Read in the DICOM images obtained by PET scanning and the DICOM images obtained by CT scanning respectively;
[0008] S2: Adaptively register the imported CT image and PET image with high precision and accuracy, including the following steps:
[0009] S21: image preprocessing;
[0010] S22: feature extraction, extracting key points in the PET image and the CT image and local features around the key points;
[0011] S23: feature matching, finding feature pairs in the PET image and the CT image that have similar descriptors and are close in space;
[0012] S24: initial transformation estimation, estimating the transformation that maps one image to another image based on the identified feature pairs;
[0013] S25: Transformation detail optimization, which refines the initial transformation to improve its accuracy and robustness;
[0014] S26: Quality assessment, evaluating the accuracy and robustness of the transformation after registration;
[0015] S27: If the quality evaluation parameter does not reach the set value or limit value, perform parameter optimization, obtain secondary transformation, and repeat steps S24-S26 until the quality evaluation reaches the set value or limit value;
[0016] S3: Import the target area information of the CT image outlined by the diagnosis and treatment planning system, normalize the PET image after adaptive image registration according to the relevant information read from the PET header file, map the CT target area to the registered PET image, calculate the weighted average of each target area or region of interest, and then convert the PET pixel value unit into the unit of boron concentration by comparing it with the standard density and material composition information of each organ in the international standard in the ICRU report.
[0017] Furthermore, S3 also includes the steps of further optimizing the boron concentration distribution model, shrinking the organ target area outlined by the CT image based on pixel gradient or geometric adaptation in all directions, shrinking to different sizes, dividing the tumor from the inside to the outside into several layers, and evaluating the boron concentration and dose in each area of the "layered" tumor after shrinkage and segmentation to ensure the evaluation of the degree of damage to different areas of the tumor.
[0018] Furthermore, the image preprocessing in S21 includes the steps of removing noise in the image and enhancing the contrast of the image, and the image preprocessing method is one of filter, sharpening, and histogram equalization or a combination thereof.
[0019] Furthermore, the feature extraction method in step S22 includes but is not limited to edge detection, corner point detection, scale-invariant feature transformation and histogram, including the steps of:
[0020] S221: define feature extraction algorithm;
[0021] S222: Detect key points in the image;
[0022] S223: Extract local features around the key points.
[0023] Furthermore, the feature extraction algorithm in S221 includes but is not limited to SIFT, SURF, ORB, HOG and LBP.
[0024] Furthermore, the key point detection algorithm in S222 includes but is not limited to Harris corner point detection, FAST corner point detection and SIFT scale space extreme value detection.
[0025] Furthermore, the extracting of local features around the key points in S223 is to extract descriptors of pixels around the key points, which can describe the color, texture and shape around the key points. The local features include SIFT descriptors, SURF descriptors and ORB descriptors.
[0026] Furthermore, the feature matching algorithms in S23 include but are not limited to nearest neighbor, random sample consistency and iterative closest point.
[0027] Furthermore, the algorithm for the initial transformation estimation in S24 includes but is not limited to the least squares method, the normalized least squares method and the iterative method.
[0028] Furthermore, the algorithm for optimizing the conversion details in S25 includes but is not limited to gradient descent, Gauss-Newton or Levenberg-Marquardt.
[0029] Furthermore, the quality assessment algorithm in S26 includes but is not limited to mean square error, correlation coefficient, mutual information or Dice similarity coefficient. When performing quality assessment, calculation time, memory usage and sensitivity to noise and artifacts should also be considered.
[0030] The beneficial effect of the present invention is that, compared with the prior art, the present invention can realize high-precision and high-accuracy adaptive image registration of PET images and CT images, and can realize the acquisition of boron concentration information of different organs with one click by only importing PET and CT scan images and CT image target area delineation results, and can also obtain the boron concentration of the range wrapped by the region of interest after scaling of a custom geometric-based region of interest. This function is conducive to evaluating the relevant dose of tumor margins or critical organs. BRIEF DESCRIPTION OF THE DRAWINGS
[0031] Figure 1 The present invention is a flow chart of a method for detecting blood boron concentration in boron neutron capture therapy. DETAILED DESCRIPTION
[0032] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0033] The purpose of the present invention is to provide a method for detecting the boron concentration in the blood of a boron neutron capture therapy based on adaptive image registration and focusing on the boron concentration distribution at the edge of a region of interest. The idea of the invention is based on adaptive image registration. After the PET image and the CT image are registered with high quality and high precision, the high resolution and pixel value of the CT image are combined with the characteristics of the composition of the human organ material and the PET image that can display the distribution characteristics of the radionuclide-labeled drug (in this technology, the radionuclide-labeled drug is 18-BPA) in the human body. After registration, the organ or target area outlined in the CT image is regarded as the range of the corresponding target area of the PET image, and the pixel value of the PET image is normalized to obtain the concentration of the radionuclide-labeled drug (18-BPA) in the human body. For the region of interest, the corresponding outlined target area is adaptively scaled based on pixel gradient (pixel value size is proportional to boron concentration) or based on geometric size, and then the boron concentration of the range between the unscaled and scaled contours of the region of interest is obtained.
[0034] See also Figure 1 The present invention provides a method for detecting blood boron concentration in boron neutron capture therapy, comprising the steps of:
[0035] S1: Read in the DICOM images obtained by PET scanning and the DICOM images obtained after CT scanning respectively; according to the DICOM files of human medical images obtained, DICOM (Digital Imaging and Communications in Medicine), namely medical digital imaging and communications, is the international standard for medical images and related information (ISO 12052), read in the DICOM images obtained by PET scanning and the DICOM images obtained after CT scanning respectively.
[0036] S2: Adaptively register the imported CT image and PET image with high precision and accuracy, including the following steps:
[0037] S21: Image preprocessing: Before registering images, it is important to preprocess them to ensure that they have similar size, resolution, and orientation. The preprocessing process includes: removing noise from the image, enhancing the contrast of the image, etc. Preprocessing methods include filters, sharpening, histogram equalization, etc.
[0038] Noise reduction: Due to the influence of factors such as image sensors, imaging devices, and image transmission, there is often some noise in the image. Noise reduction is aimed at removing this noise to reduce the impact on subsequent processing. Commonly used noise reduction methods include mean filtering, median filtering, Gaussian filtering, wavelet transform, etc.
[0039] Image enhancement: Image enhancement refers to making an image visually clearer, brighter, and more contrasty through some methods. Common image enhancement methods include histogram equalization, contrast enhancement, sharpening, etc. Image scaling and cropping: Image scaling refers to adjusting the size of an image, which is often used in image compression, image analysis, matching, etc. Cropping refers to extracting a sub-image from the original image. Common image scaling and cropping methods include bilinear interpolation, nearest neighbor interpolation, bicubic interpolation, etc.
[0040] S22: Feature extraction, extracting key points in the PET image and the CT image and local features around the key points; the first step of registration is to extract features from the image that can be used to match corresponding points or regions. Common feature extraction techniques include edge detection, corner detection, scale-invariant feature transform (SIFT) and histogram-based methods. The following steps are included:
[0041] S221: Define feature extraction algorithm: The feature extraction algorithm is the core of the feature extraction process. Different algorithms are suitable for different application scenarios. Common feature extraction algorithms include SIFT, SURF, ORB, HOG, LBP, etc. These algorithms have their own characteristics and advantages. It is necessary to select the appropriate algorithm according to the actual application scenario;
[0042] S222: Detect key points in the image: Key points refer to points with significant features in the image, such as edges, corners, spots, etc. Before feature extraction, it is necessary to first detect the key points in the image. Commonly used key point detection algorithms include Harris corner detection, FAST corner detection, SIFT scale space extreme value detection, etc.
[0043] S223: Extract local features around key points: After detecting the key points, it is necessary to extract local features around the key points. These local features are usually descriptors of the pixels around the key points, which can describe the color, texture, shape and other information around the key points. Commonly used local features include SIFT descriptors, SURF descriptors, ORB descriptors, etc.
[0044] S23: Feature matching, finding feature pairs in the PET image and the CT image that have similar descriptors and are close in space; after extracting the features, the next step is to match between the images, which involves finding feature pairs that have similar descriptors and are close in space. Common feature matching algorithms include nearest neighbor, random sample consistency (RANSAC), and iterative closest point (ICP).
[0045] S24: Initial transformation estimation, based on the identified feature pairs, estimate the transformation that maps one image to another image; once the corresponding feature pairs are identified, the next step is to estimate the transformation that maps one image to another image. The transformation can be affine (i.e., translation, rotation, scaling, shearing) or non-rigid (i.e., deformation). Common transformation estimation algorithms include least squares, normalized least squares, and iterative methods.
[0046] S25: Transformation detail optimization, refine the initial transformation to improve its accuracy and robustness; Having estimated the initial transformation, it may be necessary to refine it to improve its accuracy and robustness. This can be done using iterative optimization algorithms that minimize the distance between transformed images, such as gradient descent, Gauss-Newton, or Levenberg-Marquardt.
[0047] S26: Quality assessment, evaluate the accuracy and robustness of the transformation after registration; After registration, it is important to evaluate its accuracy and robustness. This can be done using metrics such as mean square error, correlation coefficient, mutual information, or Dice similarity coefficient (DSC). The evaluation should also consider factors such as computational time, memory usage, and sensitivity to noise and artifacts.
[0048] S27: If the quality evaluation parameter does not reach the set value or limit value, perform parameter optimization, obtain secondary transformation, and repeat steps S24-S26 until the quality evaluation reaches the set value or limit value.
[0049] S3: Import the target area information of the CT image outlined by the diagnosis and treatment planning system, normalize the PET image after adaptive image registration according to the relevant information read from the PET header file, map the CT target area to the registered PET image, calculate the weighted average of each target area or region of interest, and then convert the PET pixel value unit into the unit of boron concentration ppm by comparing it with the standard density and material composition information of each organ in the international standard in the ICRU report.
[0050] The boron concentration distribution model can be further optimized. Since boron is unevenly distributed in organs or tumors, the boron concentration at the boundary of organs or tumors is usually significantly lower than that in the central area of the tumor. In order to evaluate the killing ability at the edge of the tumor, the organ target area outlined by the above CT image can be shrunk in all directions based on pixel gradient (pixel value is proportional to boron concentration) or based on geometric adaptive shrinkage of different sizes, and the tumor can be divided into several layers from the inside to the outside. The boron concentration and dose in each area of the "layered" tumor after the shrinkage and segmentation are evaluated to ensure the evaluation of the degree of killing of different areas (edge, center) of the tumor.
[0051] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.
Claims
1. A method for detecting blood boron concentration in boron neutron capture therapy, characterized in that: Includes steps: S1: Read in the DICOM images obtained by PET scanning and the DICOM images obtained by CT scanning respectively; S2: Adaptively register the imported CT image and PET image with high precision and accuracy, including the following steps: S21: image preprocessing; S22: feature extraction, extracting key points in the PET image and the CT image and local features around the key points; S23: feature matching, finding feature pairs in the PET image and the CT image that have similar descriptors and are close in space; S24: initial transformation estimation, estimating the transformation that maps one image to another image based on the identified feature pairs; S25: Transformation detail optimization, which refines the initial transformation to improve its accuracy and robustness; S26: Quality assessment, evaluating the accuracy and robustness of the transformation after registration; S27: If the quality evaluation parameter does not reach the set value or limit value, perform parameter optimization, obtain secondary transformation, and repeat steps S24-S26 until the quality evaluation reaches the set value or limit value; S3: Import the target area information of the CT image outlined by the diagnosis and treatment planning system, normalize the PET image after adaptive image registration according to the relevant information read from the PET header file, map the CT target area to the registered PET image, calculate the weighted average of each target area or region of interest, and then convert the PET pixel value unit into the unit of boron concentration by comparing it with the standard density and material composition information of each organ in the international standard in the ICRU report.
2. The method for detecting blood boron concentration in boron neutron capture therapy according to claim 1, characterized in that: The S3 also includes the steps of further optimizing the boron concentration distribution model, shrinking the organ target area outlined by the CT image based on pixel gradient or geometric adaptive all-round, shrinking to different sizes, dividing the tumor from the inside to the outside into several layers, and evaluating the boron concentration and dose in each area of the "layered" tumor after shrinking and segmentation to ensure the evaluation of the degree of damage to different areas of the tumor.
3. The method for detecting blood boron concentration in boron neutron capture therapy according to claim 1, characterized in that: The image preprocessing in S21 includes the steps of removing noise in the image and enhancing the contrast of the image. The image preprocessing method is one of filter, sharpening, and histogram equalization or a combination thereof.
4. The method for detecting blood boron concentration in boron neutron capture therapy according to claim 1, characterized in that: The feature extraction method in step S22 includes but is not limited to edge detection, corner point detection, scale-invariant feature transformation and histogram, including the steps of: S221: define feature extraction algorithm; S222: Detect key points in the image; S223: Extract local features around the key points.
5. The method for detecting blood boron concentration in boron neutron capture therapy according to claim 4, characterized in that: The feature extraction algorithm described in S221 includes but is not limited to SIFT, SURF, ORB, HOG and LBP.
6. The method for detecting blood boron concentration in boron neutron capture therapy according to claim 4, characterized in that: The key point detection algorithms described in S222 include but are not limited to Harris corner point detection, FAST corner point detection and SIFT scale space extreme value detection.
7. The method for detecting blood boron concentration in boron neutron capture therapy according to claim 4, characterized in that: The extracting of local features around the key points in S223 is to extract descriptors of pixels around the key points, which can describe the color, texture and shape around the key points. The local features include SIFT descriptors, SURF descriptors and ORB descriptors.
8. The method for detecting blood boron concentration in boron neutron capture therapy according to claim 1, characterized in that: The feature matching algorithms in S23 include but are not limited to nearest neighbor, random sample consistency and iterative nearest point.
9. The method for detecting blood boron concentration in boron neutron capture therapy according to claim 1, characterized in that: The algorithm for the initial transformation estimation in S24 includes but is not limited to the least squares method, the normalized least squares method and the iterative method.
10. The method for detecting blood boron concentration in boron neutron capture therapy according to claim 1, characterized in that: The algorithm for optimizing the conversion details in S25 includes, but is not limited to, gradient descent, Gauss-Newton or Levenberg-Marquardt.
11. The method for detecting blood boron concentration in boron neutron capture therapy according to claim 1, characterized in that: The quality assessment algorithm in S26 includes but is not limited to mean square error, correlation coefficient, mutual information or Dice similarity coefficient. When performing quality assessment, calculation time, memory usage and sensitivity to noise and artifacts should also be considered.
Citation Information
Patent Citations
Image registration method based on automatic sketching
CN113298855A
Radiotherapy effect evaluation method based on PET and CT elastic registration technology
CN114820432A
Head and neck tumor target area sketching system for whole-body PET / CT (positron emission tomography / computed tomography) scanning
CN117351489A
Method for measuring concentration distribution of boron for BNCT using MRI, and treatment planning method for bnct
WO2018080290A1
