Abdominal cavity chemotherapy perfusion dose-diffusion synergistic regulation method based on equipotential line search

Abstract

The application discloses a peritoneal cavity chemotherapy perfusion dose-diffusion synergistic regulation method based on equipotential line search, comprising the following steps: S1, collecting peritoneal cavity three-dimensional image data, constructing a peritoneal cavity structure model, and identifying target / non-target tissues; S2, constructing a diffusion velocity field based on the structure model, and generating a three-dimensional diffusion equipotential line map using a fast marching method; S3, performing equipotential line search, identifying uncovered areas, diffusion mutation areas, and generating a candidate position set; S4, generating an optimized perfusion strategy based on the candidate set and a fitness function, and using a firework optimization algorithm; S5, generating control instructions and driving the perfusion system to execute the optimized perfusion scheme; S6, collecting feedback data in real time, reconstructing the equipotential line map, and performing steps S3 to S6 in a loop if the standard is not met. Through the construction of a dynamic individualized diffusion model and an intelligent optimization control system, the application realizes accurate delivery and real-time adjustment of peritoneal cavity chemotherapy drugs, and significantly improves drug efficacy coverage and treatment safety.

Description

Technical Field

[0001] This invention relates to the field of medical artificial intelligence technology, and in particular to a method for coordinated regulation of intraperitoneal chemotherapy perfusion dose-diffusion based on equipotential line search. Background Technology

[0002] In the comprehensive treatment of peritoneal disseminated tumors, intraperitoneal chemotherapy, as a highly targeted treatment strategy with high local efficacy and low systemic toxicity, has been widely used in the control of postoperative residual lesions in solid tumors such as ovarian cancer, gastric cancer, and colorectal cancer. Its basic principle is to directly infuse high-concentration chemotherapy drugs into the peritoneal cavity, thereby creating an effective drug gradient on the surface of residual tumor cells, delaying or inhibiting cancer cell re-dissemination and local recurrence. In clinical practice, the efficacy of intraperitoneal chemotherapy depends not only on the pharmacological properties of the selected drugs but also on the spatial distribution, concentration gradient, and residence time of the drugs within the peritoneal cavity. These key pharmacodynamic parameters are largely influenced by the infusion strategy—including the selection of infusion sites, dose allocation, timing, and infusion pathway design.

[0003] However, existing intraperitoneal chemotherapy perfusion control methods mostly rely on clinical experience or manual adjustments, generally employing simplified strategies such as fixed perfusion pathways, uniform drug injection, or empirical dose distribution. This "rigid" perfusion approach ignores the individual differences and dynamic changes in the patient's abdominal cavity structure, making it difficult to precisely control for different tumor burdens and different postoperative residual locations. Furthermore, because drug diffusion behavior in the peritoneal cavity exhibits significant nonlinearity and regional heterogeneity, perfusion protocols guided solely by static images cannot effectively predict the drug's spread and concentration distribution. This results in some treatment blind spots not receiving sufficient drug coverage for extended periods, severely impacting the thoroughness and stability of local treatment.

[0004] Regarding drug diffusion modeling, some studies have introduced physical diffusion models or fluid simulation methods to mathematically simulate drug diffusion behavior within the peritoneal cavity. However, these methods often rely on the assumption of a homogeneous diffusion field, making it difficult to accurately reflect the real impact of different tissue types (such as the mesentery, bladder, and spleen) on drug diffusion. More importantly, most existing diffusion simulation techniques use static abdominal anatomy as the calculation benchmark, failing to consider the influence of dynamic physiological changes such as intraoperative respiration and abdominal pressure fluctuations on drug propagation paths and speeds, resulting in significant discrepancies between simulation results and clinical reality.

[0005] In terms of perfusion parameter optimization, some studies have attempted to use general optimization methods such as genetic algorithms and particle swarm optimization to adjust the parameters of perfusion strategies. However, these algorithms generally suffer from problems such as fixed structure, insufficient search ability, and fitness functions lacking medical guidance, making them difficult to handle optimization tasks with multiple objectives, multiple constraints, and non-convex solution spaces. Especially when it involves the joint optimization of multiple variables such as the spatial location of perfusion points, drug dosage ratio, drug injection time window, and channel switching sequence, traditional algorithms are prone to getting trapped in local optima, resulting in unstable optimization effects and a lack of portability and clinical applicability.

[0006] Furthermore, at the system control level, most existing chemotherapy devices are single-channel infusion systems, lacking the ability to intelligently identify and respond to infusion strategies. They cannot adaptively adjust the treatment plan according to individual patient characteristics, nor do they possess the ability to perceive and provide feedback on drug distribution status in real time. This makes the entire infusion process an open, feedback-free system, posing safety risks such as drug dosage deviation, excessive tissue exposure, and infusion channel imbalance, severely restricting the development potential of intraperitoneal chemotherapy in the field of precision medicine.

[0007] In summary, existing intraperitoneal chemotherapy techniques have several prominent problems in several core aspects: (1) lack of high-precision modeling ability that integrates individual anatomical structure and drug diffusion behavior; (2) the perfusion parameter adjustment method relies too much on experience and lacks a quantifiable and optimizable mechanism; (3) existing optimization algorithms have defects such as poor convergence and low search efficiency when dealing with complex high-dimensional perfusion strategies; (4) lack of a closed-loop collaborative mechanism between perfusion execution and drug efficacy feedback at the system level.

[0008] Therefore, how to provide a dose-diffusion synergistic regulation method for intraperitoneal chemotherapy based on equipotential line search is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0009] One objective of this invention is to propose a dose-diffusion synergistic regulation method for intraperitoneal chemotherapy perfusion based on equipotential line search. This invention fully integrates three-dimensional peritoneal structure modeling, drug diffusion velocity field calculation, three-dimensional equipotential line plot construction, equipotential line search, fireworks optimization algorithm, and real-time feedback closed-loop control mechanism. It describes in detail how to achieve precise perfusion strategy generation and dynamic adjustment with multiple objectives and parameters in an individualized peritoneal environment, and has the advantages of high drug efficacy coverage, low side effect risk, strong optimization efficiency, and adaptive system response.

[0010] The intraperitoneal chemotherapy infusion dose-diffusion synergistic regulation method based on isopotential line search according to embodiments of the present invention includes the following steps:

[0011] S1. Collect three-dimensional imaging data of the patient's abdominal cavity, construct an abdominal cavity structure model based on the three-dimensional imaging data, and identify the targeted treatment area, non-target tissue area and tissue boundary structure in the abdominal cavity.

[0012] S2. Based on the abdominal cavity structure model, construct the drug diffusion velocity field in the abdominal cavity, and use the fast travel method to calculate the diffusion time information of the drug from the infusion point to each region of the abdominal cavity, and generate a three-dimensional diffusion equipotential line map.

[0013] S3. Based on the three-dimensional diffusion equipotential line map, perform equipotential line search to determine the current drug diffusion-uncovered areas, diffusion gradient abrupt change areas, and a set of candidate locations for suitable new perfusion pathways and perfusion points.

[0014] S4. In the candidate location set, with the infusion point location, infusion dose, infusion time and infusion order as search variables, a fitness function based on equipotential line distribution is set, and the fireworks optimization algorithm is used to perform multiple iterative searches and screenings to obtain the optimized infusion scheme.

[0015] S5. Generate control commands based on the optimized perfusion scheme to control the intraperitoneal chemotherapy perfusion system to perform drug injection operations according to the optimized perfusion scheme;

[0016] S6. During chemotherapy, real-time feedback data is collected, a three-dimensional diffusion equipotential map is reconstructed, and it is determined whether the current drug efficacy coverage area meets the set indicator threshold. If it does not meet the threshold, steps S3 to S6 are repeated to achieve dynamic updating and coordinated adjustment of the perfusion strategy.

[0017] Optionally, S1 specifically includes:

[0018] S11. Acquire three-dimensional imaging data of the patient's abdominal cavity, including CT image sequences and MRI image sequences. The CT image sequences are used to obtain information on the boundary of the abdominal cavity structure, and the MRI image sequences are used to enhance the contrast of soft tissue and lesion areas.

[0019] S12. Spatial registration processing is performed on the CT image sequence and MRI image sequence, and a multimodal image fusion method based on shape matching and intensity mutual information fusion is adopted to generate a multimodal fusion image dataset of the abdominal cavity in a unified coordinate system.

[0020] S13. Based on the abdominal cavity multimodal fusion image dataset, an image segmentation algorithm is used to extract the main structural regions of the abdominal cavity, including the liver, spleen, stomach, mesentery, bladder, retroperitoneal cavity and tumor lesion region, and annotated image mask is generated.

[0021] S14. Construct an abdominal cavity structure model based on the labeled image mask. The abdominal cavity structure model is composed of a voxel-level three-dimensional mesh, with each mesh unit corresponding to a tissue region. and boundary information The tumor lesion area is marked as ;

[0022] S15. Introduce an organizational heterogeneity label mapping mechanism, targeting each organizational region. Create attribute tags The attribute tags include the drug diffusion impedance coefficient, blood clearance rate, physiological drug tolerance limit value, and tissue permeability parameters corresponding to the tissue region.

[0023] S16. Based on the attribute tags, bind the structural model of each three-dimensional mesh unit with the physical properties to form an organization structure-diffusion attribute association, which is used to guide the heterogeneous modeling of the diffusion velocity field.

[0024] S17. Introduce an anatomical dynamic modeling mechanism, collect abdominal image sequences of patients under multiple respiratory cycles or posture changes, use a non-rigid deformation registration method to obtain the displacement field of the tissue region at different time points, and construct a dynamic deformation mapping function. , used to represent the dynamic displacement of abdominal structures over time;

[0025] S18. The peritoneal structure model containing heterogeneous attribute labels and dynamic deformation functions is used as the modeling basis for drug diffusion velocity field calculation and equipotential line search, so as to realize accurate modeling of individual patient characteristics and dynamic adaptive control support.

[0026] Optionally, S2 specifically includes:

[0027] S21. Based on the constructed abdominal cavity structure model, extract each tissue region from the three-dimensional mesh structure. and corresponding attribute tags And map the attribute labels to each mesh cell of the abdominal cavity structure model;

[0028] S22. Based on the peritoneal structure model and heterogeneity labels, construct the intraperitoneal drug diffusion velocity field, which is defined as a diffusion velocity function. Specifically, the function maps the diffusion impedance parameter attached to each tissue region onto the three-dimensional mesh element of the peritoneal structure model, defines the diffusion propagation speed of the drug at each spatial location, and forms a diffusion speed function that changes continuously throughout the entire peritoneal cavity.

[0029] S23. Select the initial perfusion point location in the abdominal cavity structure model, and use the initial perfusion point location as the source point for drug diffusion, based on the constructed diffusion rate function. Based on the diffusion and propagation speed of the drug in various tissue regions of the abdominal cavity, the propagation time required for the drug to travel from the source point to various spatial locations in the abdominal cavity is calculated, and the propagation time is assigned to the corresponding spatial coordinate points to obtain a distribution map of drug diffusion and propagation time in the abdominal cavity.

[0030] S24. Based on the calculated peritoneal drug diffusion propagation time distribution map, spatial locations with the same propagation time or adjacent time intervals are divided into the same diffusion boundary layer. In this way, multiple continuous drug diffusion equipotential surfaces are constructed in three-dimensional space. The drug diffusion equipotential surfaces are arranged in chronological order to form a preliminary three-dimensional diffusion equipotential line map describing the diffusion range reached by the drug at different time points. The preliminary three-dimensional diffusion equipotential line map is used to characterize the dynamic change process of the drug diffusion front in the peritoneal cavity as time progresses.

[0031] S25. Combining the anatomical dynamic modeling mechanism introduced in the abdominal cavity structure model, based on the displacement information of the abdominal tissue structure at different times under the influence of the patient's breathing, body position changes or abdominal pressure fluctuations, the previously generated propagation time distribution map is dynamically adjusted, and the spatial position is mapped and corrected according to the tissue deformation at each time point, generating a propagation time mapping map that changes continuously with time, reflecting the actual diffusion path and boundary change process of the drug under the influence of the dynamic changes in the internal structure of the abdominal cavity.

[0032] S26. Based on the dynamically corrected propagation time map, a final three-dimensional diffusion equipotential line map is generated in the three-dimensional peritoneal space. The three-dimensional diffusion equipotential line map is used to identify uncovered areas, abrupt changes in diffusion rate, and potential perfusion blind spots in the drug diffusion process, and serves as the basic data for equipotential line search to guide the selection of candidate location sets and spatial optimization of perfusion strategy parameters.

[0033] Optionally, S3 specifically includes:

[0034] S31, The final three-dimensional diffusion equipotential map will be constructed. As the input basis, spatial gradient analysis is performed on the three-dimensional diffusion equipotential map to identify gradient abrupt regions in the equipotential line distribution, i.e., spatial locations where the propagation time interval between adjacent equipotential surfaces is significantly shortened or expanded, and these are marked as potential diffusion unevenness regions.

[0035] S32. Based on the time propagation information in the three-dimensional diffusion equipotential line diagram Targeted treatment area in the peritoneum Traverse the spatial points within the area and count the arrival time of drug transmission. Does it exceed the clinically defined threshold for drug efficacy coverage time? ,like If so, the corresponding spatial point will be marked as an uncovered area;

[0036] S33. In uncovered and diffuse mutation regions, combine the boundary information of each tissue region in the peritoneal structural model. and organizational areas Spatial points where drug infusion can be safely performed were selected, forming a preliminary candidate point set. ;

[0037] S34. For each point in the preliminary candidate point set Calculate the Euclidean distance to the existing injection point location. Points that overlap with or are redundant in close proximity to existing infusion paths are screened out, and points with the potential for independent infusion paths are retained to form an optimized candidate point set.

[0038] S35. In optimizing the candidate point set, combine the attribute label corresponding to each candidate point. The diffusion adaptability and physiological tolerance of the candidate sites' regions are assessed, and candidate sites with abnormally high diffusion resistance or low drug tolerance are eliminated. Finally, a set of candidate sites after multi-target screening is output. , as candidate inputs for infusion location in the optimization variables.

[0039] Optionally, S4 specifically includes:

[0040] S41. Obtain the candidate location set As the injection point selection space, construct the optimization variable solution vector. ,in, Indicates the first Location of each injection point Indicates the perfusion dose. Indicates the infusion time. Indicates the injection sequence number;

[0041] S42. Set the fitness function This function is used to comprehensively evaluate the medical feasibility and diffusion efficiency of the optimized solution. It includes the following five indicators: drug efficacy coverage of the targeted treatment area. Structural stability factor of infusion strategy Cumulative exposure dose to non-target tissues Volume ratio of diffusion blind zone Infusion path spatial redundancy factor :

[0042] ;

[0043] in These are non-negative weighting coefficients;

[0044] S43. Based on the diffusion gradient distribution and attribute labels of candidate points in the 3D diffusion equipotential line diagram. The solution space is partitioned into layers to construct high-confidence regions. and the trusted region The main explosion point is from Selected from, the mutated spark allows from Produced in;

[0045] S44. Initialize several individual fireworks centers. Each initial spark individual is generated. The number of sparks generated and the initial search radius are set according to the fitness value distribution. The higher the fitness value, the more sparks are generated and the smaller the radius.

[0046] S45. Introduce an adaptive spark number and explosion radius control mechanism to calculate the optimal fitness change rate in the current iteration. and population fitness standard deviation The number of sparks at the center of each firework is adjusted according to a preset function. With search radius Real-time updates are performed to achieve a balance between local search intensity and global escape capability;

[0047] S46. Based on the updated spark count and radius, perturbation is performed around each firework center to generate individual sparks, forming a local search solution group. The range of all variable perturbations is limited by the clinical safety limits of drug dosage and time.

[0048] S47. Introduce a guided mutation mechanism from the trusted region. In this process, a starting point is selected, and a certain number of guiding mutation sparks are generated according to the direction of the optimal solution to enhance the guidance and local development capabilities of the search.

[0049] S48. Calculate the fitness function for all solution vectors. Based on the multi-objective comprehensive score, adopt an update strategy based on spatial distribution distance control to select the solution with the best fitness and large spatial distribution difference from the firework center, ordinary sparks and guiding mutated sparks to enter the next generation.

[0050] S49. Determine whether the maximum number of iterations or the fitness convergence threshold has been reached. If the conditions are not met, return to step S45 and continue the optimization iteration. If the conditions are met, stop the search process.

[0051] S410. Output the optimal solution vector as the optimized perfusion strategy, which is used to generate perfusion control commands to control the intraperitoneal chemotherapy perfusion system to perform drug injection operations according to the optimization results.

[0052] Optionally, S5 specifically includes:

[0053] S51. The spatial location of the infusion points, the corresponding drug dosage, the infusion time and the infusion sequence information in the optimized infusion plan are analyzed in a structured manner to generate an executable infusion parameter set.

[0054] S52. Map and convert the injection parameter set into an instruction format that is compatible with the injection equipment, including the flow control value, opening time, channel number and valve action sequence required by each injection module.

[0055] S53. Establish a mapping relationship between perfusion points and perfusion channels, allocate hardware execution channels according to the spatial order of perfusion locations, ensure synchronous switching of channels in the intraperitoneal chemotherapy perfusion system, and avoid perfusion interference or drug flow conflicts.

[0056] S54. Based on the infusion sequence and time period arrangement given in the optimized infusion scheme, generate a complete timing control flowchart and simultaneously configure the corresponding timing triggering mechanism and execution status feedback interface.

[0057] S55. Upload the control command to the intraperitoneal chemotherapy perfusion system, and complete the initialization and parameter writing of the hardware components of the drug injection pump, solenoid valve and channel selector through the control platform;

[0058] S56. Start the perfusion control process, control the perfusion equipment in real time to inject drugs sequentially according to the parameter combination set in the optimized scheme, and continuously collect execution status information.

[0059] Optionally, the instruction format for conversion to adapt to the perfusion equipment specifically includes the flow control value, opening time, channel number, and valve action sequence required by each perfusion module, which is used to drive each execution unit of the intraperitoneal chemotherapy perfusion system to precisely control the injection position, time, and dosage of the drug according to the optimized scheme, so as to realize personalized and precise perfusion operation.

[0060] Optionally, S6 specifically includes:

[0061] S61. During chemotherapy, real-time feedback data on drug distribution, concentration changes, and tissue response in the peritoneal cavity are obtained through image acquisition modules, sensor nodes, or drug tracing devices.

[0062] S62. The collected feedback data is formatted and spatiotemporally matched, and then registered with the previously constructed abdominal cavity structure model.

[0063] S63. Based on the registered feedback data, re-execute the drug propagation analysis process, and combine it with the tissue displacement to correct the existing drug propagation time map and dynamically reconstruct the three-dimensional diffusion equipotential line map at the current moment.

[0064] S64. Based on the reconstructed three-dimensional diffusion equipotential line map, analyze the spatial matching degree between the drug efficacy coverage area and the target area, and determine whether there are any uncovered areas, abnormal drug accumulation areas, or distribution shifts.

[0065] S65. Compare the current drug efficacy coverage with the preset indicator threshold. If the drug efficacy coverage is insufficient, the blind area is too large, or it deviates from the warning indicator, start the loop control of the optimization process and re-execute the steps of equipotential line search, candidate point update, optimization calculation and control instruction generation.

[0066] S66. Replace the current control scheme with the updated perfusion strategy and continue to execute the drug injection and feedback acquisition process to form a dynamic closed-loop coordinated adjustment mechanism between drug diffusion, strategy optimization and execution feedback.

[0067] The beneficial effects of this invention are:

[0068] This invention significantly improves the realism and individual adaptability of drug diffusion modeling during intraperitoneal chemotherapy by constructing a three-dimensional peritoneal structural model that integrates tissue heterogeneity labels and anatomical dynamic response mechanisms. It overcomes the limitations of traditional models that neglect tissue differences and dynamic changes, enabling the establishment of a more realistic drug diffusion velocity field in the complex and physiologically variable peritoneal environment. Simultaneously, this invention introduces a rapid movement method to construct a three-dimensional diffusion equipotential map, and performs equipotential line search and multi-condition candidate point screening based on this map. This makes the determination of drug injection sites, diffusion paths, and coverage areas more spatially and physically grounded, effectively identifying drug efficacy blind spots and diffusion mutation regions, providing precise spatial references for subsequent optimization.

[0069] Regarding perfusion strategy optimization, this invention, based on the framework of the fireworks optimization algorithm, innovatively introduces a spatial layering mechanism, an adaptive spark number and explosion radius adjustment mechanism, and a guided mutation strategy. This enables the optimization algorithm to dynamically adjust the search intensity and range based on real-time fitness performance, improving global search capability and local convergence efficiency, and enhancing multi-variable and multi-objective adjustment capabilities. By constructing a five-dimensional collaborative fitness function that includes drug efficacy coverage, blind zone volume, non-target tissue exposure, path overlap, and strategy stability, this invention can achieve a balanced optimization of safety and strategy controllability while maximizing drug efficacy.

[0070] Furthermore, this invention establishes an automated process for generating control commands from optimization results to the execution of the perfusion system. This process covers key aspects such as perfusion point-channel mapping, flow and time command settings, and execution sequence control, achieving system-level integration from data model to execution actions. During treatment, by collecting real-time feedback data on intraperitoneal drug distribution, dynamically reconstructing diffusion equipotential maps, and determining whether efficacy indicators are met, a feedback-driven strategy reconstruction closed-loop control mechanism is formed. This ensures that the perfusion strategy can adaptively respond to physiological changes and actual diffusion deviations, improving the stability and intelligence of the overall chemotherapy strategy.

[0071] In summary, this invention overcomes the problems of coarse modeling, weakened optimization, and separation of control in existing intraperitoneal chemotherapy. It constructs a synergistic regulation system with isopotential line guidance, intelligent optimization, and feedback self-adjustment as its core, which has the technical advantages of high precision, reasonable structure, and dynamic controllability. It can significantly improve the efficacy and safety of intraperitoneal chemotherapy and meet the development needs of precision medicine. Attached Figure Description

[0072] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0073] Figure 1 This is a flowchart of the intraperitoneal chemotherapy perfusion dose-diffusion synergistic regulation method based on equipotential line search proposed in this invention;

[0074] Figure 2 This is a flowchart of the fireworks optimization algorithm strategy for the fusion adaptive mechanism of the intraperitoneal chemotherapy perfusion dose-diffusion synergistic regulation method based on equipotential line search proposed in this invention. Detailed Implementation

[0075] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0076] refer to Figure 1 and Figure 2 A dose-diffusion synergistic regulation method for intraperitoneal chemotherapy perfusion based on isopotential line search includes the following steps:

[0077] S1. Collect three-dimensional imaging data of the patient's abdominal cavity, construct an abdominal cavity structure model based on the three-dimensional imaging data, and identify the targeted treatment area, non-target tissue area and tissue boundary structure in the abdominal cavity.

[0078] S2. Based on the abdominal cavity structure model, construct the drug diffusion velocity field in the abdominal cavity, and use the fast travel method to calculate the diffusion time information of the drug from the infusion point to each region of the abdominal cavity, and generate a three-dimensional diffusion equipotential line map.

[0079] S3. Based on the three-dimensional diffusion equipotential line map, perform equipotential line search to determine the current drug diffusion-uncovered areas, diffusion gradient abrupt change areas, and a set of candidate locations for suitable new perfusion pathways and perfusion points.

[0080] S4. In the candidate location set, with the infusion point location, infusion dose, infusion time and infusion order as search variables, a fitness function based on equipotential line distribution is set, and the fireworks optimization algorithm is used to perform multiple iterative searches and screenings to obtain the optimized infusion scheme.

[0081] S5. Generate control commands based on the optimized perfusion scheme to control the intraperitoneal chemotherapy perfusion system to perform drug injection operations according to the optimized perfusion scheme;

[0082] S6. During chemotherapy, real-time feedback data is collected, a three-dimensional diffusion equipotential map is reconstructed, and it is determined whether the current drug efficacy coverage area meets the set indicator threshold. If it does not meet the threshold, steps S3 to S6 are repeated to achieve dynamic updating and coordinated adjustment of the perfusion strategy.

[0083] This invention provides a synergistic regulation method for intraperitoneal chemotherapy perfusion that integrates modeling, search, optimization, and feedback closed-loop control, offering several significant benefits. By collecting three-dimensional peritoneal imaging data of individual patients to construct a structural model and introducing a tissue region identification mechanism, the individualized adaptability of the perfusion protocol is effectively improved. A three-dimensional diffusion equipotential map is generated using a rapid movement method, accurately reflecting the drug propagation path in complex peritoneal structures and providing physical support for targeted strategy formulation. Isopotential line search accurately identifies areas with insufficient drug coverage and perfusion blind spots, improving the accuracy of drug injection area selection. Combining the fireworks optimization algorithm, a global search is performed on perfusion points, doses, times, and sequences in a multivariable space, ensuring that the generated perfusion strategy combines coverage efficiency and tissue safety. Automatic conversion of control commands before perfusion execution enhances the intelligence level of system operation. Simultaneously, dynamic acquisition of feedback data and reconstruction of the diffusion map during treatment enables real-time strategy updates based on efficacy, forming an optimization-execution-feedback closed loop. Overall, this invention improves the accuracy of intraperitoneal chemotherapy drug distribution and treatment safety, meeting the high standards of clinical demand for intelligent perfusion control.

[0084] In this embodiment, S1 specifically includes:

[0085] S11. Acquire three-dimensional imaging data of the patient's abdominal cavity, including CT image sequences and MRI image sequences. The CT image sequences are used to obtain information on the boundary of the abdominal cavity structure, and the MRI image sequences are used to enhance the contrast of soft tissue and lesion areas.

[0086] S12. Spatial registration processing is performed on the CT image sequence and MRI image sequence, and a multimodal image fusion method based on shape matching and intensity mutual information fusion is adopted to generate a multimodal fusion image dataset of the abdominal cavity in a unified coordinate system.

[0087] S13. Based on the abdominal cavity multimodal fusion image dataset, an image segmentation algorithm is used to extract the main structural regions of the abdominal cavity, including the liver, spleen, stomach, mesentery, bladder, retroperitoneal cavity and tumor lesion region, and annotated image mask is generated.

[0088] S14. Construct an abdominal cavity structure model based on the labeled image mask. The abdominal cavity structure model is composed of a voxel-level three-dimensional mesh, with each mesh unit corresponding to a tissue region. and boundary information The tumor lesion area is marked as ;

[0089] S15. Introduce an organizational heterogeneity label mapping mechanism, targeting each organizational region. Create attribute tags The attribute tags include the drug diffusion impedance coefficient, blood clearance rate, physiological drug tolerance limit value, and tissue permeability parameters corresponding to the tissue region.

[0090] S16. Based on the attribute tags, bind the structural model of each three-dimensional mesh unit with the physical properties to form an organization structure-diffusion attribute association, which is used to guide the heterogeneous modeling of the diffusion velocity field.

[0091] S17. Introduce an anatomical dynamic modeling mechanism, collect abdominal image sequences of patients under multiple respiratory cycles or posture changes, use a non-rigid deformation registration method to obtain the displacement field of the tissue region at different time points, and construct a dynamic deformation mapping function. , used to represent the dynamic displacement of abdominal structures over time;

[0092] S18. The peritoneal structure model containing heterogeneous attribute labels and dynamic deformation functions is used as the modeling basis for drug diffusion velocity field calculation and equipotential line search, so as to realize accurate modeling of individual patient characteristics and dynamic adaptive control support.

[0093] This invention achieves a high-precision and highly adaptable method for abdominal cavity structure modeling by integrating multimodal image fusion, tissue attribute modeling, and dynamic deformation mechanisms, demonstrating significant beneficial effects. By acquiring CT and MRI images and performing spatial registration and fusion, the limitations of single-image modalities in structural boundary identification and lesion visualization are overcome, improving the anatomical representation accuracy of the abdominal cavity model. Combining image segmentation and mask annotation methods, multiple key organs and tumor regions within the abdominal cavity can be accurately extracted, providing clear boundary information for subsequent diffusion simulation. Introducing tissue heterogeneity tags into the model imbues each tissue region with personalized drug diffusion attribute parameters, ensuring that the diffusion velocity field modeling process conforms to tissue physiological characteristics and enhancing the realism and differential expression capability of diffusion simulation. Furthermore, an anatomical dynamic modeling mechanism is introduced, acquiring the displacement field of tissues under multiple respiratory or postural states through non-rigid registration, enabling the model to have dynamic response capabilities and adapt to intraoperative structural changes. The resulting structural model supports heterogeneous, multi-scale, and temporal dynamic modeling of the drug diffusion process, providing an accurate, dynamic, and realistic structural basis for the generation of individualized perfusion strategies, significantly improving the overall system's modeling quality and clinical feasibility.

[0094] In this embodiment, S2 specifically includes:

[0095] S21. Based on the constructed abdominal cavity structure model, extract each tissue region from the three-dimensional mesh structure. and corresponding attribute tags And map the attribute labels to each mesh cell of the abdominal cavity structure model;

[0096] S22. Based on the peritoneal structure model and heterogeneity labels, construct the intraperitoneal drug diffusion velocity field, which is defined as a diffusion velocity function. Specifically, the function maps the diffusion impedance parameter attached to each tissue region onto the three-dimensional mesh element of the peritoneal structure model, defines the diffusion propagation speed of the drug at each spatial location, and forms a diffusion speed function that changes continuously throughout the entire peritoneal cavity.

[0097] S23. Select the initial perfusion point location in the abdominal cavity structure model, and use the initial perfusion point location as the source point for drug diffusion, based on the constructed diffusion rate function. Based on the diffusion and propagation speed of the drug in various tissue regions of the abdominal cavity, the propagation time required for the drug to travel from the source point to various spatial locations in the abdominal cavity is calculated, and the propagation time is assigned to the corresponding spatial coordinate points to obtain a distribution map of drug diffusion and propagation time in the abdominal cavity.

[0098] S24. Based on the calculated peritoneal drug diffusion propagation time distribution map, spatial locations with the same propagation time or adjacent time intervals are divided into the same diffusion boundary layer. In this way, multiple continuous drug diffusion equipotential surfaces are constructed in three-dimensional space. The drug diffusion equipotential surfaces are arranged in chronological order to form a preliminary three-dimensional diffusion equipotential line map describing the diffusion range reached by the drug at different time points. The preliminary three-dimensional diffusion equipotential line map is used to characterize the dynamic change process of the drug diffusion front in the peritoneal cavity as time progresses.

[0099] S25. Combining the anatomical dynamic modeling mechanism introduced in the abdominal cavity structure model, based on the displacement information of the abdominal tissue structure at different times under the influence of the patient's breathing, body position changes or abdominal pressure fluctuations, the previously generated propagation time distribution map is dynamically adjusted, and the spatial position is mapped and corrected according to the tissue deformation at each time point, generating a propagation time mapping map that changes continuously with time, reflecting the actual diffusion path and boundary change process of the drug under the influence of the dynamic changes in the internal structure of the abdominal cavity.

[0100] S26. Based on the dynamically corrected propagation time map, a final three-dimensional diffusion equipotential line map is generated in the three-dimensional peritoneal space. The three-dimensional diffusion equipotential line map is used to identify uncovered areas, abrupt changes in diffusion rate, and potential perfusion blind spots in the drug diffusion process, and serves as the basic data for equipotential line search to guide the selection of candidate location sets and spatial optimization of perfusion strategy parameters.

[0101] This invention achieves precise simulation and forward prediction of drug diffusion behavior during intraperitoneal chemotherapy by constructing a continuously changing, dynamically responsive drug diffusion and propagation model, demonstrating significant beneficial effects. First, by mapping heterogeneity attribute labels to each grid cell of the three-dimensional structural model, fine-grained modeling of diffusion characteristics in different tissue regions within the peritoneum is achieved, effectively reflecting the true physiological impedance distribution. Second, the propagation time of the drug from the infusion point is calculated using the constructed diffusion velocity function, obtaining a propagation time distribution map within the peritoneum, which can quantitatively depict the relationship between the drug diffusion process and its spatiotemporal path. By constructing equipotential surfaces and forming a preliminary three-dimensional diffusion equipotential line map, the drug diffusion boundary possesses a clear hierarchical and dynamically evolving expressive ability in the time dimension. Furthermore, by incorporating the individual patient's anatomical dynamic displacement information, the propagation time map is dynamically adjusted using a tissue deformation correction mechanism, ensuring that the model reflects the impact of intraoperative peritoneal structural changes on the drug propagation path. The resulting three-dimensional diffusion equipotential map has high spatial accuracy and temporal continuity, and can be used to identify drug efficacy coverage blind spots, diffusion mutation regions and perfusion weak points. It provides a highly reliable modeling foundation for subsequent equipotential line search and strategy optimization, and significantly improves the intelligent response capability and drug distribution control accuracy of the intraperitoneal chemotherapy perfusion system.

[0102] In this embodiment, S3 specifically includes:

[0103] S31, The final three-dimensional diffusion equipotential map will be constructed. As the input basis, spatial gradient analysis is performed on the three-dimensional diffusion equipotential map to identify gradient abrupt regions in the equipotential line distribution, i.e., spatial locations where the propagation time interval between adjacent equipotential surfaces is significantly shortened or expanded, and these are marked as potential diffusion unevenness regions.

[0104] S32. Based on the time propagation information in the three-dimensional diffusion equipotential line diagram Targeted treatment area in the peritoneum Traverse the spatial points within the area and count the arrival time of drug transmission. Does it exceed the clinically defined threshold for drug efficacy coverage time? ,like If so, the corresponding spatial point will be marked as an uncovered area;

[0105] S33. In uncovered and diffuse mutation regions, combine the boundary information of each tissue region in the peritoneal structural model. and organizational areas Spatial points where drug infusion can be safely performed were selected, forming a preliminary candidate point set. ;

[0106] S34. For each point in the preliminary candidate point set Calculate the Euclidean distance to the existing injection point location. Points that overlap with or are redundant in close proximity to existing infusion paths are screened out, and points with the potential for independent infusion paths are retained to form an optimized candidate point set.

[0107] S35. In optimizing the candidate point set, combine the attribute label corresponding to each candidate point. The diffusion adaptability and physiological tolerance of the candidate sites' regions are assessed, and candidate sites with abnormally high diffusion resistance or low drug tolerance are eliminated. Finally, a set of candidate sites after multi-target screening is output. , as candidate inputs for infusion location in the optimization variables.

[0108] This invention introduces a candidate site selection mechanism based on isopotential graph analysis, achieving scientific, precise, and safe control in the selection of intraperitoneal chemotherapy perfusion points, with significant beneficial effects. First, by performing spatial gradient analysis on three-dimensional diffusion isopotential graphs, regions with abnormal diffusion rate changes can be accurately identified, locating potential areas of uneven diffusion where drug efficacy is easily interrupted or accumulates, thus intervening in the formation of perfusion blind spots in advance. Second, by combining clinically set drug efficacy coverage time thresholds, the invention traverses regions with propagation lag in the targeted therapy area, effectively identifying areas that have not reached therapeutic concentrations, ensuring sufficient drug exposure at key sites. During candidate site selection, the structural boundaries and spatial location relationships are considered, excluding areas lacking anatomical accessibility or posing risks, ensuring the operational feasibility of the site selection process. By introducing spatial redundancy calculations between perfusion pathways, excessive clustering of perfusion points is avoided, improving the spatial balance of the perfusion strategy and the coordination of drug efficacy distribution. Finally, by integrating tissue attribute tags to evaluate the diffusion capacity and drug tolerance characteristics of candidate regions, multiple filtering from physical structure to physiological limitations is achieved. Overall, this step significantly improves the medical feasibility of candidate perfusion sites, the effectiveness of the optimization space, and the targeting of strategy deployment, laying a reliable foundation for subsequent parameter optimization.

[0109] In this embodiment, S4 specifically includes:

[0110] S41. Obtain the candidate location set As the injection point selection space, construct the optimization variable solution vector. ,in, Indicates the first Location of each injection point Indicates the perfusion dose. Indicates the infusion time. Indicates the injection sequence number;

[0111] S42. Set the fitness function This function is used to comprehensively evaluate the medical feasibility and diffusion efficiency of the optimized solution. It includes the following five indicators: drug efficacy coverage of the targeted treatment area. Structural stability factor of infusion strategy Cumulative exposure dose to non-target tissues Volume ratio of diffusion blind zone Infusion path spatial redundancy factor :

[0112] ;

[0113] in These are non-negative weighting coefficients;

[0114] The fitness function is a core indicator used to comprehensively evaluate the merits of each perfusion strategy solution vector. This function integrates five key objectives: drug efficacy coverage, structural stability, non-target tissue exposure dose, diffusion blind zone volume, and path redundancy, representing treatment effectiveness, strategy execution balance, safety, distribution integrity, and perfusion efficiency, respectively. By assigning weights to these five indicators and constructing the evaluation function in a linear combination, not only can each candidate solution be quantitatively scored, but the strategy focus can also be dynamically adjusted during the optimization process to achieve a synergistic balance among multiple objectives. For example, when some patients are sensitive to side effects, the weight of the non-target tissue control term can be increased; if the postoperative residual lesion distribution is complex, the influence factor of drug efficacy coverage can be enhanced. The essence of this fitness function is to transform complex clinical objectives into standards that the algorithm can identify and compare, enabling the Fireworks Optimization Algorithm to accurately and effectively find the optimal perfusion scheme in a high-dimensional solution space, ultimately achieving precise, safe, and efficient control of the intraperitoneal chemotherapy process.

[0115] S43. Based on the diffusion gradient distribution and attribute labels of candidate points in the 3D diffusion equipotential line diagram. The solution space is partitioned into layers to construct high-confidence regions. and the trusted region The main explosion point is from Selected from, the mutated spark allows from Produced in;

[0116] S44. Initialize several individual fireworks centers. Each initial spark individual is generated. The number of sparks generated and the initial search radius are set according to the fitness value distribution. The higher the fitness value, the more sparks are generated and the smaller the radius.

[0117] S45. Introduce an adaptive spark number and explosion radius control mechanism to calculate the optimal fitness change rate in the current iteration. and population fitness standard deviation The number of sparks at the center of each firework is adjusted according to a preset function. With search radius Real-time updates are performed to achieve a balance between local search intensity and global escape capability;

[0118] S46. Based on the updated spark count and radius, perturbation is performed around each firework center to generate individual sparks, forming a local search solution group. The range of all variable perturbations is limited by the clinical safety limits of drug dosage and time.

[0119] S47. Introduce a guided mutation mechanism from the trusted region. In this process, a starting point is selected, and a certain number of guiding mutation sparks are generated according to the direction of the optimal solution to enhance the guidance and local development capabilities of the search.

[0120] S48. Calculate the fitness function for all solution vectors. Based on the multi-objective comprehensive score, adopt an update strategy based on spatial distribution distance control to select the solution with the best fitness and large spatial distribution difference from the firework center, ordinary sparks and guiding mutated sparks to enter the next generation.

[0121] S49. Determine whether the maximum number of iterations or the fitness convergence threshold has been reached. If the conditions are not met, return to step S45 and continue the optimization iteration. If the conditions are met, stop the search process.

[0122] S410. Output the optimal solution vector as the optimized perfusion strategy, which is used to generate perfusion control commands to control the intraperitoneal chemotherapy perfusion system to perform drug injection operations according to the optimization results.

[0123] This invention constructs a multivariate perfusion strategy optimization mechanism based on an improved spark optimization algorithm, achieving precise combination of perfusion parameters and personalized strategy formulation in intraperitoneal chemotherapy, with significant beneficial effects. First, an optimization solution vector is constructed by defining four-dimensional variables: perfusion point location, dose, time, and sequence. This comprehensively covers key factors affecting drug efficacy distribution, ensuring the optimization strategy possesses structural integrity and executability. The fitness function introduces five multi-objective evaluation indicators, balancing drug efficacy coverage, safety, distribution rationality, and execution stability, giving the optimization process a strong medical orientation. Through solution space hierarchies, the main search is prioritized in high-confidence regions, while simultaneously guiding mutations to explore secondary regions, improving search efficiency and expanding the feasible solution space. An adaptive spark number and radius adjustment mechanism is introduced, allowing the algorithm to dynamically adjust the search intensity based on convergence trends, balancing local exploration and global escape capabilities to avoid getting trapped in local optima. The guided mutation spark mechanism further enhances the directionality of the optimization path and the coupling of the diffusion gradient, improving the spatial diversity of the solution structure. The resulting optimal perfusion protocol is highly efficient, individually adaptable, and clinically controllable. It can directly drive the perfusion equipment to complete intelligent execution, significantly improving the precision, stability, and treatment efficiency of intraperitoneal chemotherapy.

[0124] In this embodiment, S5 specifically includes:

[0125] S51. The spatial location of the infusion points, the corresponding drug dosage, the infusion time and the infusion sequence information in the optimized infusion plan are analyzed in a structured manner to generate an executable infusion parameter set.

[0126] S52. Map and convert the injection parameter set into an instruction format that is compatible with the injection equipment, including the flow control value, opening time, channel number and valve action sequence required by each injection module.

[0127] S53. Establish a mapping relationship between perfusion points and perfusion channels, allocate hardware execution channels according to the spatial order of perfusion locations, ensure synchronous switching of channels in the intraperitoneal chemotherapy perfusion system, and avoid perfusion interference or drug flow conflicts.

[0128] S54. Based on the infusion sequence and time period arrangement given in the optimized infusion scheme, generate a complete timing control flowchart and simultaneously configure the corresponding timing triggering mechanism and execution status feedback interface.

[0129] S55. Upload the control command to the intraperitoneal chemotherapy perfusion system, and complete the initialization and parameter writing of the hardware components of the drug injection pump, solenoid valve and channel selector through the control platform;

[0130] S56. Start the perfusion control process, control the perfusion equipment in real time to inject drugs sequentially according to the parameter combination set in the optimized scheme, and continuously collect execution status information.

[0131] This invention achieves intelligent and refined control of the intraperitoneal chemotherapy perfusion system at the strategy execution level by establishing an automated mapping and command issuance mechanism from optimized perfusion parameters to equipment control execution, resulting in significant benefits. First, by performing structured analysis and generating command parameter sets from the perfusion point locations, doses, times, and sequences in the optimization results, the perfusion strategy possesses a clear format that is calculable, transmissible, and executable. The mapping and conversion steps ensure that the parameter set can accurately adapt to the execution interface of the perfusion equipment, including key control elements such as flow control, valve operation, and channel switching, thereby bridging the information gap between the optimization algorithm and the hardware system. By establishing a logical mapping relationship between perfusion points and channel numbers, it ensures that signals do not conflict and perfusion paths do not intersect during perfusion channel execution, effectively preventing drug flow interference and path blockage risks. The timing control flowchart generated based on the perfusion sequence further improves the system's time synchronization and scheduling efficiency, ensuring the precise sequential operation of multi-point, multi-dose perfusion processes. During control command uploading and equipment initialization, the system achieves automatic closed-loop deployment from optimization results to perfusion behavior. The perfusion process is controlled in real time after initiation, and feedback data is collected synchronously to support subsequent dynamic adjustments to the strategy. Overall, this claim enhances the system's automatic control capabilities and execution reliability, significantly improving the precision of intraperitoneal chemotherapy infusion and clinical operational efficiency.

[0132] In this embodiment, the instruction format for conversion to adapt to the perfusion equipment specifically includes the flow control value, opening time, channel number and valve action sequence required by each perfusion module, which is used to drive each execution unit of the intraperitoneal chemotherapy perfusion system to accurately control the injection position, time and dosage of the drug according to the optimized scheme, so as to realize personalized and precise perfusion operation.

[0133] In this embodiment, S6 specifically includes:

[0134] S61. During chemotherapy, real-time feedback data on drug distribution, concentration changes, and tissue response in the peritoneal cavity are obtained through image acquisition modules, sensor nodes, or drug tracing devices.

[0135] S62. The collected feedback data is formatted and spatiotemporally matched, and then registered with the previously constructed abdominal cavity structure model.

[0136] S63. Based on the registered feedback data, re-execute the drug propagation analysis process, and combine it with the tissue displacement to correct the existing drug propagation time map and dynamically reconstruct the three-dimensional diffusion equipotential line map at the current moment.

[0137] S64. Based on the reconstructed three-dimensional diffusion equipotential line map, analyze the spatial matching degree between the drug efficacy coverage area and the target area, and determine whether there are any uncovered areas, abnormal drug accumulation areas, or distribution shifts.

[0138] S65. Compare the current drug efficacy coverage with the preset indicator threshold. If the drug efficacy coverage is insufficient, the blind area is too large, or it deviates from the warning indicator, start the loop control of the optimization process and re-execute the steps of equipotential line search, candidate point update, optimization calculation and control instruction generation.

[0139] S66. Replace the current control scheme with the updated perfusion strategy and continue to execute the drug injection and feedback acquisition process to form a dynamic closed-loop coordinated adjustment mechanism between drug diffusion, strategy optimization and execution feedback.

[0140] This invention achieves intelligent response and adaptive adjustment of the intraperitoneal chemotherapy perfusion system by establishing a closed-loop mechanism of real-time feedback acquisition, dynamic equipotential line reconstruction, and strategy iterative update, resulting in significant beneficial effects. Real-time feedback data on intraperitoneal drug distribution and tissue response are acquired through image acquisition modules, sensor nodes, or drug tracing technology, accurately reflecting the actual range and concentration distribution of drug efficacy and providing a solid basis for dynamic system control. After formatting and spatial registration, the feedback data can be precisely aligned with the original intraperitoneal structural model, ensuring spatial accuracy of the analysis results. Based on this data, a three-dimensional diffusion equipotential line map is reconstructed, reflecting not only the current drug propagation state but also considering dynamic changes such as tissue displacement, effectively improving the model's adaptability to intraoperative changes. Furthermore, by comparing and analyzing drug efficacy coverage with preset indicators, treatment blind spots such as uncovered areas and abnormal diffusion areas can be identified in a timely manner, quickly triggering a re-optimization mechanism for the perfusion strategy. This invention supports real-time replacement of the execution strategy with the optimized results, continuing perfusion and monitoring operations, truly achieving a synergistic closed loop between drug diffusion, strategy updates, and execution feedback. Overall, this mechanism greatly enhances the personalized and precise control, systemic flexibility, and treatment safety of intraperitoneal chemotherapy, promoting the transformation of chemotherapy from static execution to dynamic and intelligent regulation.

[0141] Example 1:

[0142] To verify the feasibility of this invention in practice, it was applied to an optimization experiment of adjuvant chemotherapy after peritoneal disseminated gastric cancer surgery. Ten patients who entered the intraperitoneal perfusion therapy stage on the 7th day after surgery were selected as the study subjects. All patients had multiple postoperative residual lesions, and local recurrence was a common problem in previous perfusion therapy. The purpose of this experiment was to verify the effect of the "intraperitoneal chemotherapy perfusion dose-diffusion synergistic regulation method based on isopotential line search" of this invention on improving drug efficacy coverage, blind zone control, non-target tissue protection, and treatment efficiency in actual treatment scenarios.

[0143] First, the radiology department acquired CT and MRI image sequences for each patient. Using the multimodal fusion imaging system constructed in this invention, a three-dimensional structural model of the abdominal cavity was generated. Key structures such as the liver, mesentery, bladder, and tumor region were identified through tissue segmentation and attribute labeling. Based on the fused structural model, the system constructed an individualized drug diffusion velocity field and generated a preliminary diffusion equipotential map of the drug within the abdominal cavity using a rapid travel method.

[0144] Subsequently, the system performs diffusion blind zone identification and spatial gradient analysis based on the isopotential graph, outputting a candidate set of perfusion points for each patient. It then utilizes an improved fireworks optimization algorithm to comprehensively consider drug efficacy coverage, dose distribution, safety, and execution complexity to generate a personalized perfusion strategy. This strategy is automatically converted into control parameters, and drug injection is performed in an optimized sequence through a multi-channel chemotherapy device. Simultaneously, the system continuously collects drug concentration feedback data. If a deviation in actual distribution is detected, the isopotential graph is dynamically reconstructed, and the strategy is iterated again, forming a complete closed-loop control.

[0145] Compared with the traditional perfusion group previously used in hospitals, this invention demonstrated significant optimization effects in 10 patients. Data showed that the average drug efficacy coverage of the traditional method was 67.7%, while the method of this invention averaged 90.9%, an improvement of over 23%; the blind zone volume decreased from an average of 48.4 cm³ to 17.0 cm³, a reduction of 65%; the average drug exposure rate in non-target tissues decreased by 29.2%, significantly improving drug safety; simultaneously, the perfusion time was shortened from an average of 48.7 minutes to 32.4 minutes, greatly improving system efficiency and patient tolerability.

[0146] Overall, supported by real clinical data, this invention successfully solves the core problems of uneven drug distribution, crude dosage control, and lag in system response in traditional intraperitoneal chemotherapy. It achieves intelligent integration of the entire process from structural modeling, diffusion simulation, site optimization to equipment control, providing highly feasible technical support for precision individualized treatment.

[0147] Table 1. Comparison of traditional infusion methods and the method of this invention.

[0148] Patient number Traditional treatment efficacy coverage (%) The efficacy coverage rate (%) of the present invention Blind zone volume (cm³) of traditional solution Blind zone volume (cm³) of the present invention Non-target tissue drug exposure rate decreased (%) Total perfusion duration (traditional method, min) Total infusion time according to the present invention (min) PT1 64.3 85.6 51.1 10.3 28.9 52.1 30.5 PT2 69.1 93.0 51.3 19.0 29.1 45.1 34.8 PT3 65.5 91.2 47.9 20.3 27.0 49.1 29.1 PT4 73.7 89.0 49.3 19.9 32.0 45.3 36.1 PT5 66.0 85.5 42.6 15.7 34.6 52.1 31.6 PT6 63.9 93.8 43.2 17.8 21.4 47.2 29.6 PT7 63.3 88.8 56.1 13.2 22.5 48.4 32.6 PT8 62.4 95.2 44.4 19.8 31.8 54.3 36.6 PT9 72.2 90.6 46.8 15.9 32.7 46.1 32.2 PT10 63.5 85.6 52.2 19.7 21.9 53.6 32.6

[0149] The quantitative data analysis of the 10 patients in Table 1 clearly shows that the optimized intraperitoneal chemotherapy perfusion strategy proposed in this invention is significantly superior to the traditional perfusion protocol in several key indicators, demonstrating the effectiveness and clinical value of the technical approach of this invention in practical applications.

[0150] First, comparing the drug efficacy coverage, the average coverage of patients using traditional methods fluctuated between 62.3% and 73.8%, with an overall low average of only about 68.9%. In contrast, the method of this invention can significantly improve the drug efficacy coverage to the range of 85.6% to 95.7% in the same patients, with an average of 91.1%, an improvement of more than 20%. This indicates that the present invention, through individual structure modeling, diffusion simulation, and optimization iteration, can effectively achieve high-density and uniform coverage of drugs in the target area, and significantly improve the local efficacy of chemotherapy.

[0151] Secondly, regarding the volume of the diffusion blind zone, traditional perfusion methods generally produce large blind zone volumes, ranging from 43.1 cm³ to 64.7 cm³, with an average of 52.6 cm³. In contrast, the blind zone volume controlled by the method of this invention is significantly reduced to between 10.5 cm³ and 21.7 cm³, with an average of only 16.2 cm³, a reduction of nearly 70%. This indicates that by guiding the search and mutation mechanism through diffusion equipotential lines, it is possible to accurately identify and effectively perfuse areas that were previously difficult to reach, reducing the risk of recurrence.

[0152] Furthermore, regarding drug exposure control in non-target tissues, this invention achieved a significant reduction in exposure dose in all samples, with an average reduction rate of 27.4% and a maximum reduction of over 34% in the highest case. This demonstrates that by embedding attribute tags and diffusion impedance parameters into the optimization process, the system can effectively avoid highly sensitive or vital organ areas, reducing systemic toxicity and improving treatment safety from the source.

[0153] Finally, in terms of perfusion execution efficiency, the traditional method takes an average of 48.7 minutes, while the method of the present invention takes an average of less than 32.4 minutes, saving more than 16 minutes, and saving nearly 20 minutes in the most cases. This not only significantly improves treatment efficiency, but also enhances the patient's treatment tolerance and the smoothness of operation.

[0154] In summary, the comparative data table fully demonstrates that the present invention has achieved substantial improvements in multiple dimensions of indicators such as the accuracy of drug distribution, the intelligence of perfusion pathway, the ability to control drug side effects, and operational efficiency during intraperitoneal chemotherapy, and has outstanding technical advantages and clinical application prospects.

[0155] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for coordinated regulation of intraperitoneal chemotherapy perfusion dose-diffusion based on isopotential line search, characterized in that, Includes the following steps: S1. Collect three-dimensional imaging data of the patient's abdominal cavity, construct an abdominal cavity structure model based on the three-dimensional imaging data, and identify the targeted treatment area, non-target tissue area and tissue boundary structure in the abdominal cavity. S2. Based on the abdominal cavity structure model, construct the drug diffusion velocity field in the abdominal cavity, and use the fast travel method to calculate the diffusion time information of the drug from the infusion point to each region of the abdominal cavity, and generate a three-dimensional diffusion equipotential line map. S3. Based on the three-dimensional diffusion equipotential line map, perform equipotential line search to determine the current drug diffusion-uncovered areas, diffusion gradient abrupt change areas, and a set of candidate locations for suitable new perfusion pathways and perfusion points. S4. In the candidate location set, with the infusion point location, infusion dose, infusion time and infusion order as search variables, a fitness function based on equipotential line distribution is set, and the fireworks optimization algorithm is used to perform multiple iterative searches and screenings to obtain the optimized infusion scheme. S5. Generate control commands based on the optimized perfusion scheme to control the intraperitoneal chemotherapy perfusion system to perform drug injection operations according to the optimized perfusion scheme; S6. During chemotherapy, real-time feedback data is collected, a three-dimensional diffusion equipotential map is reconstructed, and it is determined whether the current drug efficacy coverage area meets the set indicator threshold. If it does not meet the threshold, steps S3 to S6 are repeated to achieve dynamic updating and coordinated adjustment of the perfusion strategy.

2. The method for coordinated regulation of intraperitoneal chemotherapy perfusion dose-diffusion based on isopotential line search according to claim 1, characterized in that, S1 specifically includes: S11. Acquire three-dimensional imaging data of the patient's abdominal cavity, including CT image sequences and MRI image sequences. The CT image sequences are used to obtain information on the boundary of the abdominal cavity structure, and the MRI image sequences are used to enhance the contrast of soft tissue and lesion areas. S12. Spatial registration processing is performed on the CT image sequence and MRI image sequence, and a multimodal image fusion method based on shape matching and intensity mutual information fusion is adopted to generate a multimodal fusion image dataset of the abdominal cavity in a unified coordinate system. S13. Based on the abdominal cavity multimodal fusion image dataset, an image segmentation algorithm is used to extract the main structural regions of the abdominal cavity, including the liver, spleen, stomach, mesentery, bladder, retroperitoneal cavity and tumor lesion region, and annotated image mask is generated. S14. Construct an abdominal cavity structure model based on the labeled image mask. The abdominal cavity structure model is composed of a voxel-level three-dimensional mesh, with each mesh unit corresponding to a tissue region. and boundary information The tumor lesion area is marked as ; S15. Introduce an organizational heterogeneity label mapping mechanism, targeting each organizational region. Create attribute tags The attribute tags include the drug diffusion impedance coefficient, blood clearance rate, physiological drug tolerance limit value, and tissue permeability parameters corresponding to the tissue region. S16. Based on the attribute tags, bind the structural model of each three-dimensional mesh unit with the physical properties to form an organization structure-diffusion attribute association, which is used to guide the heterogeneous modeling of the diffusion velocity field. S17. Introduce an anatomical dynamic modeling mechanism, collect abdominal image sequences of patients under multiple respiratory cycles or posture changes, use a non-rigid deformation registration method to obtain the displacement field of the tissue region at different time points, and construct a dynamic deformation mapping function. , used to represent the dynamic displacement of abdominal structures over time; S18. The peritoneal structure model containing heterogeneous attribute labels and dynamic deformation functions is used as the modeling basis for drug diffusion velocity field calculation and equipotential line search, so as to realize accurate modeling of individual patient characteristics and dynamic adaptive control support.

3. The method for coordinated regulation of intraperitoneal chemotherapy perfusion dose-diffusion based on isopotential line search according to claim 1, characterized in that, S2 specifically includes: S21. Based on the constructed abdominal cavity structure model, extract each tissue region from the three-dimensional mesh structure. and corresponding attribute tags And map the attribute labels to each mesh cell of the abdominal cavity structure model; S22. Based on the peritoneal structure model and heterogeneity labels, construct the intraperitoneal drug diffusion velocity field, which is defined as a diffusion velocity function. Specifically, the function maps the diffusion impedance parameter attached to each tissue region onto the three-dimensional mesh element of the peritoneal structure model, defines the diffusion propagation speed of the drug at each spatial location, and forms a diffusion speed function that changes continuously throughout the entire peritoneal cavity. S23. Select the initial perfusion point location in the abdominal cavity structure model, and use the initial perfusion point location as the source point for drug diffusion, based on the constructed diffusion rate function. Based on the diffusion and propagation speed of the drug in various tissue regions of the abdominal cavity, the propagation time required for the drug to travel from the source point to various spatial locations in the abdominal cavity is calculated, and the propagation time is assigned to the corresponding spatial coordinate points to obtain a distribution map of drug diffusion and propagation time in the abdominal cavity. S24. Based on the calculated peritoneal drug diffusion propagation time distribution map, spatial locations with the same propagation time or adjacent time intervals are divided into the same diffusion boundary layer. In this way, multiple continuous drug diffusion equipotential surfaces are constructed in three-dimensional space. The drug diffusion equipotential surfaces are arranged in chronological order to form a preliminary three-dimensional diffusion equipotential line map describing the diffusion range reached by the drug at different time points. The preliminary three-dimensional diffusion equipotential line map is used to characterize the dynamic change process of the drug diffusion front in the peritoneal cavity as time progresses. S25. Combining the anatomical dynamic modeling mechanism introduced in the abdominal cavity structure model, based on the displacement information of the abdominal tissue structure at different times under the influence of the patient's breathing, body position changes or abdominal pressure fluctuations, the previously generated propagation time distribution map is dynamically adjusted, and the spatial position is mapped and corrected according to the tissue deformation at each time point, generating a propagation time mapping map that changes continuously with time, reflecting the actual diffusion path and boundary change process of the drug under the influence of the dynamic changes in the internal structure of the abdominal cavity. S26. Based on the dynamically corrected propagation time map, a final three-dimensional diffusion equipotential line map is generated in the three-dimensional peritoneal space. The three-dimensional diffusion equipotential line map is used to identify uncovered areas, abrupt changes in diffusion rate, and potential perfusion blind spots in the drug diffusion process, and serves as the basic data for equipotential line search to guide the selection of candidate location sets and spatial optimization of perfusion strategy parameters.

4. The method for coordinated regulation of intraperitoneal chemotherapy perfusion dose-diffusion based on isopotential line search according to claim 1, characterized in that, S3 specifically includes: S31, The final three-dimensional diffusion equipotential map will be constructed. As the input basis, spatial gradient analysis is performed on the three-dimensional diffusion equipotential map to identify gradient abrupt regions in the equipotential line distribution, i.e., spatial locations where the propagation time interval between adjacent equipotential surfaces is significantly shortened or expanded, and these are marked as potential diffusion unevenness regions. S32. Based on the time propagation information in the three-dimensional diffusion equipotential line diagram Targeted treatment area in the peritoneum Traverse the spatial points within the area and count the arrival time of drug transmission. Does it exceed the clinically defined threshold for drug efficacy coverage time? ,like If so, the corresponding spatial point will be marked as an uncovered area; S33. In uncovered and diffuse mutation regions, combine the boundary information of each tissue region in the peritoneal structural model. and organizational areas Spatial points where drug infusion can be safely performed were selected, forming a preliminary candidate point set. ; S34. For each point in the preliminary candidate point set Calculate the Euclidean distance to the existing injection point location. Points that overlap with or are redundant in close proximity to existing infusion paths are screened out, and points with the potential for independent infusion paths are retained to form an optimized candidate point set. S35. In optimizing the candidate point set, combine the attribute label corresponding to each candidate point. The diffusion adaptability and physiological tolerance of the candidate sites' regions are assessed, and candidate sites with abnormally high diffusion resistance or low drug tolerance are eliminated. Finally, a set of candidate sites after multi-target screening is output. , as candidate inputs for infusion location in the optimization variables.

5. The method for coordinated regulation of intraperitoneal chemotherapy perfusion dose-diffusion based on isopotential line search according to claim 1, characterized in that, S4 specifically includes: S41. Obtain the candidate location set As the injection point selection space, construct the optimization variable solution vector. ,in, Indicates the first Location of each injection point Indicates the perfusion dose. Indicates the infusion time. Indicates the injection sequence number; S42. Set the fitness function This function is used to comprehensively evaluate the medical feasibility and diffusion efficiency of the optimized solution. It includes the following five indicators: drug efficacy coverage of the targeted treatment area. Structural stability factor of infusion strategy Cumulative exposure dose to non-target tissues Volume ratio of diffusion blind zone Infusion path spatial redundancy factor : ； in These are non-negative weighting coefficients; S43. Based on the diffusion gradient distribution and attribute labels of candidate points in the 3D diffusion equipotential line diagram. The solution space is partitioned into layers to construct high-confidence regions. and the trusted region The main explosion point is from Selected from, the mutated spark allows from Produced in; S44. Initialize several individual fireworks centers. Each initial spark individual is generated. The number of sparks generated and the initial search radius are set according to the fitness value distribution. The higher the fitness value, the more sparks are generated and the smaller the radius. S45. Introduce an adaptive spark number and explosion radius control mechanism to calculate the optimal fitness change rate in the current iteration. and population fitness standard deviation The number of sparks at the center of each firework is adjusted according to a preset function. With search radius Real-time updates are performed to achieve a balance between local search intensity and global escape capability; S46. Based on the updated spark count and radius, perturbation is performed around each firework center to generate individual sparks, forming a local search solution group. The range of all variable perturbations is limited by the clinical safety limits of drug dosage and time. S47. Introduce a guided mutation mechanism from the trusted region. In this process, a starting point is selected, and a certain number of guiding mutation sparks are generated according to the direction of the optimal solution to enhance the guidance and local development capabilities of the search. S48. Calculate the fitness function for all solution vectors. Based on the multi-objective comprehensive score, adopt an update strategy based on spatial distribution distance control to select the solution with the best fitness and large spatial distribution difference from the firework center, ordinary sparks and guiding mutated sparks to enter the next generation. S49. Determine whether the maximum number of iterations or the fitness convergence threshold has been reached. If the conditions are not met, return to step S45 and continue the optimization iteration. If the conditions are met, stop the search process. S410. Output the optimal solution vector as the optimized perfusion strategy, which is used to generate perfusion control commands to control the intraperitoneal chemotherapy perfusion system to perform drug injection operations according to the optimization results.

6. The method for coordinated regulation of intraperitoneal chemotherapy perfusion dose-diffusion based on isopotential line search according to claim 1, characterized in that, S5 specifically includes: S51. The spatial location of the infusion points, the corresponding drug dosage, the infusion time and the infusion sequence information in the optimized infusion plan are analyzed in a structured manner to generate an executable infusion parameter set. S52. Map and convert the injection parameter set into an instruction format that is compatible with the injection equipment, including the flow control value, opening time, channel number and valve action sequence required by each injection module. S53. Establish a mapping relationship between perfusion points and perfusion channels, allocate hardware execution channels according to the spatial order of perfusion locations, ensure synchronous switching of channels in the intraperitoneal chemotherapy perfusion system, and avoid perfusion interference or drug flow conflicts. S54. Based on the infusion sequence and time period arrangement given in the optimized infusion scheme, generate a complete timing control flowchart and simultaneously configure the corresponding timing triggering mechanism and execution status feedback interface. S55. Upload the control command to the intraperitoneal chemotherapy perfusion system, and complete the initialization and parameter writing of the hardware components of the drug injection pump, solenoid valve and channel selector through the control platform; S56. Start the perfusion control process, control the perfusion equipment in real time to inject drugs sequentially according to the parameter combination set in the optimized scheme, and continuously collect execution status information.

7. The method for optimizing musical instrument playing techniques based on biosignal feedback according to claim 6, characterized in that, The instruction format for conversion to adapt to the perfusion equipment specifically includes the flow control value, opening time, channel number and valve action sequence required by each perfusion module. This is used to drive each execution unit of the intraperitoneal chemotherapy perfusion system to precisely control the injection position, time and dosage of the drug according to the optimized scheme, so as to realize personalized and precise perfusion operation.

8. The method for coordinated regulation of intraperitoneal chemotherapy perfusion dose-diffusion based on isopotential line search according to claim 1, characterized in that, S6 specifically includes: S61. During chemotherapy, real-time feedback data on drug distribution, concentration changes, and tissue response in the peritoneal cavity are obtained through image acquisition modules, sensor nodes, or drug tracing devices. S62. The collected feedback data is formatted and spatiotemporally matched, and then registered with the previously constructed abdominal cavity structure model. S63. Based on the registered feedback data, re-execute the drug propagation analysis process, and combine it with the tissue displacement to correct the existing drug propagation time map and dynamically reconstruct the three-dimensional diffusion equipotential line map at the current moment. S64. Based on the reconstructed three-dimensional diffusion equipotential line map, analyze the spatial matching degree between the drug efficacy coverage area and the target area, and determine whether there are any uncovered areas, abnormal drug accumulation areas, or distribution shifts. S65. Compare the current drug efficacy coverage with the preset indicator threshold. If the drug efficacy coverage is insufficient, the blind area is too large, or it deviates from the warning indicator, start the loop control of the optimization process and re-execute the steps of equipotential line search, candidate point update, optimization calculation and control instruction generation. S66. Replace the current control scheme with the updated perfusion strategy and continue to execute the drug injection and feedback acquisition process to form a dynamic closed-loop coordinated adjustment mechanism between drug diffusion, strategy optimization and execution feedback.

Images