Tympanic membrane repair postoperative wound assessment method based on artificial intelligence

Abstract

The invention discloses a tympanic membrane repair postoperative wound assessment method based on artificial intelligence, and relates to the technical field of medical assessment, and the method comprises the steps: obtaining multi-modal image data of a reference tympanic membrane and a target tympanic membrane through a multi-modal dynamic imaging fusion technology, determining a target biomechanical feature ratio through a biomechanical feature fusion technology, and obtaining a target tympanic membrane target biomechanical feature ratio; the method further comprises the steps of obtaining a physiological parameter set of the reference tympanic membrane and the target tympanic membrane, training an associated parameter prediction model to construct a metabolism-mechanical associated model, and achieving accurate evaluation of tympanic membrane wound healing conditions. The invention also covers related electronic equipment, a computer readable storage medium and a computer program product.

Description

Technical Field

[0001] The present invention relates to the field of medical assessment technology, and more specifically, to an artificial intelligence-based method for assessing wounds after tympanic membrane repair. Background Art

[0002] Tympanoplasty is an important treatment for ear diseases such as tympanic membrane perforation. After surgery, accurate assessment of tympanic membrane wound healing is crucial, as it directly impacts the patient's hearing recovery and postoperative rehabilitation process. Traditional methods for evaluating tympanic membrane healing after surgery rely primarily on visual observation and basic audiometric testing. For example, doctors use otoscopy to observe changes in the tympanic membrane's morphology and color, combined with pure-tone audiometry to assess the healing status of the tympanic membrane. However, these traditional methods have numerous limitations. Visual observation can only capture macroscopic information about the tympanic membrane surface, but cannot provide in-depth and accurate assessment of the tympanic membrane's microstructure, internal biomechanical properties, and metabolic state. For example, it is difficult to discern key information such as the growth of tympanic membrane healing tissue at different levels and the density of microvascular regeneration. While audiometric testing can reflect the recovery of hearing function, it lacks direct and detailed quantitative indicators for the specific degree and quality of wound healing. Differences in experience among different doctors can also lead to highly subjective assessment results.

[0003] With the continuous development of medical imaging technology and artificial intelligence, people have higher expectations for using advanced technology to improve the accuracy, objectivity and comprehensiveness of wound assessment after tympanic membrane repair. Multimodal imaging technology can obtain rich information about the tympanic membrane in different dimensions. For example, polarization-sensitive OCT imaging can clearly present the layered structure of the tympanic membrane, and laser speckle blood flow imaging can reflect the density of microvessels. These technologies provide new methods for in-depth exploration of the healing mechanism of tympanic membrane wounds. At the same time, biomechanical characteristic analysis can reveal important mechanical properties of the tympanic membrane, such as tension distribution and elastic modulus. These properties are closely related to the functional recovery of the tympanic membrane. In addition, the microenvironment metabolic assessment model helps to understand the metabolic changes in the wound healing process, and artificial intelligence algorithms can fuse, analyze and mine massive amounts of multimodal data to achieve more accurate wound assessment.

[0004] Therefore, existing methods for wound assessment after tympanic membrane repair surgery are difficult to accurately quantify the tympanic membrane microstructure, biomechanical properties, and metabolic state, and have problems such as strong subjectivity and poor continuity. Summary of the Invention

[0005] In order to overcome the problems of existing wound assessment methods, such as the difficulty in quantification and subjectivity of existing tympanic membrane assessment, the present invention discloses an artificial intelligence-based wound assessment method for tympanic membrane repair surgery, which can effectively solve the above technical problems.

[0006] In order to solve the above technical problems, the technical solutions of the present invention are as follows:

[0007] A post-tympanoplasty wound assessment method based on artificial intelligence comprises the following steps:

[0008] Using multimodal dynamic imaging fusion technology, first multimodal image data of a reference tympanic membrane at a target evaluation time and second multimodal image data of a target tympanic membrane to be evaluated at a target evaluation time are acquired, wherein the target tympanic membrane is a post-repair tympanic membrane, and the reference tympanic membrane is a healthy or well-healed tympanic membrane; the first multimodal image data includes an image of the tympanic membrane layer structure acquired by polarization-sensitive optical coherence tomography (OCT) imaging and an image of microvessel density acquired by laser speckle blood flow imaging, and the second multimodal image data corresponds to the acquisition of the aforementioned two modal images;

[0009] determining, based on the first multimodal image data and the second multimodal image data, a target biomechanical characteristic ratio between a reference tympanic membrane and a target tympanic membrane at a target assessment time using a biomechanical characteristic fusion technique; the biomechanical characteristics comprising tympanic membrane tension distribution calculated using curvature modal analysis and healing tissue elastic modulus inferred using acoustic impedance inversion;

[0010] Based on the target biomechanical characteristic ratio, the first healing standard of the reference tympanic membrane, the second healing expectation of the target tympanic membrane, the postoperative duration of the target tympanic membrane at the target evaluation time, and the microenvironment metabolic evaluation model, the wound healing evaluation result of the target tympanic membrane relative to the reference tympanic membrane at the target evaluation time is obtained; the microenvironment metabolic evaluation model is determined based on the same relationship between the metabolite index ratio and the biomechanical characteristic ratio between the reference tympanic membrane and the target tympanic membrane under normal healing and current repair states.

[0011] Preferably, obtaining the wound healing assessment result of the target tympanic membrane relative to the reference tympanic membrane at the target assessment time includes:

[0012] The original microenvironment metabolic assessment model is updated according to the model parameter values of the metabolic-mechanical correlation model constructed during the current assessment cycle to obtain a target microenvironment metabolic assessment model; wherein the metabolic-mechanical correlation model is a coupled model of the metabolite concentrations and biomechanical indicators of the reference and target tympanic membranes over time;

[0013] The target evaluation time, the first healing standard of the reference tympanic membrane, the postoperative duration of the target tympanic membrane at the target evaluation time, and the second healing expectation are input into the target microenvironment metabolic evaluation model to obtain the wound healing evaluation result of the target tympanic membrane relative to the reference tympanic membrane at the target evaluation time.

[0014] Preferably, the method further comprises:

[0015] Acquire a first physiological parameter set of a reference tympanic membrane and a second physiological parameter set of a target tympanic membrane to be evaluated; wherein the first physiological parameter set includes first sample multimodal image data, first sample biomechanical index values, and first sample metabolite concentration values of the reference tympanic membrane at multiple historical evaluation times; and the second physiological parameter set includes second sample multimodal image data, second sample biomechanical index values, and second sample metabolite concentration values of the target tympanic membrane at multiple historical evaluation times;

[0016] Based on the first-sample multimodal image data at each historical evaluation time, the first-sample tympanic membrane tension distribution of the reference tympanic membrane at each historical evaluation time is determined through curvature modal analysis and point cloud reconstruction, and the first-sample healing tissue elastic modulus of the reference tympanic membrane at each historical evaluation time is determined through acoustic impedance inversion combined with pure tone audiometry data; based on the second-sample multimodal image data at each historical evaluation time, the second-sample tympanic membrane tension distribution and the second-sample healing tissue elastic modulus of the target tympanic membrane at each historical evaluation time are correspondingly determined;

[0017] The metabolite concentration value of the first sample at each historical evaluation time was obtained based on metabolite spectral analysis, and the pH value and lactic acid concentration of the wound surface were detected using a Raman probe. The collagen arrangement direction was identified through SHG imaging, and the metabolite concentration value of the second sample at each historical evaluation time was obtained.

[0018] Based on the first sample biomechanical index value, the first sample metabolite concentration value, the second sample biomechanical index value and the second sample metabolite concentration value at each historical evaluation time, the association parameter prediction model is trained to obtain the target association parameter value, and the target association parameter value is introduced into the coupling model to construct the metabolism-mechanics association model.

[0019] Preferably, obtaining the target associated parameter value includes:

[0020] Determine the first sample biomechanical index value, the first sample metabolite concentration value, the second sample biomechanical index value, and the second sample metabolite concentration value at each historical evaluation time as sample physiological index values, respectively, to obtain a sample physiological index value set;

[0021] For each sample physiological indicator value in the sample physiological indicator value set, input it into the correlation parameter prediction model to be trained to obtain a predicted correlation parameter value;

[0022] Determining a predicted value of a sample physiological indicator based on the predicted associated parameter value and the coupling model;

[0023] Inputting the sample physiological index value and the sample physiological index predicted value into a loss function to obtain a loss function value;

[0024] If the loss function value is less than or equal to the loss function threshold, the predicted associated parameter value is determined as the target associated parameter value.

[0025] Preferably, the method further comprises:

[0026] In a next evaluation cycle, obtaining a first updated set of physiological parameters of the reference tympanic membrane and a second updated set of physiological parameters of the target tympanic membrane;

[0027] Based on the first updated physiological parameter set and the second updated physiological parameter set, an updated metabolic-mechanical correlation model is determined.

[0028] Preferably, determining the target biomechanical characteristic ratio between the reference tympanic membrane and the target tympanic membrane at the target evaluation time comprises:

[0029] Based on the first multimodal image data, a curvature modal analysis of the tympanic membrane layered structure image is performed using a finite element model to calculate the first tympanic membrane tension distribution of the reference tympanic membrane at the target evaluation time. The first healing tissue elastic modulus of the reference tympanic membrane at the target evaluation time is inferred through acoustic impedance inversion combined with pure tone audiometry data.

[0030] calculating a second tympanic membrane tension distribution and a second healing tissue elastic modulus of the target tympanic membrane at a target evaluation time based on the second multimodal image data;

[0031] The ratio of the first tympanic membrane tension distribution to the second tympanic membrane tension distribution and the ratio of the first callus elastic modulus to the second callus elastic modulus are weightedly fused to determine a target biomechanical characteristic ratio between the reference tympanic membrane and the target tympanic membrane at a target evaluation time.

[0032] Preferably, calculating the first tympanic membrane tension distribution of the reference tympanic membrane at the target evaluation time includes:

[0033] The tympanic membrane hierarchical structure image obtained by polarization-sensitive OCT imaging of the reference tympanic membrane was reconstructed into a three-dimensional point cloud to construct a tympanic membrane finite element model;

[0034] Based on the finite element model and combined with the physiological boundary conditions of the tympanic membrane, the first tympanic membrane tension distribution of the reference tympanic membrane at the target evaluation time is calculated using a curvature modal analysis algorithm.

[0035] An electronic device includes a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the above-mentioned evaluation method.

[0036] A computer-readable storage medium stores a computer program thereon, wherein the computer program implements the above-mentioned evaluation method when executed by a processor.

[0037] A computer program product includes a computer program, wherein the computer program implements the above-mentioned evaluation method when executed by a processor.

[0038] Compared with the existing technology, the beneficial effects of the present invention are: the present invention can accurately quantify the microstructure, biomechanical properties and metabolic state of the tympanic membrane by adopting multimodal dynamic imaging fusion technology, which solves the problem that the existing wound assessment methods after tympanic membrane repair are difficult to accurately quantify. First, the tympanic membrane layered structure image obtained by polarization-sensitive OCT imaging and the microvascular density image obtained by laser speckle blood flow imaging can clearly show the microstructure and blood flow of the tympanic membrane. Compared with traditional assessment methods that can only be observed by the naked eye or simple imaging, the accuracy of microstructure assessment is improved, and more detailed basic data is provided for wound healing assessment. Second, the target biomechanical feature ratio is determined by using biomechanical feature fusion technology, in which the tympanic membrane tension distribution is calculated by curvature modal analysis and the elastic modulus of the healing tissue is inferred by acoustic impedance inversion. Compared with the traditional subjective assessment of biomechanical properties, Method: This technology can present the mechanical properties of the tympanic membrane in a numerical form, avoiding the subjectivity of human judgment and making the biomechanical property evaluation more objective and accurate. Third, based on the microenvironment metabolic assessment model, it is combined with the biomechanical characteristic ratio to integrate the metabolic and mechanical factors of the tympanic membrane wound. Compared with the existing assessment method that only focuses on a single dimension, it realizes a multi-dimensional and comprehensive assessment, which is more in line with the actual complex physiological process of tympanic membrane wound healing, thereby improving the comprehensiveness and accuracy of wound healing assessment. Fourth, by constructing a metabolic-mechanical correlation model and updating it based on historical assessment data, continuous monitoring and prediction of the tympanic membrane wound status are achieved, overcoming the disadvantage of poor continuity of existing assessment methods, and providing clinicians with dynamic and continuous wound healing information, which helps to adjust the treatment plan in time and improve the treatment effect and patient prognosis after tympanic membrane repair. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the implementation methods of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the implementation methods or the description of the prior art. Obviously, the drawings described below are merely exemplary. For ordinary technicians in this field, other implementation drawings can be derived based on the provided drawings without any creative work.

[0040] Figure 1 This figure shows the steps of an artificial intelligence-based method for wound assessment after tympanoplasty. DETAILED DESCRIPTION

[0041] The accompanying drawings are for illustrative purposes only and are not to be construed as limiting this patent;

[0042] In order to better illustrate this embodiment, some parts in the drawings may be omitted, enlarged, or reduced, and do not represent the actual product size;

[0043] It is understandable to those skilled in the art that some well-known structures and descriptions thereof may be omitted in the drawings.

[0044] The technical solution of the present invention is further described below with reference to the accompanying drawings and embodiments.

[0045] Example

[0046] A post-tympanoplasty wound assessment method based on artificial intelligence comprises the following steps:

[0047] Using multimodal dynamic imaging fusion technology, first multimodal image data of a reference tympanic membrane at a target evaluation time and second multimodal image data of a target tympanic membrane to be evaluated at a target evaluation time are acquired, wherein the target tympanic membrane is a post-repair tympanic membrane, and the reference tympanic membrane is a healthy or well-healed tympanic membrane; the first multimodal image data includes an image of the tympanic membrane layer structure acquired by polarization-sensitive optical coherence tomography (OCT) imaging and an image of microvessel density acquired by laser speckle blood flow imaging, and the second multimodal image data corresponds to the acquisition of the aforementioned two modal images;

[0048] determining, based on the first multimodal image data and the second multimodal image data, a target biomechanical characteristic ratio between a reference tympanic membrane and a target tympanic membrane at a target assessment time using a biomechanical characteristic fusion technique; the biomechanical characteristics comprising tympanic membrane tension distribution calculated using curvature modal analysis and healing tissue elastic modulus inferred using acoustic impedance inversion;

[0049] Based on the target biomechanical characteristic ratio, the first healing standard of the reference tympanic membrane, the second healing expectation of the target tympanic membrane, the postoperative duration of the target tympanic membrane at the target evaluation time, and the microenvironment metabolic evaluation model, the wound healing evaluation result of the target tympanic membrane relative to the reference tympanic membrane at the target evaluation time is obtained; the microenvironment metabolic evaluation model is determined based on the same relationship between the metabolite index ratio and the biomechanical characteristic ratio between the reference tympanic membrane and the target tympanic membrane under normal healing and current repair states.

[0050] Obtaining the wound healing assessment result of the target tympanic membrane relative to the reference tympanic membrane at the target assessment time includes:

[0051] The original microenvironment metabolic assessment model is updated according to the model parameter values of the metabolic-mechanical correlation model constructed during the current assessment cycle to obtain a target microenvironment metabolic assessment model; wherein the metabolic-mechanical correlation model is a coupled model of the metabolite concentrations and biomechanical indicators of the reference and target tympanic membranes over time;

[0052] The target evaluation time, the first healing standard of the reference tympanic membrane, the postoperative duration of the target tympanic membrane at the target evaluation time, and the second healing expectation are input into the target microenvironment metabolic evaluation model to obtain the wound healing evaluation result of the target tympanic membrane relative to the reference tympanic membrane at the target evaluation time.

[0053] The method further comprises:

[0054] Acquire a first physiological parameter set of a reference tympanic membrane and a second physiological parameter set of a target tympanic membrane to be evaluated; wherein the first physiological parameter set includes first sample multimodal image data, first sample biomechanical index values, and first sample metabolite concentration values of the reference tympanic membrane at multiple historical evaluation times; and the second physiological parameter set includes second sample multimodal image data, second sample biomechanical index values, and second sample metabolite concentration values of the target tympanic membrane at multiple historical evaluation times;

[0055] Based on the first-sample multimodal image data at each historical evaluation time, the first-sample tympanic membrane tension distribution of the reference tympanic membrane at each historical evaluation time is determined through curvature modal analysis and point cloud reconstruction, and the first-sample healing tissue elastic modulus of the reference tympanic membrane at each historical evaluation time is determined through acoustic impedance inversion combined with pure tone audiometry data; based on the second-sample multimodal image data at each historical evaluation time, the second-sample tympanic membrane tension distribution and the second-sample healing tissue elastic modulus of the target tympanic membrane at each historical evaluation time are correspondingly determined;

[0056] The metabolite concentration value of the first sample at each historical evaluation time was obtained based on metabolite spectral analysis, and the pH value and lactic acid concentration of the wound surface were detected using a Raman probe. The collagen arrangement direction was identified through SHG imaging, and the metabolite concentration value of the second sample at each historical evaluation time was obtained.

[0057] Based on the first sample biomechanical index value, the first sample metabolite concentration value, the second sample biomechanical index value and the second sample metabolite concentration value at each historical evaluation time, the association parameter prediction model is trained to obtain the target association parameter value, and the target association parameter value is introduced into the coupling model to construct the metabolism-mechanics association model.

[0058] Obtaining the target associated parameter value includes:

[0059] Determine the first sample biomechanical index value, the first sample metabolite concentration value, the second sample biomechanical index value, and the second sample metabolite concentration value at each historical evaluation time as sample physiological index values, respectively, to obtain a sample physiological index value set;

[0060] For each sample physiological indicator value in the sample physiological indicator value set, input it into the correlation parameter prediction model to be trained to obtain a predicted correlation parameter value;

[0061] Determining a predicted value of a sample physiological indicator based on the predicted associated parameter value and the coupling model;

[0062] Inputting the sample physiological index value and the sample physiological index predicted value into a loss function to obtain a loss function value;

[0063] If the loss function value is less than or equal to the loss function threshold, the predicted associated parameter value is determined as the target associated parameter value.

[0064] The method further comprises:

[0065] In a next evaluation cycle, obtaining a first updated set of physiological parameters of the reference tympanic membrane and a second updated set of physiological parameters of the target tympanic membrane;

[0066] Based on the first updated physiological parameter set and the second updated physiological parameter set, an updated metabolic-mechanical correlation model is determined.

[0067] Determining the target biomechanical characteristic ratio between the reference tympanic membrane and the target tympanic membrane at the target evaluation time includes:

[0068] Based on the first multimodal image data, a curvature modal analysis of the tympanic membrane layered structure image is performed using a finite element model to calculate the first tympanic membrane tension distribution of the reference tympanic membrane at the target evaluation time. The first healing tissue elastic modulus of the reference tympanic membrane at the target evaluation time is inferred through acoustic impedance inversion combined with pure tone audiometry data.

[0069] calculating a second tympanic membrane tension distribution and a second healing tissue elastic modulus of the target tympanic membrane at a target evaluation time based on the second multimodal image data;

[0070] The ratio of the first tympanic membrane tension distribution to the second tympanic membrane tension distribution and the ratio of the first callus elastic modulus to the second callus elastic modulus are weightedly fused to determine a target biomechanical characteristic ratio between the reference tympanic membrane and the target tympanic membrane at a target evaluation time.

[0071] The calculating of the first tympanic membrane tension distribution of the reference tympanic membrane at the target evaluation time comprises:

[0072] The tympanic membrane hierarchical structure image obtained by polarization-sensitive OCT imaging of the reference tympanic membrane was reconstructed into a three-dimensional point cloud to construct a tympanic membrane finite element model;

[0073] Based on the finite element model and combined with the physiological boundary conditions of the tympanic membrane, the first tympanic membrane tension distribution of the reference tympanic membrane at the target evaluation time is calculated using a curvature modal analysis algorithm.

[0074] An electronic device includes a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the above-mentioned evaluation method.

[0075] A computer-readable storage medium stores a computer program thereon, wherein the computer program implements the above-mentioned evaluation method when executed by a processor.

[0076] A computer program product includes a computer program, wherein the computer program implements the above-mentioned evaluation method when executed by a processor.

[0077] In the specific implementation, please refer to Figure 1 In order to accurately evaluate the healing of the tympanic membrane wound of a patient who had undergone tympanic membrane repair surgery, this artificial intelligence-based evaluation method was used to provide a basis for adjusting the subsequent treatment plan.

[0078] Equipped with polarization-sensitive OCT imaging equipment, laser speckle blood flow imaging equipment, Raman probe, SHG imaging equipment and pure tone audiometry equipment.

[0079] The selected reference tympanic membrane comes from a clinically verified healthy and well-healed tympanic membrane sample database. This database has been established in advance and covers multimodal image data and related physiological parameters of healthy tympanic membranes of different age groups and genders.

[0080] At the target evaluation time point (14 days after surgery), polarization-sensitive OCT imaging was performed on the reference tympanic membrane to obtain its layered structural image, which can clearly show the thickness and morphological information of each layer such as the tympanic membrane epithelium and fibrous layer. At the same time, laser speckle blood flow imaging was performed to obtain a microvascular density image, which shows the microvascular distribution and blood perfusion conditions in the tympanic membrane area. The above constitutes the first multimodal image data.

[0081] The target tympanic membrane to be evaluated (the patient's postoperative tympanic membrane) is simultaneously imaged using the same modality to obtain the corresponding second multimodal image data. The parameter settings, imaging angles, and positions are controlled throughout the imaging process to ensure that the data are accurate and comparable.

[0082] A finite element model was used to reconstruct a three-dimensional point cloud of the hierarchical structure image of the reference tympanic membrane obtained by polarization-sensitive OCT imaging. A finite element model of the tympanic membrane was constructed. Combined with the physiological boundary conditions of the tympanic membrane, a curvature modal analysis algorithm was used to calculate the first tympanic membrane tension distribution of the reference tympanic membrane at the target evaluation time, and the size and change trend of the tension in different areas of the tympanic membrane were presented in a digital quantitative form.

[0083] The first healing tissue elastic modulus of the reference tympanic membrane at the target assessment time was calculated using acoustic impedance inversion combined with pure tone audiometry data, reflecting the elasticity and mechanical support capacity of the tympanic membrane tissue.

[0084] According to the same operation process, based on the second multimodal image data of the target tympanic membrane, a second tympanic membrane tension distribution and a second healing tissue elastic modulus of the target tympanic membrane at the target evaluation time are calculated.

[0085] The ratio of the first tympanic membrane tension distribution to the second tympanic membrane tension distribution, and the ratio of the first healing tissue elastic modulus to the second healing tissue elastic modulus are weightedly fused according to preset weights (the weights can be determined in advance based on clinical experience and data verification, for example, a tension distribution ratio weight of 0.6 and an elastic modulus ratio weight of 0.4), and finally the target biomechanical characteristic ratio of the reference tympanic membrane and the target tympanic membrane at the target evaluation time is determined.

[0086] The hospital has built a microenvironment metabolic assessment model in advance. It is based on the relationship between the ratio of metabolite indicators and the ratio of biomechanical characteristics between the reference tympanic membrane and the target tympanic membrane under normal healing and current repair conditions. During the model construction process, a large amount of similar data accumulated in the early clinical stage has been input for training and optimization, so that it can accurately reflect the metabolic and mechanical coupling laws in the tympanic membrane healing process.

[0087] The target biomechanical characteristic ratios determined above, the first healing standard of the reference tympanic membrane, including the biomechanical characteristic thresholds and metabolite concentration ranges of the healthy tympanic membrane at different stages, the second healing expectation of the target tympanic membrane (the postoperative healing target estimated based on the patient's individual conditions, such as age and basic health status, expressed in the form of biomechanical and metabolite indicators), and the postoperative duration of the target tympanic membrane at the target assessment time (14 days) are now substituted into the microenvironment metabolic assessment model. After model calculation and analysis, the wound healing assessment results of the target tympanic membrane relative to the reference tympanic membrane at the target assessment time are obtained, including detailed information such as the percentage of tympanic membrane wound healing, healing trends, and healing differences in each region.

[0088] A first set of physiological parameters of the reference tympanic membrane was collected, covering the first-sample multimodal image data of the reference tympanic membrane at multiple historical evaluation times, such as the 3rd, 7th, 10th, and 14th days after surgery. These data were the same as the aforementioned polarization-sensitive OCT imaging and laser speckle blood flow imaging data, and the first-sample biomechanical index values, such as the tympanic membrane tension distribution and healing tissue elastic modulus obtained through curvature modal analysis and acoustic impedance inversion at each historical time point. The first-sample metabolite concentration values were obtained through metabolite spectral analysis, and the pH value and lactic acid concentration of the wound surface were detected by Raman probe for auxiliary calibration. SHG (second harmonic generation) imaging was used to identify the collagen arrangement direction, indirectly determine the distribution direction of metabolites in the tympanic membrane, and comprehensively determine the metabolite concentration.

[0089] The second physiological parameter set of the target tympanic membrane is collected synchronously, and the second sample multimodal image data, second sample biomechanical index values, and second sample metabolite concentration values of the target tympanic membrane at multiple historical evaluation times are recorded in detail. The acquisition method is consistent with that of the reference tympanic membrane to ensure the homology and comparability of the data.

[0090] Based on the first-sample multimodal image data at each historical evaluation time, curvature modal analysis and point cloud reconstruction technology are used to determine the first-sample tympanic membrane tension distribution of the reference tympanic membrane at each historical evaluation time, and the first-sample healing tissue elastic modulus of the reference tympanic membrane at each historical evaluation time is determined by acoustic impedance inversion combined with pure tone audiometry data; correspondingly, based on the second-sample multimodal image data of the target tympanic membrane, the second-sample tympanic membrane tension distribution and the second-sample healing tissue elastic modulus of the target tympanic membrane at each historical evaluation time are obtained.

[0091] Metabolite spectral analysis was used in conjunction with Raman probes to detect wound pH and lactic acid concentration, and the first sample metabolite concentration value at each historical assessment time was accurately obtained. Similarly, SHG imaging was used to identify the collagen arrangement direction and correlate and deduce the second sample metabolite concentration value at each historical assessment time.

[0092] The first sample biomechanical index value, the first sample metabolite concentration value, the second sample biomechanical index value, and the second sample metabolite concentration value at each historical evaluation time are summarized to form a sample physiological index value set, which is then input into the association parameter prediction model to be trained. The model uses a deep learning algorithm, such as a convolutional neural network, and after data training, can capture the intrinsic correlation between biomechanical and metabolite indicators.

[0093] For each sample physiological indicator value in the sample physiological indicator value set, it is input into the trained correlation parameter prediction model to obtain the predicted correlation parameter value. Based on this predicted correlation parameter value and the preset coupling model, the functional relationship between the metabolite concentration and the biomechanical index changing over time is described to determine the predicted value of the sample physiological indicator.

[0094] The actual and predicted values of the sample physiological indicators are substituted into the loss function, such as the mean square error loss function, and the loss function value is calculated. If the loss function value is less than or equal to the pre-set loss function threshold, which is set according to the model accuracy requirements and clinical error tolerance, such as 0.05, then the predicted correlation parameter value is determined to be reliable at this time, and it is determined as the target correlation parameter value and introduced into the coupling model to finally complete the construction of the metabolic-mechanical correlation model.

[0095] In the next evaluation cycle (evaluation on the 21st day after surgery), the first updated set of physiological parameters of the reference tympanic membrane was obtained again, including the sample multimodal image data, biomechanical index values, and metabolite concentration values at the new time point; the second updated set of physiological parameters of the target tympanic membrane was collected simultaneously. Based on these updated data, the model was retrained and calibrated according to the same process and method of constructing the metabolic-mechanical correlation model in the early stage, and the updated metabolic-mechanical correlation model was determined, so that it can more accurately adapt to the dynamic changes of tympanic membrane healing and provide a more reliable basis for subsequent longer-term wound assessment.

[0096] The same or similar reference numerals correspond to the same or similar components;

[0097] The terms used in the drawings to describe positional relationships are for illustrative purposes only and should not be construed as limiting this patent;

[0098] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not limitations on the implementation methods of the present invention. For ordinary technicians in the field, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to enumerate all the implementation methods here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the claims of the present invention.

Claims

1. A method for evaluating wound surface after tympanic membrane repair surgery based on artificial intelligence, characterized in that: The following steps are involved: Using multimodal dynamic imaging fusion technology, first multimodal image data of a reference tympanic membrane at a target evaluation time and second multimodal image data of a target tympanic membrane to be evaluated at a target evaluation time are acquired, wherein the target tympanic membrane is a post-repair tympanic membrane, and the reference tympanic membrane is a healthy or well-healed tympanic membrane; the first multimodal image data includes an image of the tympanic membrane layer structure acquired by polarization-sensitive optical coherence tomography (OCT) imaging and an image of microvessel density acquired by laser speckle blood flow imaging, and the second multimodal image data corresponds to the acquisition of the aforementioned two modal images; determining, based on the first multimodal image data and the second multimodal image data, a target biomechanical characteristic ratio between a reference tympanic membrane and a target tympanic membrane at a target assessment time using a biomechanical characteristic fusion technique; the biomechanical characteristics comprising tympanic membrane tension distribution calculated using curvature modal analysis and healing tissue elastic modulus inferred using acoustic impedance inversion; Based on the target biomechanical characteristic ratio, the first healing standard of the reference tympanic membrane, the second healing expectation of the target tympanic membrane, the postoperative duration of the target tympanic membrane at the target evaluation time, and the microenvironment metabolic evaluation model, the wound healing evaluation result of the target tympanic membrane relative to the reference tympanic membrane at the target evaluation time is obtained; the microenvironment metabolic evaluation model is determined based on the same relationship between the metabolite index ratio and the biomechanical characteristic ratio between the reference tympanic membrane and the target tympanic membrane under normal healing and current repair states.

2. The evaluation method according to claim 1, wherein: Obtaining the wound healing assessment result of the target tympanic membrane relative to the reference tympanic membrane at the target assessment time includes: The original microenvironment metabolic assessment model is updated according to the model parameter values of the metabolic-mechanical correlation model constructed during the current assessment cycle to obtain a target microenvironment metabolic assessment model; wherein the metabolic-mechanical correlation model is a coupled model of the metabolite concentrations and biomechanical indicators of the reference and target tympanic membranes over time; The target evaluation time, the first healing standard of the reference tympanic membrane, the postoperative duration of the target tympanic membrane at the target evaluation time, and the second healing expectation are input into the target microenvironment metabolic evaluation model to obtain the wound healing evaluation result of the target tympanic membrane relative to the reference tympanic membrane at the target evaluation time.

3. The evaluation method according to claim 2, wherein: The method further comprises: Acquire a first physiological parameter set of a reference tympanic membrane and a second physiological parameter set of a target tympanic membrane to be evaluated; wherein the first physiological parameter set includes first sample multimodal image data, first sample biomechanical index values, and first sample metabolite concentration values of the reference tympanic membrane at multiple historical evaluation times; and the second physiological parameter set includes second sample multimodal image data, second sample biomechanical index values, and second sample metabolite concentration values of the target tympanic membrane at multiple historical evaluation times; Based on the first-sample multimodal image data at each historical evaluation time, the first-sample tympanic membrane tension distribution of the reference tympanic membrane at each historical evaluation time is determined through curvature modal analysis and point cloud reconstruction, and the first-sample healing tissue elastic modulus of the reference tympanic membrane at each historical evaluation time is determined through acoustic impedance inversion combined with pure tone audiometry data; based on the second-sample multimodal image data at each historical evaluation time, the second-sample tympanic membrane tension distribution and the second-sample healing tissue elastic modulus of the target tympanic membrane at each historical evaluation time are correspondingly determined; The metabolite concentration value of the first sample at each historical evaluation time was obtained based on metabolite spectral analysis, and the pH value and lactic acid concentration of the wound surface were detected using a Raman probe. The collagen arrangement direction was identified through SHG imaging, and the metabolite concentration value of the second sample at each historical evaluation time was obtained. Based on the first sample biomechanical index value, the first sample metabolite concentration value, the second sample biomechanical index value and the second sample metabolite concentration value at each historical evaluation time, the association parameter prediction model is trained to obtain the target association parameter value, and the target association parameter value is introduced into the coupling model to construct the metabolism-mechanics association model.

4. The evaluation method according to claim 3, wherein: Obtaining the target associated parameter value includes: Determine the first sample biomechanical index value, the first sample metabolite concentration value, the second sample biomechanical index value, and the second sample metabolite concentration value at each historical evaluation time as sample physiological index values, respectively, to obtain a sample physiological index value set; For each sample physiological indicator value in the sample physiological indicator value set, input it into the correlation parameter prediction model to be trained to obtain a predicted correlation parameter value; Determining a predicted value of a sample physiological indicator based on the predicted associated parameter value and the coupling model; Inputting the sample physiological index value and the sample physiological index predicted value into a loss function to obtain a loss function value; If the loss function value is less than or equal to the loss function threshold, the predicted associated parameter value is determined as the target associated parameter value.

5. The evaluation method according to claim 3, wherein: The method further comprises: In a next evaluation cycle, obtaining a first updated set of physiological parameters of the reference tympanic membrane and a second updated set of physiological parameters of the target tympanic membrane; Based on the first updated physiological parameter set and the second updated physiological parameter set, an updated metabolic-mechanical correlation model is determined.

6. The evaluation method according to claim 1, wherein: Determining the target biomechanical characteristic ratio between the reference tympanic membrane and the target tympanic membrane at the target evaluation time includes: Based on the first multimodal image data, a curvature modal analysis of the tympanic membrane layered structure image is performed using a finite element model to calculate the first tympanic membrane tension distribution of the reference tympanic membrane at the target evaluation time. The first healing tissue elastic modulus of the reference tympanic membrane at the target evaluation time is inferred through acoustic impedance inversion combined with pure tone audiometry data. calculating a second tympanic membrane tension distribution and a second healing tissue elastic modulus of the target tympanic membrane at a target evaluation time based on the second multimodal image data; The ratio of the first tympanic membrane tension distribution to the second tympanic membrane tension distribution and the ratio of the first callus elastic modulus to the second callus elastic modulus are weightedly fused to determine a target biomechanical characteristic ratio between the reference tympanic membrane and the target tympanic membrane at a target evaluation time.

7. The evaluation method according to claim 6, characterized in that The calculating of the first tympanic membrane tension distribution of the reference tympanic membrane at the target evaluation time comprises: The tympanic membrane hierarchical structure image obtained by polarization-sensitive OCT imaging of the reference tympanic membrane was reconstructed into a three-dimensional point cloud to construct a tympanic membrane finite element model; Based on the finite element model and combined with the physiological boundary conditions of the tympanic membrane, the first tympanic membrane tension distribution of the reference tympanic membrane at the target evaluation time is calculated using a curvature modal analysis algorithm.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the evaluation method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the evaluation method according to any one of claims 1 to 7 is implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the evaluation method according to any one of claims 1 to 7 is implemented.

Images