Image identifying system and method

Abstract

The present invention provides a novel image identifying system and method, comprising: an anatomical image module for reading anatomical images; a contour module for identifying the human body contour based on the anatomical images; a pixel value module for setting a first pixel value and a second pixel value, which correspond and filter the first pixel region and the second pixel region within the human body contour respectively. Finally, an identification and analysis module for expanding the first pixel region and the second pixel region outward by a voxel range to form a first expanded region and a second expanded region. The first expanded region is then adjusted by excluding the overlapping area between the first expanded region and the second expanded region to form a preprocessing region. Additionally, the pixel value setting module defines a mucosal pixel value, allowing the identification and analysis module to recognize the mucosal region within the preprocessing region based on the mucosal pixel value. The voxel range is set between 0.2mm and 10mm.

Description

Image calibration system and method This case proposes an image calibration system and method, more specifically, an image calibration system and method for enhancing the identification and calibration of mucosal or soft tissues in anatomical images through image processing technology. Boron neutron capture therapy (BNCT) is an innovative radiotherapy for cancer that uses boron-containing drugs (such as those using amino acids as boron carriers). These drugs are designed to specifically accumulate in the patient's tumor cells, while normal cells absorb less. When a patient receives epithermal neutron radiation, the tumor cells with accumulated boron-containing drugs absorb neutrons more readily than normal healthy cells. When the neutrons interact with the boron-containing drugs, a nuclear reaction occurs, producing lithium nuclei (…). 7 Li) and alpha particles, generating alpha particles and lithium nuclei ( 7 Li atoms have an extremely short range (about 5-9 micrometers) that can destroy tumor cells. This range is roughly equal to the diameter of a cell, allowing energy to be concentrated inside the tumor cells with minimal damage to adjacent healthy tissues. Therefore, BNCT treatment has advantages such as high selectivity for tumor sites, low toxicity, fewer treatment sessions than traditional radiotherapy, less burden on the human body, and faster recovery. For traditional radiation medicine, or the aforementioned boron neutron capture therapy, the use of medical imaging is crucial. This is because it helps to accurately locate the patient's tumor, protect normal tissue during treatment, assess treatment effectiveness, and develop new medical tools in the medical engineering field. Medical imaging is currently mainly divided into anatomical imaging (such as computed tomography and magnetic resonance imaging, providing information about physical geometry) and functional imaging (such as positron emission tomography, gamma scintillation detectors, and single-photon emission computed tomography, providing information about the distribution of targeted drugs or other calibrated substances in the human body). In conventional techniques, computed tomography (CT) scans are typically used as anatomical images to assess the distribution of organs, tissues, tumors, bones, mucous membranes, and air within the patient's body. Medical personnel usually use image processing techniques to automatically or manually annotate regions of interest (ROIs) to provide the input data required for treatment planning simulations. These data include parameters such as CT numbers (in Hounsfield units), tissue composition, and density, which are used to describe the composition of the human body and to further calculate the dose distribution for the patient during actual treatment. For example, let's discuss Chinese patent CN101458826B, which proposed a digital human body modeling method in 2008 using anatomical imaging technology and CT values ​​to assign density and composition. The method steps of this '826 patent include the following: collecting image data in both CT images and color photographs; manually outlining the outer contours of various organs and tissues on the color photographs and marking them with different colors to obtain geometric information about the shape, size, and location of each organ and tissue; then, superimposing the CT images and color photographs; obtaining the outer contour information of organs and tissues on the CT images based on the outlined organ and tissue contours segmented from the color photographs; next, calculating the composition of unknown voxels using linear interpolation based on the CT values, density, and composition of known voxels; and finally, generating a data input file that meets the requirements for radiation dose calculation based on the obtained density and composition information for subsequent simulation calculations. However, conventional anatomical image processing techniques have several problems: 1. The selection of regions of interest (ROIs) on CT images manually by different medical personnel is easily influenced by their prior knowledge and subjective judgment, leading to inconsistent annotation results and thus errors; 2. Anatomical images alone are insufficient to provide comprehensive biological information and may have limitations; 3. Mucosal regions or soft tissues are often difficult to determine solely from functional images, and there is currently no method to define these regions using explicit parameters (because mucosal or soft tissue regions are relatively fragile, the radiation dose during X-ray imaging is limited, resulting in generally less than ideal resolution, making direct manual or automated interpretation difficult). Therefore, to improve these problems, the field of radiology, especially in the boron neutron capture therapy market, still urgently needs an image calibration system and method to effectively improve the accuracy and consistency of image analysis and enhance the scientific rigor and reliability of treatment plans. For the reasons mentioned above, the main objective of this invention is to provide a novel image calibration system and method. Based on anatomical images (e.g., CT images), it proposes a specific parameter setting for calibrating organs, tissues, tumors, bones, mucosal regions, and air. These parameters conform to the structure of human anatomy, making the output results more aligned with treatment needs. Another objective of this invention is to automate the above method through an automated analysis system, improving upon the limitations of traditional anatomical image interpretation, which requires manual positioning by skilled medical engineers or physicians, and the poor automated interpretation results due to the limited photon dose from anatomical image irradiation. This also addresses the shortcomings of conventional techniques, such as susceptibility to human error and lack of objective and unified standards for judgment. A further objective of this invention is to compensate for the insufficient tissue resolution of anatomical images by performing image fusion (registration) between functional and anatomical images, thereby improving the accuracy and applicability of treatment planning. The detailed technical solution of this invention, in summary, is described below. To achieve the purpose of this invention, in a broad embodiment, the present invention provides an image calibration system comprising the following system components: an anatomical image module for reading anatomical images; a contour module for identifying human body contours based on the anatomical images; a pixel value module for setting a first pixel value and a second pixel value, respectively corresponding to and filtering out the first pixel region and the second pixel region in the human body contour; and an identification and analysis module connected to the pixel value module, protruding the first pixel region and the second pixel region outward by a voxel range to form a first expanded region and a second expanded region, and subtracting the intersection area of ​​the first expanded region and the second expanded region from the first expanded region to form a preprocessing region; wherein, the pixel value setting module sets a mucosal pixel value, enabling the identification and analysis module to identify the mucosal region in the preprocessing region based on the mucosal pixel value and output the identification result, wherein the voxel range is between 0.2mm and 10mm. To achieve the objectives of this invention, the present invention also provides a novel image calibration method, comprising the following steps: preparing an anatomical image; identifying the human body contour using a contour module based on the anatomical image; defining a first pixel region and a second pixel region in the human body contour using a pixel value module based on a first pixel value and a second pixel value; protruding the first pixel region and the second pixel region outward by at least one voxel range respectively using an identification and analysis module to form a first expanded region and a second expanded region; subtracting the area where the first expanded region and the second expanded region intersect using the first expanded region, and defining the area remaining after subtraction as a preprocessing region; and processing the preprocessing region by setting a mucosal pixel value range, thereby identifying the mucosal region in the anatomical image, wherein the voxel range is between 0.2 mm and 10 mm. According to an embodiment of the present invention, the setting of the first pixel value and the second pixel value can be determined by selecting any region such as the required organ, tissue, tumor, bone, or air, as needed for the application. In the above embodiment, the first pixel region is defined based on the location of air, and the second pixel region is defined based on the location of bone, as an example for illustration. This invention will be described in detail with reference to preferred embodiments and viewpoints so that the reader may fully understand how these embodiments are implemented. However, those skilled in the art will understand that this invention can also be implemented without these details. Furthermore, this invention can also be used and implemented through other specific embodiments, and the various details set forth in this specification can be applied based on different needs, and various modifications or changes can be made without departing from the spirit of this invention. Therefore, this invention will be described with reference to preferred embodiments and viewpoints. Such descriptions are for explaining the structure of the invention and are only for illustration and not for limiting the scope of the patent application of this invention. The terminology used in the following description will be interpreted in the broadest reasonable sense. Those skilled in the art can adjust the structure of this invention to meet the needs of actual industry according to the requirements of manufacturing or therapeutic applications, which is hereby stated in advance. The terms "about," "approximately," or "substantially" as used in this invention generally refer to within 20 percent of a given value or range, preferably within 10 percent, and more preferably within 5 percent. The values ​​given herein are approximate, meaning that unless explicitly stated otherwise, the meaning of the terms "about," "approximately," or "substantially" can be inferred. Furthermore, the terms "front and back," "left and right," and "up and down" as used in this invention can be understood in conjunction with the accompanying drawings. Unless otherwise stated in this invention, the meaning can be obtained from the context of the description after reading it. Finally, for the sake of simplicity, some conventional structures and components are shown in a simplified schematic manner in the drawings, and for ease of viewing, the dimensions of the components in the drawings are not drawn to scale. Furthermore, in the description of this invention, the method procedure numbers described are only the procedure numbers of a certain method flow unless specifically indicated in the description of this invention. They are used only for clarity and convenience in the description, and are not used to describe the sequential order of the method flow. This is also stated in advance. Please refer to Figures 1, 6, and 7. This invention provides an image calibration system 100, comprising: an anatomical image module 110 for reading anatomical images; a contour module 120 for identifying human body contours based on anatomical images; a pixel value module 130 for setting a first pixel value and a second pixel value, corresponding to the first pixel region 620 and the second pixel region 630 in the human body contour, respectively; and an identification analysis module 140 connected to the pixel value module 130, protruding the first pixel region 620 and the second pixel region 630 outward by a voxel range Δ to form a first expanded region 640 and a second expanded region 650, and subtracting the area where the first expanded region 640 and the second expanded region 650 intersect. The pixel value setting module 130 sets a mucosal pixel value, enabling the identification analysis module 140 to identify the mucosal region 610 in the preprocessed area based on the mucosal pixel value, and outputting the identification result. The voxel range Δ is between 0.2 mm and 10 mm. In one embodiment of the present invention, referring to Figures 1 and 6, an example of the application of the image calibration system 100 is taken using the oral mucosa. The voxel range Δ can be set between 0.2 mm and 4 mm. Since the oral mucosa is located between the lining of the oral cavity and the nasal cavity from an anatomical perspective, the identification target of the image calibration system 100 is to output the oral mucosa in the range from the nasal cavity to the epipharynx in the anatomical image as an identification result (CT images are also used as one embodiment of the anatomical image in this embodiment). It should be noted that, according to the embodiments of the present invention, the setting of the first pixel value and the second pixel value can be customized according to the application needs, and any region such as organs, tissues, tumors, bones, and air can be selected for definition. In the above embodiment, air is selected to define the first pixel region, and bone is selected to define the second pixel region. Accordingly, the anatomical imaging module 110 reads an anatomical image, which is synthesized from X-rays using optical slices at various angles, and transmits the anatomical image to the contour module 120. The contour module 120 identifies the human body contour based on the anatomical image file, as a medically meaningful region of interest (ROI). Next, the pixel value module 130 sets a first pixel value and a second pixel value. In one embodiment of the present invention, each pixel value in the ROI is a CT value expressed in Henry units (of course, in other embodiments, different types of anatomical images, such as magnetic resonance imaging, can also be processed using pixel values ​​commonly used in the medical field or defined as needed). In this embodiment, the first pixel value and the second pixel value are respectively used to filter out the first pixel region 620 and the second pixel region 630 in the human body contour, so as to filter out the range of air and bone in the human body contour of the anatomical image. Accordingly, the identification and analysis module 140, based on the setting results of the first pixel region 620 and the second pixel region 630, protrudes outward by a voxel range Δ. The voxel range Δ can be defined according to the application requirements. In a typical embodiment, such as the identification of the mucosal region 610 of the oral mucosa, it can be between 0.2mm and 4mm, so that after protrusion, it forms a first expansion region 640 and a second expansion region 650. The first expansion region 640 is then subtracted from the area where the first expansion region 640 and the second expansion region 650 intersect. The pixel value setting module 130 sets a mucosal pixel value, so that the identification and analysis module 140 can identify the mucosal region 610 in the preprocessed area based on the mucosal pixel value and output the identification result. According to one aspect of the present invention, the aforementioned voxel range Δ for the mucosal region 610 can be defined according to the required application location. In this embodiment, that is, for the oral mucosa applied in oral medicine, since the hard palate is usually thicker, and considering that the oral mucosal region also contains biological tissues such as the buccal mucosa and the dorsal surface of the tongue, 0.2mm-4mm is a better parameter setting for processing the oral mucosa. This specific parameter setting can achieve the purpose of standardization and automation proposed in the present invention.According to another aspect of the present invention, since the oral mucosa is used as the basis for the image calibration system 100, the location of the sinus may be misjudged by the image calibration system 100 during the formation of the first expansion region 640. Therefore, in the subsequent processing, the sinus portion can be filtered out by subtracting the area where the first expansion region 640 and the second expansion region 650 intersect, so as to avoid misjudgment. This approach makes the processing of the image calibration system 100 more in line with the principles of anatomy. Furthermore, it should be noted that the aforementioned 0.2mm-4mm is merely one embodiment, used only to illustrate the method of marking the oral mucosa in this invention. If it is necessary to mark the mucosa of other tissues, such as the intestines, stomach, uterus, foreskin, respiratory tract, vagina, conjunctiva, etc., an appropriate range can be defined based on the actual anatomical structure. In a broader sense, the voxel range Δ selected for the overall system application can be chosen from 0.2mm to 10mm to adapt to different embodiments. Those skilled in the art can modify it according to common knowledge after reading the description of this invention, which is specifically stated here. Please refer to Figures 2A and 6. According to one embodiment of the present invention, the identification and analysis module 140 further includes a smoothing unit 140A. When the mucosal region 610 is generated in the region of interest (ROI), the smoothing unit 140 performs smoothing and noise reduction processing on the edges of the mucosal region 610 through a smoothing algorithm to eliminate jagged edges of the mucosal region 610 in the image. The smoothing algorithm used by the smoothing unit 140A can be selected from, but is not limited to, Gaussian blur, bilateral filter, mean filter, median filter, guided filter, non-local means (NLM), anisotropic diffusion ( / perona-malik filter), curvature-driven diffusion (CDD), and other smoothing algorithms. The selection principle of the smoothing algorithm can be based on computing resources, computing time, whether real-time calculation is required, the amount of image details to be retained, and the quality of noise processing. Referring to Figures 1 and 8, according to one embodiment of the present invention, the image calibration system 100 further includes a functional imaging module 160 for reading functional images; a tissue analysis module 170 for identifying and calibrating mucosal locations 660 in the functional images using image-visible tumor volume (GTV) and standardized uptake value (SUV); and a fusion analysis module 180, coupled to the identification analysis module 140, for superimposing (registration) the functional images and anatomical images, taking the union of the mucosal region 610 and the mucosal location 660 to form a calibration result, and outputting the calibration result. In this embodiment of the present invention, the functional image is exemplified by a positron emission tomography (PET) image (hereinafter referred to as a PET image), and the specific source of the functional image may be, but is not limited to, positron emission tomography, gamma scintillation detector, or single-photon emission computed tomography. According to one aspect of the present invention, since PET imaging uses isotopes ingested by the patient to mark lesions, the principle is that during the β decay of the isotope, the detector detects the annihilation reaction produced by the combination of the positron emitted by β decay with the surrounding electrons, and the emitted photon localization signal source (i.e., the lesion, such as a tumor) is generated. Since the isotope tracker will accumulate at specific marked locations or areas with high blood flow and metabolism, it will have better resolution for marking the location of mucosa or soft tissue within the region of interest (ROI). Therefore, by combining the mucosal location 660 in the functional image and the mucosal region 610 marked in the anatomical image, the location of the mucosa can be more accurately identified, achieving the purpose of the present invention of improving tissue resolution by fusing functional and anatomical images. According to one embodiment of the present invention, the tissue analysis module 170 identifies and marks the mucosal location 660 in functional images using the gross tumor volume (GTV) and standardized uptake value (SUV). Specifically, after subtracting the region of gross tumor volume from the region of gross tumor volume in the functional image where the standardized uptake value approximates the tumor tissue value, the mucosal location 660 can be obtained. Based on one aspect of the present invention and related general medical knowledge, in functional images, i.e., the PET images of this embodiment, the standardized uptake value of the mucosa is approximately approximate to that of the tumor tissue, i.e., the tumor-to-normal tissue ratio (T / N value). For example, assuming the current T / N value is 2.5, in this embodiment, after extracting the region with a standardized uptake value greater than or equal to 2.5 and subtracting the region of gross tumor volume, the mucosal location 660 can be obtained in the functional image. Similarly, it should be noted that the T / N value of 2.5 mentioned above is only an example parameter and is not used to limit the selection of T / N value. Other parts of the human body, such as the intestines, stomach, uterus, foreskin, respiratory tract, vagina, conjunctiva, etc., can be modified or changed by those skilled in the art according to common medical knowledge after reading the description of this invention, so that the tissue analysis module 170 can be adjusted according to the needs of the application. According to one aspect of the present invention, the image calibration system 100 further includes an image output module 150, coupled with an identification and analysis module 140 and a fusion analysis module 180, which outputs the identification results of the mucosal region 610 in the anatomical image and the mucosal position 660 in the functional image. The image output module 150 can be deployed as a proximal or distal system element in the image calibration system 100. If proximal, the identification results of the mucosal region 610 and the mucosal position 660 can be output and stored in the storage device of a local computer system; if distal, they can be output and stored in a cloud or a data storage server system. Please refer to Figure 2B. According to one embodiment of the present invention, the image calibration system 100 further includes an intelligent module 190, coupled to an image output module 150, an anatomical image module 110, and a functional image module 160. The intelligent module monitors the identification results of the mucosal region 610 and mucosal location 660 in the identification and analysis module 140 and the fusion analysis module, and the fusion analysis module 180, after overlaying (registration) functional and anatomical images, takes the calibration result of the union set of the mucosal region 610 and mucosal location 660. When the intelligent module 190 determines the mucosal region 610, mucosal location 660... When the calibration results of the joint set of the membrane position 660 and the mucosal region 610 and the mucosal position 660 meet the detection standards required for medical testing, the intelligent module 190 controls the image output module 150 to output the above results. However, when the intelligent module 190 determines that the accuracy of the calibration results of the mucosal region 610, the mucosal position 660, and the joint set of the mucosal region 610 and the mucosal position 660 does not meet the detection standards required for medical testing, the intelligent module 190 adjusts the operating mode or related parameters of the image calibration system 100, such as the number of voxel range Δ, until the medical testing meets the preset detection standards. The algorithms required during the operation of the intelligent module 190 can be, but are not limited to, supervised learning networks, unsupervised learning networks, hybrid learning networks, associative learning networks, optimization application networks, or any combination thereof. According to one embodiment of the present invention, the image calibration system 100 may be software, hardware, or a combination of both. If it is software, when its workflow or system components are installed on a computer system, the computer system immediately becomes a tool for executing the present invention. If it is hardware, the image calibration system 100 includes a typical processing chip, memory, temporary memory, display device, network communication interface, operating system, and application program, which are interconnected in a generally known manner to perform calculations, temporary storage, display, and data transmission. Each of its system components may also be a combination of hardware and software. Based on the above, which belong to a generally known architecture, further details are omitted here. Please refer to Figures 3-7. To achieve the aforementioned purpose, this invention provides an image calibration method 300, comprising various program steps in a first subroutine 310 and a second subroutine 320. The specific program steps include: in step S1, preparing an anatomical image; in step S2, identifying the human body contour using a contour module 120 based on the anatomical image; in step S3, using a pixel value module 130, setting a first pixel region 620 and a second pixel region 630 in the human body contour according to the ranges of the first pixel value and the second pixel value, respectively; in step S4, the identification and analysis module 140 identifies the first pixel region... The first expansion region 640 and the second expansion region 650 are respectively protruded outward by at least one voxel range Δ to form the second expansion region 640 and the second expansion region 650. In step S5, the area where the first expansion region 640 and the second expansion region 650 intersect is subtracted from the first expansion region 640, and the area remaining after subtraction is defined as a preprocessing region. In addition, a range of mucosal pixel values ​​is set to process and filter the preprocessing region, thereby identifying the mucosal region 610 in the anatomical image, wherein the voxel range Δ is between 0.2mm and 10mm. In step S8, the identification result of the mucosal region 610 is output. Similarly, it should be noted that the voxel range Δ in the above embodiments, which is between 0.2mm and 10mm, is only an example. If it is to be applied to the oral mucosa, the voxel range Δ can be between 0.2mm and 4mm. The above description is only to illustrate the method of labeling the oral mucosa in this invention. If it is necessary to label the mucosa of other tissues, such as the intestines, stomach, uterus, foreskin, respiratory tract, vagina, conjunctiva, etc., an appropriate range can be defined by selecting the actual anatomical structure. In a broad sense, the voxel range Δ selected by the overall method can be selected between 0.2mm and 10mm to adapt to different embodiments. Those skilled in the art can modify it according to the common knowledge in the biomedical field after reading the description of this invention, which is specifically stated here. According to one embodiment of the present invention, the image calibration method 300 further includes step S7, in which the identification result of the mucosal region 610 is processed by a smoothing unit 140A through a smoothing algorithm to smooth the edges and remove noise, so as to eliminate the jagged edges of the mucosal region 610 on the image. The smoothing algorithm used by the smoothing unit 140A can be selected from, but is not limited to, Gaussian blur, bilateral filter, mean filter, median filter, guided filter, non-local means (NLM), anisotropic diffusion (perona-malik filter), curvature-driven diffusion (CDD), and other smoothing algorithms. The selection principle of the smoothing algorithm can be based on computing resources, computing time, whether real-time calculation is required, the amount of image details to be retained, and the quality of noise processing. According to one embodiment of the present invention, the image calibration method 300 further includes, in step S9, an intelligent module adjusting the processing method of the anatomical image based on the processing result of the anatomical image, namely the identification result of the mucosal region 610, such as adjusting the number of the voxel range Δ. The algorithm required by the intelligent module 190 during operation can be, but is not limited to, supervised learning network, unsupervised learning network, hybrid learning network, associative learning network, optimization learning network, etc., or any combination thereof. According to one embodiment of the present invention, the image calibration method 300 further includes: in step S10, preparing a functional image; in step S11, the tissue analysis module 170 identifies the mucosal location 660 by means of the visible tumor volume (GTV) and standard uptake value (SUV) in the functional image, wherein the identification method is that the standard uptake value is approximately equal to the value of the tumor tissue, and the area of ​​the visible tumor volume is subtracted from the area of ​​the functional image to obtain the mucosal location 660; in step S12, the functional image and the anatomical image are superimposed, and the mucosal location 660 is compared with the mucosal region 610; and in step S13, the comparison result is output. According to one embodiment of the present invention, the image calibration method 300 further includes: in step S14, a smart module 190 monitors the identification and comparison results of the mucosal region 610 and the mucosal position 660 using artificial intelligence; when the smart module 190 determines that the calibration results of the mucosal region 610, the mucosal position 660, and the union of the mucosal region 610 and the mucosal position 660 meet the detection standards required for medical testing, the smart module 190 controls the image output module 150 to output the above results; however, when the smart module 190 determines that the accuracy of the calibration results of the mucosal region 610, the mucosal position 660, and the union of the mucosal region 610 and the mucosal position 660 does not meet the detection standards required for medical testing, the smart module 190 adjusts the operating mode or related parameters of the image calibration system 100, such as the number of voxel range Δ, until the medical testing meets the preset detection standards. Similarly, in the embodiments of the present invention, the algorithms required during the operation of the smart module 190 can be, but are not limited to, supervised learning networks, unsupervised learning networks, hybrid learning networks, associative learning networks, optimization learning networks, etc., or any combination thereof. In summary, the image calibration system and method proposed in this invention aim to improve the accuracy and automation of medical image analysis. By utilizing anatomical images (such as CT images) to calibrate mucosal or soft tissues in the human body, and by proposing specific parameter standards, it addresses the problem of poor resolution of mucosal tissues by anatomical images. If necessary, functional images (such as PET images) can also be combined, leveraging the superior resolution of mucosal tissues by PET to assist in the interpretation and cross-validation of CT image results, further enhancing the accuracy required for treatment planning and analysis. Therefore, this invention, in addition to its high industrial applicability, also possesses a novel system and method architecture and improved performance, hence this application. It should be noted that those skilled in the art are free to make various modifications after reading this invention, but all such modifications will not depart from the scope of protection sought in the appended claims. 100: Image calibration system; 110: Anatomical image module; 120: Contour module; 130: Pixel value module; 140: Identification and analysis module; 140A: Smoothing unit; 150: Image output module; 160: Functional image module; 170: Tissue analysis module; 180: Fusion analysis module; 190: Smart module; 300: Image calibration method; 310: First subroutine; 320: Second subroutine; S1-S14: Program steps; 610: Mucosal region; 620: First pixel region; 630: Second pixel region; 640: First dilated region; 650: Second dilated region; 660: Mucosal location; ROI: Region of interest; Δ: Voxel range. The following detailed description of the invention and the illustrated embodiments are intended to enable a fuller understanding of the invention; however, it should be understood that this is limited to providing a reference for understanding the application of the invention and is not intended to limit the invention to a particular embodiment. Figure 1 illustrates the system architecture of the image calibration system. Figure 2A illustrates the detailed system architecture of the identification and analysis module. Figure 2B further illustrates one system implementation architecture of the intelligent module combined with the image calibration system. Figure 3 illustrates the methodological architecture of the image calibration method. Figure 4 illustrates the detailed sub-architecture of the image calibration method. Figure 5 illustrates the detailed sub-architecture of the image calibration method. Figure 6 illustrates the calibration results of the mucosal region in one embodiment of this invention using anatomical images (illustrated as CT images). Figure 7 illustrates the image calibration process in one embodiment of the present invention. Figure 8 illustrates the sources of anatomical images (illustrated as CT images) and functional images (illustrated as PET images) during image fusion in one embodiment of the present invention, where it is clearly visible that the resolution of the functional images is lower than that of the anatomical images. 100: Image Calibration System 110: Anatomical Imaging Module 120: Contour Module 130: Pixel Value Module 140: Identification and Analysis Module 150: Image Output Module 160: Functional Imaging Module 170: Organization Analysis Module 180: Fusion Analysis Module

Claims

1. An image calibration system includes: an anatomical image module for reading anatomical images; a contour module for identifying human body contours based on the anatomical images; a pixel value module for setting a first pixel value and a second pixel value to filter out a first pixel region and a second pixel region in the human body contour; and an identification and analysis module for protruding the first pixel region and the second pixel region outward by a pixel range to form a first expanded region and a second expanded region, and subtracting the intersection area of ​​the first expanded region and the second expanded region from the first expanded region to form a preprocessed region; wherein... The pixel value setting module sets a mucosal pixel value, enabling the identification and analysis module to identify the mucosal region in the preprocessed area, and the voxel range is between 0.2mm and 10mm. The image calibration system as described in claim 1, wherein when applied to oral mucosa, the voxel range is set by the identification and analysis module to be between 0.2 mm and 4 mm. The image calibration system as described in claim 1, wherein the identification and analysis module further includes a smoothing unit that performs smoothing and noise reduction processing through a smoothing algorithm. The image calibration system as described in claim 3, wherein the smoothing algorithm is selected from Gaussian blur, bilateral filtering, average filtering, median filtering, guided filtering, nonlocal average filtering, anisotropic diffusion, or curvature-driven diffusion. The image calibration system as described in claim 1 further includes: a functional imaging module for reading functional images; a tissue analysis module for identifying and calibrating mucosal locations based on the functional images; and a fusion analysis module for overlaying functional images and anatomical images, and taking the union of mucosal regions and mucosal locations to form a calibration result. The image calibration system described in claim 5, wherein the tissue analysis module identifies and calibrates the mucosal location in functional images using two parameters: image-visible tumor volume (GTV) and standard uptake value (SUV). The image calibration system described in claim 6, wherein the tissue analysis module outputs the mucosal location by approximating the value of the tumor tissue to a standard uptake value and subtracting the area of ​​visible tumor volume from the area on the functional image. The image calibration system as described in claim 5 further includes an image output module coupled to the identification analysis module and the fusion analysis module, which outputs the identification results of the mucosal region and the mucosal position to the proximal or distal end. The image calibration system as described in claim 8 further includes an intelligent module coupled to the image output module, the anatomical image module, and the functional image module, which monitors the identification results of mucosal regions and mucosal locations, and the calibration results of the union set of mucosal regions and mucosal locations, and thereby adjusts the operating mode of the image calibration system. As described in claim 9, the algorithm of the intelligent module is selected from supervised learning networks, unsupervised learning networks, hybrid learning networks, associative learning networks, optimization learning networks, or any combination thereof. An image calibration method includes: preparing an anatomical image; identifying a human body contour using a contour module based on the anatomical image; defining a first pixel region and a second pixel region in the human body contour using a pixel value module based on a first pixel value and a second pixel value; protruding the first pixel region and the second pixel region outward by a voxel range to form a first expanded region and a second expanded region using an identification and analysis module; subtracting the intersection area of ​​the first expanded region and the second expanded region from the first expanded region to define a preprocessing region; and identifying a mucosal region in the preprocessing region using the pixel value module by setting a mucosal pixel value, wherein the voxel range is between 0.2 mm and 10 mm. The image calibration method as described in claim 11, wherein when applied to oral mucosa, the voxel range is set by the identification and analysis module to be between 0.2 mm and 4 mm. The image calibration method described in claim 11 further includes smoothing and noise removal of the edges of the mucosal region by passing it through a smoothing unit. The image calibration method as described in claim 13, wherein the smoothing algorithm of the smoothing unit is selected from Gaussian blur, bilateral filtering, average filtering, median filtering, guided filtering, nonlocal average filtering, anisotropic diffusion, or curvature-driven diffusion. The image calibration method described in claim 11 further includes adjusting the voxel range by means of a smart module based on the identification results of the mucosal region. The image calibration method as described in claim 11 further includes: preparing a functional image; calibrating the mucosal location based on the functional image using a tissue analysis module; and overlaying the functional image and the anatomical image using a fusion analysis module, and taking the union of the mucosal region and the mucosal location to form a calibration result. The image calibration method described in claim 16, wherein the calibration basis for the mucosal location is to use the visible tumor volume (GTV) and standard uptake value (SUV) to set the standard uptake value to a value approximating that of tumor tissue through the tissue analysis module, and to obtain the mucosal location by subtracting the area of ​​the visible tumor volume from the area of ​​the functional image marker. The image calibration method described in claim 17 further includes superimposing the functional image and the anatomical image, and comparing them according to the mucosal location and the mucosal region. The image calibration method described in claim 18 further includes using a smart module to monitor the identification results of the mucosal region and the mucosal location, and the calibration results of the union set of the mucosal region and the mucosal location to adjust the operation mode of the image calibration system method. As described in claim 19, the algorithm of the intelligent module is selected from supervised learning networks, unsupervised learning networks, hybrid learning networks, associative learning networks, optimization learning networks, or any combination thereof.

Images