Three-dimensional medical imaging system for kidney surgical planning and method thereof

Abstract

A three-dimensional medical imaging system for planning kidney surgery and a method for generating the medical images are disclosed. The three-dimensional medical imaging system includes an image processing unit that receives a plurality of contrast CT images of a patient, a 3D modeling unit that generates a three-dimensional model of a kidney and surrounding organs based on the CT images, a user interface unit that independently controls the visibility of each organ element in the three-dimensional model and displays a resection path of the kidney according to the kidney surgery plan, and a function prediction unit that predicts the estimated glomerular filtration rate (eGFR) after surgery.

Description

Technology Field

[0001] The present invention relates to medical image processing technology, and more specifically, to a 3D medical imaging system and a method for generating medical images that generate a 3D model based on a plurality of contrast CT images and AI technology for planning kidney surgery, and integrally provide organ-specific visualization control, surgical path simulation, and eGFR prediction functions. Background Technology

[0002] Renal tumor resection is a complex surgery that must simultaneously achieve two goals: preserving the patient's renal function and removing the tumor. Particularly in the case of partial nephrectomy, precise preoperative planning directly affects the success rate of the surgery and the patient's postoperative renal function.

[0003] Conventional kidney surgery planning relies primarily on 2D CT images or simple 3D reconstructions, which has limitations in accurately identifying complex anatomical structures and the positional relationship of tumors. Additionally, since only some of the four phases (Pre-contrast (N Phase), A Phase, P Phase, D Phase) are often captured, it is difficult to generate a complete 3D model.

[0004] A more significant issue is the lack of quantitative tools to predict changes in renal function after resection during the surgical planning phase. Currently, clinical practice relies primarily on the experience of medical professionals to determine the extent of resection, which limits the ability to establish optimized surgical plans for each patient. Prior art literature

[0005] Republic of Korea Published Patent Application No. 10-2024-0167458 (Published November 27, 2024) The problem to be solved

[0006] The objective of the present invention to solve the aforementioned problems is to provide a 3D medical imaging system and a method for generating medical images that generate an accurate 3D elongation model by integrally utilizing multiple phases of contrast CT images and automatically supplementing any missing phases using AI technology.

[0007] Another objective of the present invention is to provide a three-dimensional medical imaging system and a method for generating medical images that provide intuitive visualization control functions specialized for kidney surgery, thereby enabling medical staff to establish an optimal surgical plan through independent control of each organ element, such as fat, kidney, and tumor.

[0008] Another objective of the present invention is to provide a three-dimensional medical imaging system and a method for generating medical images that enable the establishment of a precise surgical plan by providing a real-time parameter adjustment function and a resection plane deformation function.

[0009] An additional objective of the present invention is to provide a three-dimensional medical imaging system and a method for generating medical images that can pre-evaluate the rate of preservation of renal function after surgery by predicting changes in eGFR in conjunction with a surgical plan. means of solving the problem

[0010] To achieve the above-mentioned purpose, a three-dimensional medical imaging system for planning kidney surgery according to one embodiment of the present invention comprises: an image processing unit that receives a plurality of contrast CT images of a patient; a 3D modeling unit that generates a three-dimensional model of a kidney and surrounding organs based on the CT images; a user interface unit that independently controls the visibility of each organ element in the three-dimensional model to set the kidney surgery plan and displays the resection path of the kidney according to the kidney surgery plan; and a function prediction unit that predicts the estimated glomerular filtration rate (eGFR) before and after the surgery.

[0011] The above plurality of contrast CT images include phase images of Pre-contrast Phase, A Phase, P Phase, and D Phase, and the image processing unit generates the missing phase image based on the remaining phase images using an artificial intelligence model when at least one phase image among the contrast CT images is missing.

[0012] The user interface unit includes a visualization unit that independently displays or hides the organ elements, such as fat, kidney, and tumor, in the three-dimensional model, and a surgical planning unit that displays the resection path of the kidney according to the kidney surgery plan in the three-dimensional model.

[0013] The visualization unit provides user interfaces corresponding to the organ elements and, depending on whether the user interface is selected, adjusts or removes the transparency of the selected organ in real time.

[0014] The surgical planning unit above provides a user interface for generating resection guidelines, setting parameters, and modifying the shape of the resection guidelines, and displays the resection path of the kidney based on the values ​​set from the user interface.

[0015] The above-mentioned function prediction unit uses a pre-trained AI model to predict the post-eGFR based on the planned resection range and the pre-eGFR, and calculates the renal function preservation rate as a percentage.

[0016] To achieve the above-mentioned purpose, a method for generating three-dimensional medical images for kidney surgery planning according to one embodiment of the present invention comprises the steps of: receiving a plurality of contrast CT images of a patient (S100); generating a three-dimensional model of a kidney and surrounding organs based on the CT images (S200); independently controlling the visibility of each organ element in the three-dimensional model to set the kidney surgery plan and displaying the resection path of the kidney according to the kidney surgery plan (S300); and predicting the estimated glomerular filtration rate (eGFR) after surgery (S400).

[0017] The above step (S100) includes a step (S110) of receiving a plurality of contrast CT images of a patient classified into four phases (Pre-contrast Phase, A Phase, P Phase, D Phase) and a step (S120) of checking if there is a missing phase among the four phases and, if there is a missing phase, generating an image of the missing phase based on the CT images of the remaining phases using an artificial intelligence model.

[0018] The above step (S300) includes the step (S310) of independently displaying or hiding the organ elements, including fat, kidney, and tumor, in the three-dimensional model according to user input, and the step (S320) of displaying the resection path of the kidney in the three-dimensional model according to user input.

[0019] The above step (S310) includes providing a user interface for controlling the transparency of the organ element and changing the visibility of the organ element in real time according to the operation of the user interface (S311).

[0020] The above step (S320) includes providing a user interface for setting the resection angle, adjusting the resection interval, and modifying the shape of the resection guideline, and a step (S321) of changing the kidney surgery plan according to the operation of the user interface and updating it on the three-dimensional model.

[0021] The above step (S400) includes a step (S410) of predicting changes in glomerular filtration rate according to a set surgical plan and displaying kidney function before and after surgery as a percentage. Effects of the invention

[0022] According to the 3D medical imaging system and the method for generating medical images according to the present invention, a high-resolution 3D renal model is generated by integrating multiple phases of contrast CT images, and a missing phase is automatically supplemented by AI, thereby enabling the establishment of a complete surgical plan even with limited image data.

[0023] According to the 3D medical imaging system and the method for generating medical images of the present invention, medical staff can clearly understand complex anatomical structures and determine the optimal surgical approach through intuitive visualization control functions specialized for kidney surgery.

[0024] According to the 3D medical imaging system and the medical image generation method of the present invention, it is possible to establish a customized surgical plan for each patient through a real-time parameter adjustment function and a resection plane deformation function.

[0025] According to the 3D medical imaging system and the method for generating medical images of the present invention, patient safety can be significantly improved by quantitatively evaluating changes in kidney function after surgery in advance through an AI-based eGFR prediction function. Brief explanation of the drawing

[0026] FIG. 1 is a block diagram showing each configuration of a three-dimensional medical imaging system according to the present invention. FIG. 2 is a diagram showing the process of generating a three-dimensional model of a 3D modeling unit according to the present invention. FIG. 3 is a diagram showing the overall screen configuration of a user interface provided by a user interface unit according to the present invention. Figure 4 is a diagram showing a detailed image verification function through the SEG button in the CT image list area according to the present invention. FIG. 5 is a diagram showing the rotation, enlargement, and movement operations of an extended model in a 3D viewport area according to the present invention. Figure 6 is a diagram showing the transparency control function for each organ through the visualization control panel according to the present invention. FIG. 7 is a diagram showing the surgical plan setting and resection depth indication according to the present invention. FIG. 8 is a diagram showing a real-time resection depth update according to the resection plane change according to the present invention. FIG. 9 is a diagram showing the function of changing the cutting angle among the parameter setting functions according to the present invention. FIG. 10 is a diagram showing the setting of a safety margin among the parameter setting functions according to the present invention. FIG. 11 is a drawing showing the initial state of the deformation guide function according to the present invention. FIG. 12 is a drawing showing the deformation of the cut surface through the deformation guide according to the present invention. FIG. 13 is a diagram showing the result screen after executing the ablation simulation according to the present invention. FIG. 14 is a diagram showing the change and prediction of the glomerular filtration rate provided by the function prediction unit according to the present invention. FIG. 15 is a flowchart of a method for generating a three-dimensional medical image according to the present invention. FIG. 16 is a detailed flowchart of a method for generating a three-dimensional medical image according to the present invention. Specific details for implementing the invention

[0027] Some embodiments of the present invention will be described in detail below with reference to the exemplary drawings. It should be noted that in assigning reference numerals to the components of each drawing, the same components are given the same reference numeral whenever possible, even if they are shown in different drawings.

[0028] Furthermore, in describing the embodiments of the present invention, if it is determined that a detailed description of related known configurations or functions would hinder understanding of the embodiments of the present invention, such detailed description is omitted.

[0029] In addition, terms such as first, second, A, B, (a), (b), etc., may be used when describing the components of the embodiments of the present invention. These terms are used merely to distinguish the components from other components, and the essence, order, or sequence of the components is not limited by the terms used.

[0030] In this specification, the singular form includes the plural form unless specifically stated otherwise in the text. As used in this specification, “comprising” and / or “comprising” does not exclude the presence or addition of one or more other components in addition to the mentioned components.

[0031] Hereinafter, the present invention will be described in more detail with reference to the attached drawings.

[0033] FIG. 1 is a block diagram showing each configuration of a three-dimensional medical imaging system according to the present invention.

[0034] Referring to FIG. 1, the three-dimensional medical imaging system (100) according to the present invention includes an image processing unit (110) that receives a plurality of contrast CT images of a patient, a 3D modeling unit (120) that generates a three-dimensional model of a kidney and surrounding organs based on the input CT images, a user interface unit (130) that independently controls the visibility of each organ element in the generated three-dimensional model to set a kidney surgery plan and displays the resection path of the kidney on a display device according to the kidney surgery plan, and a function prediction unit (140) that predicts the glomerular filtration rate (eGFR) before and after surgery.

[0036] The image processing unit (110) is a component that receives and preprocesses multiple phase contrast CT images of a patient, and systematically classifies and manages images for four phases: Pre-contrast Phase (N Phase), A Phase, P Phase, and D Phase. The image processing unit (110) performs preprocessing operations such as noise removal, contrast enhancement, and resolution normalization on the CT images of each input phase to convert them into a form optimized for 3D modeling.

[0037] The image processing unit (110) detects cases where some of the four phases are missing, and if the missing is confirmed, automatically activates an AI-based image synthesis process.

[0038] According to an embodiment, the image processing unit (110) analyzes the DICOM header information of the input CT images to determine the time of capture and can detect cases where some of the four phases are missing. For example, the image processing unit (110) can identify the absence of images corresponding to the elapsed time after contrast agent administration: 25-30 seconds (Phase A), 65-70 seconds (Phase P), and after 180 seconds (Phase D). If a missing image is identified, the AI ​​synthesis module uses a U-Net-based encoder-decoder structure and can generate an image of the missing phase by learning the pixel intensity distribution, contrast pattern, and anatomical structure information of the existing phases. Through this, the contrast state at the time of the missing image can be predicted, and an image with image characteristics similar to the missing phase can be generated. However, it is not limited thereto.

[0040] FIG. 2 is a diagram showing the process of generating a three-dimensional model of a 3D modeling unit according to the present invention.

[0041] According to one embodiment of the present invention, the 3D modeling unit (120) first generates a segmentation mask for each anatomical structure through an AI model specialized for each phase based on the input 4-phase abdominal CT images (N, A, P, D) of the patient.

[0042] Specifically, the Vessel Model receives Phase A and Phase P CTs as input and generates artery, vein, and kidney masks. The Ureter Model receives Phase D CTs as input and generates ureter and kidney masks. The Tumor Model receives CTs from all phases as input and generates kidney and tumor masks. Each of these phase-specific models uses a deep learning-based segmentation algorithm to identify the interface between tumors and normal kidney tissues with sub-millimeter accuracy.

[0043] Instead of a separate AI model, the fat mask performs distance-based dilation around the kidney and tumor and automatically derives anatomically valid fat regions through the determination of the nearest distance to other organs. Finally, segmentation masks are generated for a total of six elements: tumor, kidney, artery, vein, ureter, and fat.

[0044] The generated individual masks select the mask elements to be finally used in masks of different phases based on set priorities, and generate a single integrated mask through a resizing process to compensate for resolution differences.

[0045] Based on the integrated mask, the 3D modeling unit (120) generates initial volume data by stacking 2D CT slices of each Phase and converts the medical image file (.nii) into a 3D model (.glb) by applying the Marching Cubes algorithm.

[0046] Next, the 3D modeling unit (120) smooths the surface using Poisson Reconstruction and Taubin Smoothing algorithms to remove stair-step artifacts and create natural curved surfaces. Finally, unique colors and textures are mapped to each organ to complete a 3D model with realistic visual quality.

[0048] The user interface section (130) includes a visualization section (131) and a surgical planning section (133), and provides an intuitive interface that allows medical staff to effectively manipulate a three-dimensional model and establish a surgical plan.

[0049] The visualization unit (131) provides the ability to independently control each organ element of the generated 3D model and individually controls a total of 8 anatomical structures, including artery (ARTERY), left fat (FAT-L), right fat (FAT-R), left kidney (KIDNEY-L), right kidney (KIDNEY-R), tumor (TUMOR), ureter (URETER), and vein (VEIN).

[0050] The surgical planning unit (133) provides tools that can simulate the actual surgical process, supports a function to generate resection guidelines, provides a user interface for real-time parameter setting, changing the resection plane and shape modification of the resection guidelines, and provides a function to display the resection path of the kidney based on the values ​​set from the user interface, thereby supporting the planning of an optimal surgical path.

[0052] FIGS. 3 to 12 are drawings illustrating a screen (200) displayed on a display device by a user interface unit (130) according to the present invention. Hereinafter, the user interface (200) provided by the user interface unit (130) will be described with reference to FIGS. 3 to 12.

[0053] FIG. 3 is a diagram showing the overall screen configuration of a user interface provided by the user interface unit (130) according to the present invention.

[0054] Referring to FIG. 3, the user interface (200) is composed of several functional areas. In the CT image list area (210), four Phase CT images of a patient, received and processed by the image processing unit (110), are displayed in an orderly fashion, and detailed information can be checked in each image. In the toolbar area (220), buttons are arranged to initialize the camera viewpoint and adjust view options.

[0055] The central 3D viewport area (230) is the main workspace where a 3D extended model created by the 3D modeling unit (120) is rendered in real time, and rotation, zooming in, zooming out, and movement operations using a mouse are possible. The overall brightness of the 3D model can be adjusted in real time through the brightness control bar (240) at the bottom of the screen.

[0056] On the right panel, the visualization control area (250), surgical planning area (260), parameter setting area (270), and eGFR prediction area (280) are systematically arranged to provide their respective specialized functions.

[0058] FIG. 4 is a diagram showing a detailed image verification function through the SEG button (211) in the CT image list area (210) according to the present invention.

[0059] Referring to FIG. 4, by clicking the SEG button (211) for each Phase CT image (210a to 210d) in the CT image list area (210), one can view the details of each Phase CT image. Through this function, medical staff can check the quality of the original CT data and, if necessary, sequentially review other slice images by scrolling. Each image clearly displays a contrast pattern according to the characteristics of the corresponding Phase, allowing for accurate identification of the angiographic status or the degree of tumor contrast enhancement.

[0061] FIG. 5 is a diagram showing the rotation, enlargement, and movement operations of an extended model in a 3D viewport area according to the present invention.

[0062] Referring to FIG. 5, in the 3D viewport area (230), the user can freely rotate the kidney model by using a mouse. Rotation via left-click drag, zooming in / out via wheel scroll, and movement via right-click drag are supported, and all operations are reflected in real time. Through this free viewpoint manipulation, medical staff can observe the complex three-dimensional structure of the kidney from various angles and identify the spatial relationship between the tumor and surrounding tissues.

[0064] FIG. 6 is a diagram showing the transparency control function for each organ through the visualization control panel (250) according to the present invention.

[0065] Referring to FIG. 6, clicking each button (artery, left fat (FAT-L), right fat (FAT-R), left kidney (KIDNEY-L), right kidney (KIDNEY-R), tumor (TUMOR), ureter (URETER), vein (VEIN)) of the visualization control panel (250) can adjust the transparency of the corresponding organ elements in the 3D model of the 3D viewport area (230). In one embodiment, each time a button for each organ is clicked, the transparency can cycle between a fully visible state, a translucent state, and a fully hidden state.

[0066] For example, when observing kidneys or tumors in detail, adipose tissue can be made transparent to secure a clear view, and when identifying relationships with vascular structures, only arteries and veins can be selectively displayed. Through these selective visualization functions, medical professionals can focus on areas of interest within complex anatomical structures and clearly identify the relationships between important structures when planning surgery.

[0068] FIG. 7 is a diagram showing the surgical plan setting and resection depth indication according to the present invention.

[0069] Referring to FIG. 7, when the Cutting Guide is selected in the surgical planning area (260), the resection guideline is displayed on the 3D model in the 3D viewport area (230). The resection guideline visualizes the path for removing a tumor located mainly inside the kidney by approaching it from the outside, and clearly indicates the direction of the surgery.

[0070] Along with the resection guidelines, the resection depth (230a) is displayed numerically, allowing medical staff to quantitatively determine how much healthy kidney tissue can be preserved during the resection process. The resection depth is calculated in real time based on the location and size of the tumor and established safety margins, and is provided as an accurate value in millimeters.

[0072] FIG. 8 is a diagram showing a real-time resection depth update according to the resection plane change according to the present invention.

[0073] Referring to FIG. 8, the position or angle of the resection plane can be directly adjusted by clicking the mouse on the 3D model in the 3D viewport area (230), and the resection depth (230a) is updated in real time according to the changed settings. Through this intuitive operation method, medical staff can seek a realistic approach that considers not only theoretical plans but also actual surgical situations.

[0075] FIG. 9 is a diagram showing the function of changing the cutting angle among the parameter setting functions according to the present invention, and FIG. 10 is a diagram showing the setting of the safety margin among the parameter setting functions according to the present invention.

[0076] Referring to FIGS. 9 and 10, resection-related parameters can be precisely set through the adjustment bar provided in the parameter setting area (270). According to an embodiment, the upper adjustment bar can adjust the resection angle in increments of 1 degree in the range of -90 degrees to 90 degrees, and as the angle increases, the resection surface widens to include more surrounding tissue.

[0077] The lower adjustment bar provides a function to set a safety margin from the tumor boundary to the resection margin. The safety margin can be adjusted in increments of 0.1 according to the embodiment, and the unit may be cm, but is not limited thereto.

[0078] While a larger safety margin increases the likelihood of complete removal of tumor cells, it also increases the loss of healthy kidney tissue; therefore, optimization tailored to each patient's specific situation is necessary.

[0079] As shown in Fig. 10, when the parameters are adjusted, the resection guidelines are changed immediately, and the resection depth (230a) according to the new settings is updated. Through this real-time feedback, medical staff can quickly compare various options and select the surgical method most suitable for the patient.

[0081] FIG. 11 is a drawing showing the initial state of the deformation guide function according to the present invention, and FIG. 12 is a drawing showing the deformation of the cut surface through the deformation guide according to the present invention.

[0082] Referring to FIG. 11, when the Deformable Guide is selected in the surgical planning area (260), the system switches to a mode in which the user can directly modify the shape of the resection guideline through mouse operation. The Deformable Guide can be activated after the Cutting Guide is selected, and preferably, it can be utilized after parameter settings. The Deformable Guide mode supports medical staff in finely adjusting the resection plane by taking into account the irregular shape of the tumor or surrounding vascular structures.

[0083] FIG. 12 illustrates a user selecting and modifying a specific area of ​​the resection guideline by clicking and dragging a mouse. The modified resection plane is reflected in the 3D model in real time, and the resection depth (230a) and resection volume are also immediately recalculated and displayed. Through this deformable guide function, medical staff can establish a surgical plan optimized for the patient's individual anatomical characteristics, rather than being limited to standardized resection shapes.

[0084] The modification guide is designed to adjust the resection margin by comprehensively considering the irregular boundaries of the tumor, the location of adjacent blood vessels, and the thickness of the renal parenchyma, thereby enabling the establishment of a surgical plan that achieves complete tumor removal while remaining minimally invasive.

[0086] FIG. 13 is a diagram showing the result screen after executing the ablation simulation according to the present invention.

[0087] Referring to FIG. 13, after completing all parameter settings in the surgical planning area (260) and the parameter setting area (270) and optimizing the shape of the resection surface through the Deformable Guide if necessary, if the RESECTION button in the surgical planning area (260) is selected, the actual resection result is simulated according to the set resection guideline and displayed on the 3D model in the 3D viewport area (230). Through the displayed simulation result, the shape of the kidney after surgery can be checked in advance.

[0088] In the resection simulation, the volume of the tumor to be removed, the volume of normal kidney tissue removed along with it, and the volume of preserved kidney tissue are calculated and provided as numerical values. Additionally, the preservation of major blood vessels or ureters can be visually verified.

[0089] If the simulation results are unsatisfactory, you can return to the initial state via the RESTORE button to formulate a new plan, and through this iterative planning process, you can derive the optimal surgical strategy.

[0091] FIG. 14 is a diagram showing the change and prediction of the glomerular filtration rate provided by the function prediction unit according to the present invention.

[0092] Referring to FIG. 14, the function prediction unit (140) according to the present invention performs an important function of predicting changes in kidney function after surgery based on a set resection plan. In the eGFR prediction area (280), the medical staff first inputs the patient's pre-operative eGFR value into the input window (281).

[0093] When the AI ​​prediction button (283) is clicked after inputting the pre-operative eGFR, the machine learning model embedded in the function prediction unit (140) is activated. This model predicts post-operative kidney function by comprehensively considering the location, volume, and blood vessel distribution of the kidney tissue to be resected.

[0094] The eGFR prediction process of the function prediction unit (140) according to one embodiment of the present invention is described in more detail as follows.

[0095] First, a virtual resection simulation is performed based on the tumor resection margin set by the medical team in the 3D viewer. Through this simulation, the volume of the residual kidney after resection is accurately calculated, which serves as key input data for eGFR prediction.

[0096] The AI ​​prediction model embedded in the function prediction unit (140) can predict postoperative eGFR based on three feature values: Pre-eGFR (preoperative eGFR value entered by medical staff), preoperative kidney volume (calculated from simulation results), and postoperative / preoperative kidney volume ratio (calculated from simulation results).

[0097] These three feature values ​​undergo a normalization process and are input into a regression neural network model. The regression neural network is pre-trained with data from multiple kidney surgery cases and predicts postoperative kidney function by considering complex factors such as the location, volume, and vascular distribution of the kidney tissue to be resected.

[0098] The output result of the AI ​​model, i.e., the prediction result (285), is provided in two forms: Post-eGFR and Predicted eGFR. Post-eGFR is the absolute eGFR value (in ml / min / 1.73 m² units) expected after resection, and Predicted eGFR is a percentage representing the renal function preservation rate, calculated as “(Post-eGFR / Pre-eGFR) × 100”.

[0099] For example, if a patient's preoperative eGFR is 96 ml / min / 1.73㎡, the postoperative eGFR is predicted to be 92.97 ml / min / 1.73㎡ according to the established resection plan, which means a renal function preservation rate of 96.84%. Through this quantitative predictive information, medical staff can objectively evaluate the risks and benefits of surgery and explain them to the patient based on specific figures during consultations.

[0101] FIG. 15 is a flowchart of a method for generating a three-dimensional medical image according to the present invention.

[0102] Referring to FIG. 15, the method for generating a three-dimensional medical image according to the present invention is a method for generating a three-dimensional medical image for planning kidney surgery, comprising the steps of receiving a plurality of contrast CT images of a patient (S100), generating a three-dimensional model of a kidney and surrounding organs based on the CT images (S200), independently controlling the visibility of each organ element in the three-dimensional model to set the kidney surgery plan and displaying the resection path of the kidney according to the kidney surgery plan (S300), and predicting the estimated glomerular filtration rate (eGFR) after surgery (S400).

[0103] In the above step (S100), contrast CT scan data obtained from the patient is classified according to chronological order, and the completeness of the images for each phase is examined to determine the direction of subsequent processing. If all four phases are completely present, the process proceeds directly to the 3D modeling stage, and if some phases are missing, the missing data is supplemented through an AI-based synthesis process.

[0104] In the above step (S200), preprocessed CT images are used as input to generate segmentation masks for each anatomical structure using a specialized AI model for each phase. The generated individual masks are integrated according to priority to create a single mask, and a volume rendering technique is applied based on this. Based on the integrated mask and volume data, a 3D mesh model is generated using a marching cube algorithm, and finally, a 3D model suitable for visualization is completed.

[0105] In the above step (S300), an interactive surgical plan is established through a user interface based on the generated 3D model. The medical team adjusts the transparency of each organ to intensively observe the area of ​​interest, and plans the optimal surgical path by setting the resection method and parameters.

[0106] In the above step (S400), postoperative kidney function is predicted through a machine learning model based on the confirmed surgical plan, and the results are presented as quantitative indicators to support the medical team's final decision-making.

[0108] Hereinafter, with reference to FIG. 16, a method for generating a three-dimensional medical image according to the present invention will be described in more detail.

[0109] FIG. 16 is a detailed flowchart of a method for generating a three-dimensional medical image according to the present invention.

[0110] Referring to FIG. 16, the above step (S100) includes a step (S110) of receiving a plurality of contrast CT images of a patient classified into four phases (Pre-contrast Phase, A Phase, P Phase, D Phase) and a step (S120) of checking if there is a missing phase among the four phases and, if there is a missing phase, generating an image of the missing phase based on the CT images of the remaining phases using an artificial intelligence model.

[0112] In the above step (S110), the metadata of the input CT images is analyzed to determine the time of shooting, and each image is assigned to an appropriate Phase based on the temporal relationship before and after the administration of the contrast agent. The Pre-contrast Phase is the basic image before the administration of the contrast agent, Phase A is the arterial phase image around 25-30 seconds after the administration of the contrast agent, Phase P is the portal phase image around 65-70 seconds, and Phase D is the delayed phase image after 180 seconds.

[0113] In the above step (S120), the completeness of the classified phases is checked, and if a missing phase is found, a deep learning-based image synthesis algorithm is executed. In the synthesis process, a generative adversarial network that has learned the contrast patterns of the existing phases models the expected contrast state at the time of the missing phase to generate a new image. The generated synthetic image has the same resolution and contrast as the original image and ensures quality suitable for clinical interpretation.

[0115] The above step (S300) includes the step (S310) of independently displaying or hiding the organ elements, including fat, kidneys, and tumors, in the three-dimensional model according to user input.

[0116] In the above step (S310), the visibility of each anatomical structure is controlled through organ-specific control buttons on the user interface. Independent transparency control is possible for eight major structures, including arteries, fat, kidneys, tumors, ureters, and veins, and each structure is controlled in three stages: fully displayed, semi-transparent, and fully hidden. Through this selective visualization, medical staff can focus on areas of interest in complex three-dimensional structures and clearly identify important anatomical relationships when planning surgery.

[0117] The above step (S310) includes providing a user interface for controlling the transparency of the organ element and changing the visibility of the organ element in real time according to the operation of the user interface (S311).

[0118] In the above step (S311), when the user clicks the button, the transparency of the corresponding organ is immediately changed and reflected in the 3D viewport. In addition, simultaneous control of multiple organs is supported, allowing medical staff to set optimal observation conditions through a combination of complex visualizations.

[0120] The above step (S300) further includes the step (S320) of displaying the resection path of the kidney in the three-dimensional model according to the user's input.

[0121] In the above step (S320), the resection guidelines are visualized on a 3D model according to the surgical method set by the medical team. The resection guidelines include the resection area facing the tumor and are displayed in a translucent color to clearly show the resection range without obscuring internal structures.

[0122] The above step (S320) includes providing a user interface for setting the resection angle, adjusting the resection interval, and modifying the shape of the resection guideline, and a step (S321) of changing the kidney surgery plan according to the operation of the user interface and updating it on the three-dimensional model.

[0123] In the above step (S321), resection-related parameters can be adjusted in real time through a parameter adjustment interface. The resection angle is adjusted in 1-degree increments via a slider, and the safety margin is precisely adjusted in 0.1-degree increments. Whenever a parameter is changed, the resection guideline is immediately updated, and the resection volume and expected result according to the changed settings are calculated and displayed in real time.

[0124] In addition, through the Deformable Guide interface, users can directly modify the shape of the resection guideline or the resection plane by clicking and dragging with a mouse, thereby enabling the establishment of a customized resection plan that takes into account the irregular shape of the tumor or surrounding vascular structures.

[0126] The above step (S400) includes a step (S410) of predicting changes in glomerular filtration rate according to a set surgical plan and displaying kidney function before and after surgery as a percentage.

[0127] In the above step (S410), feature values ​​calculated through a virtual resection simulation are input into a regression neural network model to predict postoperative eGFR. Specifically, three feature values—Pre-eGFR (preoperative eGFR), preoperative kidney volume, and postoperative / preoperative kidney volume ratio—are input into the model after undergoing a normalization process.

[0128] The prediction model is pre-trained with data from multiple kidney surgery cases and estimates final kidney function by considering complex factors such as the location, volume, and vascular distribution of the resected kidney tissue. The prediction results are provided simultaneously as the absolute Post-eGFR value (ml / min / 1.73㎡) and the relative Predicted eGFR value (preservation rate %), supporting medical staff in comprehensively assessing the risks and benefits of the surgery.

[0130] The present invention can be implemented in a parallel computing environment utilizing a high-performance graphics processing unit (GPU), and can support real-time rendering and AI computation through accelerated processing based on CUDA or OpenCL. The 3D rendering engine provides high-quality visualization using shader technology and maximizes the three-dimensionality of anatomical structures through multiple lighting and shadow effects.

[0131] The database can support DICOM standards and ensure compatibility with various medical imaging systems. To ensure the security of patient information and the protection of personal information, an encryption system compliant with HIPAA standards may be applied, and access control management and audit trail functions may be provided.

[0132] The user interface is implemented based on responsive web technology, enabling it to provide an optimized screen on displays of various resolutions. It supports multi-touch gestures to enable intuitive operation in touchscreen environments, and optionally supports hands-free operation through voice command recognition.

[0133] In this specification, the image processing unit (110), 3D modeling unit (120), user interface unit (130), and function prediction unit (140) may be processors that execute a series of processes stored in memory. Alternatively, they may operate as software modules driven and controlled by the processor. Furthermore, the processor may be a hardware device.

[0134] For reference, a method for generating a three-dimensional medical image according to one embodiment of the present invention may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., either individually or in combination. The program instructions recorded on the medium may be those specifically designed and configured for the present invention, or they may be those known and available to those skilled in the art of computer software. Examples of computer-readable media may include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of program instructions include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. The aforementioned hardware devices may be configured to operate as one or more software modules to perform the operation of the present invention, and vice versa.

[0135] In this specification, the term "part" includes a unit realized by hardware, a unit realized by software, and a unit realized using both. Additionally, one unit may be realized using two or more hardware, and two or more units may be realized by one hardware.

[0136] The scope of protection of the present invention is not limited to the descriptions and expressions of the embodiments explicitly described above. Furthermore, the scope of protection of the present invention cannot be limited by obvious modifications or substitutions within the technical field to which the present invention pertains. Explanation of the symbols

[0137] 100 : 3D medical imaging system 110 : Image processing unit 120: 3D Modeling Section 130: User Interface Section 131 : Visualization Section 133 : Surgical Planning Section 140 : Function Prediction Section 200 : User Interface 210: CT Image List Area 211: SEG Button 220: Toolbar area 230: 3D viewport area 230a : Resection depth 240 : Brightness adjustment bar 250: Visualization Control Area 260: Surgical Planning Area 270: Parameter setting area 280: eGFR prediction area 281 : eGFR Input Window 283 : AI Prediction Button 285 : Prediction result display area

Claims

Claim 1 A three-dimensional medical imaging system for planning kidney surgery comprises: an image processing unit that receives multiple contrast CT images of a patient; a 3D modeling unit that generates a three-dimensional model of the kidney and surrounding organs based on the CT images; and a user interface unit that independently controls the visibility of each organ element in the three-dimensional model to set the kidney surgery plan and displays the resection path of the kidney according to the kidney surgery plan. A 3D medical imaging system comprising: a function prediction unit for predicting the glomerular filtration rate (eGFR) after surgery; wherein the user interface unit simulates the resection results according to set resection guidelines and displays them on the 3D model, and calculates and provides the volume of the tumor to be removed, the volume of normal kidney tissue removed together, and the volume of the kidney tissue to be preserved; wherein the function prediction unit predicts the post-eGFR by inputting a plurality of feature values, including the pre-eGFR, pre-eGFR, and the ratio of post-eGFR to pre-eGFR, into a pre-trained regression neural network model based on the residual kidney volume calculated from the simulation, and simultaneously provides the absolute value of the post-eGFR and the kidney function preservation rate calculated as the ratio of the post-eGFR to the pre-eGFR as a percentage. Claim 2 A three-dimensional medical imaging system according to claim 1, wherein the plurality of contrast CT images include phase images of a Pre-contrast Phase, A Phase, P Phase, and D Phase, and the image processing unit generates the missing phase image based on the remaining phase images using an artificial intelligence model when at least one phase image among the contrast CT images is missing. Claim 3 A three-dimensional medical imaging system according to claim 1, wherein the user interface unit comprises: a visualization unit that independently displays or hides the organ elements, including fat, kidney, and tumor, in a three-dimensional model in a semi-transparent manner; and a surgical planning unit that displays the resection path of the kidney according to the kidney surgery plan in the three-dimensional model. Claim 4 A three-dimensional medical imaging system according to claim 3, wherein the visualization unit provides user interfaces corresponding to the organ elements, and, depending on whether the user interface is selected, adjusts the transparency of the selected organ in real time or hides the selected organ element from the screen. Claim 5 A three-dimensional medical imaging system according to claim 3, wherein the surgical planning unit provides a user interface for generating resection guidelines, setting parameters, and modifying the shape of the resection guidelines, and displays a resection path of the kidney based on values ​​set from the user interface. Claim 6 delete Claim 7 A method for generating a three-dimensional medical image for planning kidney surgery comprises: an image processing unit receiving a plurality of contrast CT images of a patient (S100); a 3D modeling unit generating a three-dimensional model of the kidney and surrounding organs based on the CT images (S200); and a user interface unit independently controlling the visibility of each organ element in the three-dimensional model to set the kidney surgery plan, and displaying the resection path of the kidney according to the kidney surgery plan (S300). The method comprises a step (S400) in which a function prediction unit predicts the glomerular filtration rate (eGFR) after surgery; wherein, in step (S300), the user interface unit simulates the resection results according to the set resection guidelines and displays them on the 3D model, and calculates and provides the volume of the tumor to be removed, the volume of normal kidney tissue to be removed together, and the volume of the kidney tissue to be preserved; and in step (S400), the function prediction unit predicts the post-eGFR by inputting a plurality of feature values, including the pre-eGFR, pre-eGFR, and the ratio of post-eGFR to pre-eGFR, into a pre-trained regression neural network model based on the residual kidney volume calculated from the simulation, and the absolute value of the post-eGFR and the ratio of the post-eGFR to the pre-eGFR A 3D medical image generation method characterized by simultaneously providing a kidney function preservation rate calculated as a ratio as a percentage. Claim 8 A method for generating three-dimensional medical images according to claim 7, wherein the step (S100) comprises: receiving a plurality of contrast CT images of a patient classified into four phases including Pre-contrast Phase, A Phase, P Phase, and D Phase (S110); and checking whether there is a missing phase among the four phases, and if there is a missing phase, generating an image of the missing phase based on the CT images of the remaining phases using an artificial intelligence model (S120). Claim 9 A method for generating a three-dimensional medical image according to claim 7, wherein the step (S300) comprises: a step (S310) of independently displaying or hiding the organ elements, including fat, kidney, and tumor, in the three-dimensional model according to user input; and a step (S320) of displaying the resection path of the kidney in the three-dimensional model according to user input. Claim 10 A method for generating a three-dimensional medical image according to claim 9, wherein the step (S310) comprises providing a user interface for a transparency control function of the organ element and changing the visibility of the organ element in real time according to the operation of the user interface (S311). Claim 11 A method for generating a three-dimensional medical image according to claim 9, wherein the step (S320) includes providing a user interface for setting the resection angle, adjusting the resection interval, and modifying the shape of the resection guideline, and a step (S321) of changing the kidney surgery plan according to the operation of the user interface and updating it on the three-dimensional model. Claim 12 delete

Images