Method and system for evaluating curative effect of antibacterial drug

Abstract

The invention relates to the technical field of antibacterial drug curative effect evaluation, in particular to an antibacterial drug curative effect evaluation method and system, and the method comprises the following steps: collecting various optical information from an infected region; preprocessing the multiple optical information to inhibit thin film interference and extract deep biological characteristics; fusing the deep biological characteristics to form a deep infected tissue state description; performing redundancy check on the deep infected tissue state description, and evaluating the reliability of the deep infected tissue state description to obtain an antibacterial drug curative effect evaluation result; various optical information from an infected area is collected and preprocessed to inhibit film interference and extract deep biological characteristics, so that the problems of strong scattering and absorption effects of a surface film on optical signals and signal attenuation and distortion caused by the strong scattering and absorption effects in the prior art are effectively solved.

Description

Technical Field

[0001] This invention relates to the field of antimicrobial drug efficacy evaluation technology, and in particular to a method and system for evaluating the efficacy of antimicrobial drugs. Background Technology

[0002] In modern medicine, non-invasive methods for assessing the efficacy of antimicrobial drugs are widely used to monitor the treatment progress of infectious diseases due to their advantages of convenience, repeatability, and minimal trauma. This method involves emitting a specific wavelength light beam through an optical sensor probe, receiving the reflected / transmitted light signals from the infected area, and analyzing the tissue oxygen saturation, hemoglobin concentration, fluorescence intensity of cellular metabolites (such as NADH and FAD), and optical characteristics of pathogen-specific pigments (such as bacterial porphyrins) carried in the signals. This indirectly assesses the antimicrobial effect to guide adjustments to clinical protocols.

[0003] However, in complex and long-term infections such as chronic diabetic foot ulcers, pressure sores, deep burn wounds, and refractory skin infections, a thin film often forms on the surface of the infected area, containing necrotic tissue fragments, coagulated plasma proteins (such as fibrin), dried exudate, keratinocytes, and a bacterial biofilm polymer matrix. The physical and optical properties of this film dynamically change with infection progression, treatment response, and wound healing, with thicknesses ranging from tens of micrometers to several millimeters, and its optical characteristics differ significantly from those of surrounding healthy tissue or exposed infected areas. This film causes strong scattering and absorption of probe light signals: when a beam of light is incident, multiple scattering occurs due to the film's inhomogeneous structure and rough surface; oxidized hemoglobin derivatives, bacterial pigments, melanin, and necrotic tissue degradation products within the film also strongly absorb specific wavelengths of light. This leads to a significant reduction in the intensity of the effective light signal returned from deep infected tissue, changes in spectral characteristics, and a sharp deterioration in the signal-to-noise ratio.

[0004] Faced with raw signals severely scattered and attenuated by thin films, the computational steps within the device used to filter out background noise and the identification rules for extracting key feature values ​​from the signal are designed based on relatively direct and stable optical signal models. These models typically assume that light propagation in tissues is relatively predictable and that background noise mainly originates from ambient light or electronic devices. However, the presence of the thin film introduces a new, irregular "optical noise" or "interference signal," the intensity of which may even far exceed the effective signal from deep tissues. This makes traditional background subtraction methods unable to distinguish between the optical effects of the thin film and the actual biological changes in deep tissues, and the feature extraction rules fail when the signal is severely distorted.

[0005] Ultimately, the aforementioned issues with physical-optical interaction and device logic processing lead to a significant discrepancy between the antimicrobial efficacy assessment results output by this non-invasive method when evaluating infected areas covered by a thin film and the actual judgments obtained by clinicians through other means. This inconsistency severely impacts healthcare professionals' trust in and willingness to apply this non-invasive assessment method in specific complex infection scenarios.

[0006] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0007] The purpose of this invention is to address the shortcomings of existing technologies by proposing a method and system for evaluating the efficacy of antibacterial drugs.

[0008] In a first aspect, the present invention provides a method for evaluating the efficacy of antibacterial drugs, the method comprising the following steps: Collect various optical information from the infected area; The various optical information is preprocessed to suppress thin-film interference and extract deep biological features; The deep biological features are fused to form a description of the state of deep infected tissue; Redundancy checks are performed on the description of the deep-infected tissue state, and the reliability of the description of the deep-infected tissue state is evaluated to obtain the results of the antimicrobial drug efficacy evaluation.

[0009] Secondly, an antibacterial drug efficacy evaluation system is provided, the system comprising: The information acquisition module is used to acquire various optical information from the infected area, including spectral information, spatial resolution information, and temporal resolution information. A preprocessing module is used to preprocess the various optical information to suppress thin-film interference and extract deep biological features; The feature fusion module is used to fuse the deep biological features to form a description of the state of deep infected tissue; The verification and evaluation module is used to perform redundancy verification on the description of the deep infection tissue state and evaluate the reliability of the description of the deep infection tissue state to obtain the evaluation results of the antimicrobial drug efficacy.

[0010] Compared with the prior art, the present invention has the following beneficial effects: By collecting various optical information from the infected area and preprocessing this information to suppress thin-film interference and extract deep biological features, this method effectively solves the problems of strong scattering and absorption of optical signals by surface thin films in existing technologies, as well as the resulting signal attenuation and distortion. Furthermore, by fusing deep biological features to form a description of the deep infected tissue state and performing redundancy verification and reliability assessment on this description, this application overcomes the limitations of traditional methods in dealing with complex infected lesions, where the assessment results are inconsistent with actual clinical judgment due to "optical noise" or "interference signals" introduced by the thin film.

[0011] This method can distinguish between the absorption and scattering effects of the thin film itself and the actual biological changes in deep tissues, ensuring that the identification rules for key feature values ​​extracted from the signal remain effective even under severe signal distortion. Therefore, this application significantly improves the accuracy and reliability of non-invasive antimicrobial drug efficacy assessment, providing clinicians with a more credible basis for adjusting treatment plans, thereby enhancing healthcare professionals' trust in and willingness to apply this assessment method in specific complex infection scenarios. Attached Figure Description

[0012] Figure 1 This is a flowchart of the method of the present invention.

[0013] Figure 2 This is a schematic diagram of the system structure of the present invention.

[0014] In the diagram: 201, Information Acquisition Module; 202, Preprocessing Module; 203, Feature Fusion Module; 204, Verification and Evaluation Module. Detailed Implementation

[0015] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0016] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0017] This application significantly improves the accuracy and reliability of antimicrobial drug efficacy evaluation in complex infection scenarios by preprocessing various optical information to suppress thin film interference and fusing and redundancy-checking the extracted deep biological features, thereby effectively solving the problem of inaccurate evaluation results caused by thin film interference in the prior art.

[0018] The "multiple optical information" mentioned in this application refers to a set of optical signals collected from the infected area that can reflect the physiological state of the tissue. This information may include, but is not limited to, spectral information, spatially resolved information, and temporally resolved information. "Spectral information" typically refers to the characteristics of light interacting with tissue at different wavelengths, such as absorption or fluorescence spectra, which can reflect tissue components such as blood oxygen saturation, hemoglobin concentration, and cellular metabolites. "Spatially resolved information" refers to the characteristics of the spatial distribution or intensity of light as it propagates through tissue, which can be used to assess the scattering characteristics and structural heterogeneity of the tissue. "Temporally resolved information" involves the time delay distribution of photons propagating through tissue, i.e., the time spectrum of photons reaching the detector, which can provide in-depth information about photon propagation paths and tissue optical parameters. "Deep biomarkers" refer to biological indicators extracted from the multiple optical information after preprocessing, which can accurately reflect the physiological state of deeply infected tissue. These features eliminate interference from surface films and more realistically represent the biological changes within the infection focus, such as blood perfusion, oxygenation status, cell density, and inflammatory factor concentration in deep tissues. The "deep infection tissue state description" is a comprehensive and quantitative indicator formed by the integrated analysis and fusion of these deep biological characteristics. It is used to characterize the overall physiological and pathological state of the infected area and its response to antimicrobial drugs. "Redundancy verification" refers to the introduction of additional validation mechanisms to check the internal consistency and external compliance of the deep infection tissue state description to ensure its accuracy. "Reliability assessment," based on the results of redundancy verification, quantifies the trustworthiness of the deep infection tissue state description, ultimately outputting the antimicrobial drug efficacy evaluation results.

[0019] Therefore, this application is as follows Figure 1 The method shown is an antibacterial drug efficacy evaluation method, which includes the following steps: S101. Collect various optical information from the infected area; It should be noted that this step, for example, can use a multimodal optical probe that integrates a spectrometer, a spatially resolved detector, and a temporally resolved detector. The spectrometer emits broadband light and receives reflected or transmitted light to obtain spectral information of the infected area. The spatially resolved detector can consist of multiple light-receiving units at different distances from the light source, used to measure the light signal intensity at different spatial locations, thereby obtaining spatially resolved information. The temporally resolved detector emits ultrashort pulses of light and records the photon arrival time spectrum to obtain temporally resolved information. After acquisition, this optical information is transmitted to a data processing unit for subsequent analysis.

[0020] S102. Preprocess various optical information to suppress thin film interference and extract deep biological features; It should be noted that one approach to this step is based on an optical model. For example, a simplified two-layer model (a thin film layer and a deep tissue layer) can be established, and an iterative algorithm can be used to remove the influence of the thin film from the original optical signal based on the film's light scattering and absorption characteristics. Specifically, an initial thin film optical parameter can be assumed, and the original signal can be preliminarily corrected based on this parameter to obtain preliminary deep biological features. If inconsistencies exist among these preliminary features (e.g., significant differences in blood oxygen saturation reflected by spectral and spatial resolution information), the thin film optical parameter can be adjusted, and the correction process can be repeated until the deep biological features achieve a high degree of consistency. Another approach is to utilize machine learning algorithms. By training a large amount of sample data containing and without thin film interference, the model learns how to identify and extract true deep biological features from signals affected by the thin film. For example, a deep neural network can be constructed, taking the original optical information as input and outputting deep biological features after thin film interference suppression.

[0021] S103. Integrate deep biological characteristics to form a description of the state of deep infected tissue; It should be noted that one approach to this fusion step is a simple weighted average method. For example, different weights can be assigned to different deep biomarkers based on their importance to efficacy assessment or the accuracy of their extraction. These weighted features are then linearly combined to obtain a comprehensive description of the deep infection tissue state. For instance, if blood oxygen saturation is considered a more direct reflection of the infection state, it can be given a higher weight. Another approach is to use a more complex nonlinear fusion algorithm, such as neural network-based fusion. By training a multilayer perceptron, multiple deep biomarkers are used as input, and the output is a description that comprehensively reflects the infection tissue state. This method can capture the nonlinear relationships between features, thus forming a more accurate description.

[0022] S104. Redundancy check is performed on the description of deep infection tissue state, and the reliability of the description of deep infection tissue state is evaluated to obtain the evaluation results of antimicrobial drug efficacy.

[0023] It should be noted that, in this step, one verification method is to introduce additional physical perturbations to verify the stability of deep biomarkers. For example, a slight mechanical vibration can be applied to the surface of the infected area while simultaneously monitoring the response of the optical signal. If the description of the deep infected tissue state remains stable before and after the vibration, its reliability is considered high. Significant fluctuations may indicate that film interference has not been completely suppressed or that there are errors in the extraction of deep biomarkers. Another verification method is to utilize the inherent correlation between different optical information sources for cross-validation. For example, if the inflammatory indicators reflected by the spectral information are inconsistent with the cell density change trend reflected by the time-resolved information, it may be necessary to revise the description of the deep infected tissue state and lower its reliability score. Finally, based on the results of redundancy verification and combined with preset evaluation criteria, a reliability score for the description of the deep infected tissue state is calculated, and the evaluation result of antimicrobial drug efficacy is given based on this score. For example, when the reliability score is higher than a certain threshold, the evaluation result is considered credible and can be used to guide clinical decision-making.

[0024] The antimicrobial drug efficacy evaluation method proposed in this application effectively overcomes the serious interference of surface films on optical signals in existing technologies by introducing a series of innovative steps, including multimodal optical information acquisition, thin-film interference suppression preprocessing, deep biological feature fusion, redundancy verification, and reliability assessment. Traditional methods often struggle to distinguish between the absorption and scattering effects of the thin film itself and the actual biological changes in deep tissues, leading to evaluation results that do not match clinical reality.

[0025] Compared with existing technologies, the core innovation of this application lies in its refined processing of film interference and the quantification of the reliability of evaluation results. Existing technologies, when facing infected areas covered by films, suffer from a lack of effective film interference suppression mechanisms, leading to severe attenuation and distortion of effective light signals returning from deep tissues, making subsequent feature extraction and efficacy assessment unreliable. This application, through preprocessing of various optical information, effectively suppresses film interference, thereby extracting more realistic deep biological features. Furthermore, the redundancy verification and reliability assessment mechanisms introduced in this application further enhance the credibility of the evaluation results, enabling physicians to more accurately determine the efficacy of antimicrobial drugs. For example, by applying mechanical vibration and capturing the synchronous optical signal response, micron-level physical interface disturbances between the film and deep tissue can be identified, and redundancy verification of the deep infected tissue state description can be performed accordingly, thus avoiding misjudgments caused by dynamic changes in the film. This multi-dimensional, multi-level verification mechanism significantly improves the accuracy and clinical applicability of the evaluation results, providing a more reliable solution for evaluating the efficacy of antimicrobial drugs in complex infectious lesions.

[0026] As one embodiment of the present invention, the various optical information includes spectral information, spatial resolution information, and temporal resolution information.

[0027] It should be noted that spectral information refers to information obtained by analyzing the absorption, scattering, or fluorescence characteristics at different wavelengths after light interacts with biological tissue. Its purpose is to reveal the composition and concentration changes of biomolecules within the infected area, such as hemoglobin, water, lipids, and bacterial metabolites. In practical applications, spectral information can be obtained by illuminating the infected area with a broadband light source and using a spectrometer to collect reflected, transmitted, or fluorescent spectra.

[0028] Spatial resolution information refers to information obtained by analyzing the spatial distribution characteristics of light as it propagates through biological tissues. Its purpose is to reveal the tissue structure, cell density, and distribution of light scattering centers within the infected area. For example, by measuring the light signal intensity or spot size at different distances, the optical heterogeneity of the tissue can be inferred. Specifically, spatial resolution information can be obtained by using multiple spatially separated light sources and / or detector arrays to collect light signals from different locations.

[0029] Time-resolved information refers to information obtained by analyzing the arrival time distribution characteristics of light as it propagates through biological tissues. Its purpose is to reveal the average propagation path of photons in tissues, the number of scattering events, and absorption characteristics, thereby reflecting the tissue's microstructure and physiological state. For example, by measuring the photon arrival time spectrum after ultrashort pulses of light propagate through tissues, it is possible to distinguish between early-arriving ballistic photons and diffuse photons that have undergone multiple scattering. Specifically, time-resolved information can be acquired by using an ultrafast laser to emit femtosecond or picosecond pulses, combined with time-correlated single-photon counting (TCSPC) technology.

[0030] This application's approach concretizes various optical information into spectral, spatially resolved, and temporally resolved information, enabling a comprehensive capture of the biological and physical characteristics of the infected area from multiple dimensions. Spectral information provides insights into biochemical composition and metabolic state, spatially resolved information reveals changes in tissue structure and cell density, while temporally resolved information deeply reflects the microscopic propagation characteristics of photons within the tissue and the physiological environment. These different types of information complement each other, collectively constructing a more complete and accurate description of the infected area's state, laying a solid data foundation for subsequent deep biomarker extraction and fusion.

[0031] The aforementioned technical solution ensures that the acquired optical information possesses high specificity and comprehensiveness, thereby more accurately reflecting the deep biological characteristics of the infected area. Specifically, spectral information helps identify pathogen-specific metabolites or biomarkers of host inflammatory responses; spatially resolved information can distinguish the boundaries and internal structural changes of infection foci; and temporally resolved information can effectively penetrate surface membranes to obtain microscopic physiological parameters of deep tissues. This combination of multimodal optical information significantly improves the ability to perceive complex physiological and pathological processes in the infected area, providing more reliable and abundant data support for the accurate evaluation of antimicrobial drug efficacy.

[0032] As one embodiment of the present invention, the steps of performing redundancy verification on the description of deep infected tissue state and evaluating the reliability of the description of deep infected tissue state include: Mechanical vibrations are applied to the surface of the infected area, and optical signal responses synchronized with the mechanical vibrations are captured to identify whether there are micron-level physical interface disturbances between the thin film and deep tissue. It should be noted that applying mechanical vibration to the surface of the infected area refers to applying a controllable, minute physical disturbance to the surface layer of the infected area through an external excitation source. For example, devices such as piezoelectric ceramic actuators, miniature loudspeakers, or pneumatic pulse generators can be used to generate mechanical waves with adjustable frequency and amplitude, which can then be applied to the skin or tissue surface of the infected area. Capturing the optical signal response synchronized with the mechanical vibration refers to using an optical detector to monitor changes in the light signals reflected, transmitted, or scattered from the infected area in real time while the mechanical vibration is applied. These optical signal responses can be changes in light intensity, phase, polarization state, or photon arrival time distribution. The purpose is to detect whether there are micrometer-level physical interface disturbances between the film and the deeper tissue by observing how the light signal changes with mechanical vibration. Micrometer-level physical interface disturbances can be understood as structural discontinuities, differences in adhesion, or tiny gaps existing between the film and the deeper tissue at the micrometer scale. These disturbances will exhibit specific optical response characteristics under the action of mechanical vibration.

[0033] When micron-level physical interface perturbations are identified, the distortion patterns caused by the micron-level physical interface perturbations to spectral information, spatial resolution information, and temporal resolution information are recorded. It should be noted that distortion patterns refer to specific variations in optical signals across the spectral, spatial, and temporal dimensions caused by perturbations at micrometer-level physical interfaces. For example, perturbations may lead to shifts or broadening of spectral absorption peaks, blurring or artifacts in spatially resolved images, and changes in the distribution of early-arriving photons in the photon arrival time spectrum of temporally resolved signals. The purpose of recording these distortion patterns is to provide crucial correction or compensation criteria for subsequent redundancy checks and reliability assessments, thereby enabling a more accurate understanding of the impact of perturbations on deep biomarker extraction.

[0034] By comparing and quantifying the consistency deviations among deep biological characteristics reflected by spectral, spatially resolved, and temporally resolved information, redundancy checks are performed on descriptions of deep-infected tissue states.

[0035] It should be noted that the above step can be understood as cross-validating identical or related deep biological features extracted from different optical modalities (spectral, spatial, and temporal). For example, if there is a significant difference between the blood oxygen saturation indicated by spectral information and that indicated by spatially or temporally resolved information, it indicates a discrepancy in consistency. This discrepancy can be quantified by calculating the statistical differences in feature values ​​between different modalities (such as root mean square error, correlation coefficient, or information entropy). The purpose is to identify inaccuracies or inconsistencies in deep biological feature extraction caused by thin-film interference or other factors through cross-verification of multi-source information, thereby making a preliminary judgment on the reliability of the description of the state of deeply infected tissues.

[0036] This application's solution introduces an active detection mechanism—applying mechanical vibration and capturing synchronous optical signal responses—to effectively identify micron-level physical interface perturbations that may exist between the thin film and deep tissue. This active detection method overcomes the hidden interference that traditional passive observation may fail to detect, thus providing more comprehensive environmental information for subsequent redundancy verification. It is precisely because these micron-level perturbations can be identified that it becomes possible to record the distortion patterns they produce on spectral, spatial, and temporal resolution information. Accurate recording of these distortion patterns allows for a deeper understanding of the specific impacts on different optical modes, laying the foundation for subsequent feature correction and reliability assessment. Based on this, comparing and quantifying the consistency deviations between deep biological features reflected by different optical information allows for multi-dimensional cross-verification of the reliability of deep biological features. When micron-level physical interface perturbations exist, optical signals of different modes may be subject to varying degrees or types of interference, leading to inconsistencies between the extracted deep biological features. By quantifying this consistency deviation, this application can more accurately determine the reliability of the description of the state of deeply infected tissue, avoiding misjudgments caused by interference with single-mode information.

[0037] Through the above technical solutions, this application can significantly improve the accuracy and reliability of antimicrobial drug efficacy evaluation results. Specifically, by actively identifying micron-level physical interface perturbations between the film and deep tissues and recording their distortion patterns on multimodal optical information, these potential interference factors can be fully considered and compensated for during redundancy verification. This avoids inaccurate extraction of deep biological features due to complex physical environmental factors such as film interference, thereby improving the intrinsic consistency and reliability of the description of deep infected tissue states. Furthermore, by comparing and quantifying the consistency deviations between deep biological features reflected by different optical information, this application provides a more robust redundancy verification mechanism that can promptly detect and correct evaluation biases caused by damage to single-modal information, ensuring that the final antimicrobial drug efficacy evaluation results are more objective and credible.

[0038] In some preferred embodiments, a specific example is illustrated below. Suppose it is necessary to evaluate the efficacy of antimicrobial drugs in a deep skin infection area of ​​a patient. First, a small piezoelectric vibrator is used to apply slight mechanical vibration to the surface of the infection area at a frequency of approximately 150 Hz and a micrometer-level amplitude. Simultaneously, a multimodal optical imaging system integrating spectral, spatially resolved, and temporally resolved detectors continuously acquires optical information from the infection area. During the application of mechanical vibration, the system analyzes the response of the optical signal in real time. For example, if periodic fluctuations in light intensity synchronized with the vibration frequency or specific changes in the photon arrival time spectrum are detected, a micrometer-level physical interface perturbation between the film and the deep tissue can be identified. Once this perturbation is identified, the system immediately records the specific effects of the perturbation on different optical information. For example, the instantaneous redshift or blueshift amplitude of specific absorption peaks in the spectral information, the dynamic changes in the degree of light spot diffusion in the spatially resolved information, and the reduction in the number of early-arriving photons or changes in their distribution patterns in the temporally resolved information are recorded. These recorded features constitute the distortion mode. Subsequently, the system extracts deep biological features based on these perturbation-affected optical information, such as blood oxygen saturation, tissue scattering coefficient, and metabolite concentration. Then, these deep biological features extracted from the spectral, spatial, and temporal dimensions are compared. For example, if the blood oxygen saturation obtained from spectral analysis is 85%, while the blood oxygen saturation obtained from spatial resolution analysis after distortion mode correction is 70%, this 15% consistency deviation is calculated and quantified. By comprehensively considering the identified micron-level physical interface perturbations, the recorded distortion modes, and the quantified consistency deviation, the system can more accurately assess the reliability of the description of deep-infected tissue states, thus providing more reliable data support for the evaluation of antimicrobial drug efficacy.

[0039] As one embodiment of the present invention, the step of evaluating the reliability of the description of the state of deep-infected tissue to obtain the evaluation results of antimicrobial drug efficacy includes: When there are inconsistencies among deep biological characteristics, the root cause of the inconsistencies can be identified by combining the perturbation characteristics of micron-level physical interfaces and distortion patterns. It should be noted that "deviations in consistency among deep biological features" refers to the inconsistencies in the data of deep biological features reflected by spectral, spatially resolved, and temporally resolved information across different dimensions or measurement channels. These deviations may manifest as differences in measurement results for the same biological feature from different optical information sources, or fluctuations in measurement results for the same biological feature from the same optical information source at different time points.

[0040] "Micron-level physical interface perturbation characteristics" refer to the specific attributes of micron-level physical interface perturbations between the thin film and deep tissue, identified by applying mechanical vibrations to the surface of the infected area and capturing the synchronous optical signal response. These attributes include the frequency, amplitude, duration, spatial distribution of the perturbation, and its impact on the light transmission path. "Distortion patterns" refer to specific, identifiable signal distortion or deviation patterns caused by micron-level physical interface perturbations on spectral, spatially resolved, and temporally resolved information. Examples include spectral peak shifts, spatial image blurring, broadening of the photon arrival time spectrum, or specific shape changes. Identifying the root cause of consistency deviations involves analyzing whether the observed consistency deviation matches known micron-level physical interface perturbation characteristics and their corresponding distortion patterns. This helps distinguish whether the deviation originates from genuine deep biological changes or is caused by thin film interference or measurement artifacts. For example, if the consistency deviation pattern highly matches a distortion pattern caused by a specific thin film vibration, it can be determined that the deviation is primarily caused by thin film interference.

[0041] Based on the consistency of deep biological characteristics and the effect of micron-level physical interface perturbation, the reliability of the description of deep infected tissue state is calculated to obtain the results of antimicrobial drug efficacy evaluation.

[0042] It should be noted that "the degree of consistency of deep biological features" can be understood as the degree to which deep biological features corroborate each other across different optical information dimensions or channels. It is the inverse indicator of "consistency deviation," and can be quantified, for example, by calculating the correlation coefficient, mean square error, or standardized difference between different feature measurements. The higher the degree of consistency, the more reliable the measurement results of deep biological features. "Micron-level physical interface perturbation processing effect" refers to the effectiveness of suppressing or compensating for film interference and micron-level physical interface perturbations during the preprocessing stage or redundancy verification process. For example, it can be quantified by evaluating the intensity of residual perturbation signals, the degree of distortion mode elimination, or the deviation between the corrected data and the true value. Calculating the reliability of the description of the state of deep infected tissue specifically refers to combining the identified root causes, consistency degree, and perturbation processing effect, and generating a quantitative reliability score through a preset algorithm or model. For example, when the consistency deviation is determined to be mainly caused by film interference and this interference has been effectively processed, the reliability score can be maintained at a high level; conversely, if the consistency deviation is determined to be a real biological change and the consistency degree is low, the reliability score will decrease accordingly.

[0043] This application's solution effectively addresses the problem of accurately assessing the reliability of descriptions of deep-infected tissue states when inconsistencies exist in deep biomarkers by introducing a mechanism to discriminate the root causes of inconsistencies. Specifically, when inconsistencies arise between deep biomarkers, this solution does not simply mark the data as unreliable. Instead, it further analyzes the root causes of the inconsistencies by combining micrometer-level physical interface perturbation characteristics and distortion patterns. This analysis enables the system to distinguish between measurement artifacts caused by external factors such as thin-film interference and data differences caused by actual biological changes. It is precisely this root cause discrimination that allows subsequent reliability calculations to be more accurate and targeted.

[0044] Building upon this foundation, by comprehensively considering the consistency of deep biological features and the effectiveness of micron-level physical interface perturbation processing, this approach provides a more comprehensive and robust reliability assessment. The consistency of deep biological features directly reflects the inherent stability of measurements of the same biological feature from different optical information sources, while the effectiveness of micron-level physical interface perturbation processing quantifies the system's ability to suppress external interference. Combining these two aspects ensures that the reliability assessment not only focuses on the intrinsic quality of the data itself but also fully considers the impact of the external environment on data acquisition and the degree to which it is effectively controlled. This multi-dimensional, contextualized assessment approach makes the reliability score of the final description of deep-infected tissue state more convincing, thus providing a solid data foundation for the evaluation of antimicrobial drug efficacy.

[0045] Through the aforementioned technical solutions, this application can significantly improve the accuracy and reliability of antimicrobial drug efficacy assessment results. Specifically, by identifying the root causes of deviations in the consistency of deep biological features, it can effectively avoid misjudging measurement artifacts caused by non-biological factors such as membrane interference as real biological changes, thereby reducing the risk of misdiagnosis or misjudgment. Furthermore, by combining the consistency of deep biological features with the effect of micron-level physical interface perturbation processing to calculate reliability, the assessment results become more refined and contextualized, more accurately reflecting the true credibility of the description of deep-infected tissue states. This more rigorous reliability assessment mechanism provides clinicians with more valuable antimicrobial drug efficacy assessment results, helping to develop more precise treatment plans and optimize patient treatment outcomes.

[0046] As one embodiment of the present invention, the step of preprocessing various optical information to suppress thin-film interference and extract deep biological features includes: Based on the preset initial thin-film optical parameters, the spectral information, spatial resolution information and temporal resolution information are preliminarily processed to estimate the preliminary physiological characteristics of deep infected tissues. It should be noted that the initial thin-film optical parameters can be understood as preliminary estimates of the optical properties of the thin film on the surface of the infected area, such as the film thickness, refractive index, scattering coefficient, or absorption coefficient. These parameters can be obtained based on empirical data, typical values, or through preliminary measurements. Preliminary processing of spectral, spatially resolved, and temporally resolved information refers to constructing a preliminary thin-film optical model using these initial parameters, and then using this model to perform preliminary thin-film interference removal on the various optical information collected. This allows for the estimation of preliminary physiological characteristics of the deep-infected tissue from the optical information after preliminary thin-film interference removal. These preliminary physiological characteristics may include, but are not limited to, the oxygen saturation, hemoglobin concentration, cell density, or metabolite concentration of the deep tissue.

[0047] Compare the consistency among preliminary physiological characteristics; It should be noted that this step refers to assessing whether there are significant differences or contradictions between the same or related preliminary physiological characteristics estimated using different optical information (e.g., spectral information, spatially resolved information, and temporally resolved information) or different analytical methods. For example, if there is a large deviation between the oxygen saturation estimated using spectral information and the oxygenation status indirectly inferred using spatially resolved or temporally resolved information, it indicates that the preliminary film interference suppression may not be accurate enough.

[0048] When there are differences among preliminary physiological characteristics, the optical parameters of the thin film are adjusted to reduce the differences among preliminary physiological characteristics; It should be noted that when there are differences between the preliminary physiological characteristics, the thin film optical parameters will be adjusted. This adjustment process aims to optimize the thin film optical parameters so that the consistency between the re-estimated preliminary physiological characteristics under the new thin film model is improved. This can be an iterative process, for example, by using optimization algorithms (such as gradient descent, genetic algorithms, etc.) or lookup tables to systematically adjust the thin film optical parameters until the consistency between the preliminary physiological characteristics reaches a preset threshold or converges.

[0049] Adjusted thin-film optical parameters are used to process various optical information to extract deep biological features.

[0050] It should be noted that the above step means that, under optimized and verified thin-film optical parameters, more precise thin-film interference suppression is performed on the original collection of various optical information, thereby enabling more accurate extraction of biological features that represent the true state of deep-infected tissue.

[0051] This application's solution effectively addresses the problem of traditional methods struggling to accurately suppress thin-film interference when thin-film optical parameters are unknown or dynamically changing by introducing an adaptive thin-film optical parameter adjustment mechanism. Specifically, it first processes various optical information based on initially estimated thin-film optical parameters and estimates preliminary physiological characteristics of deeply infected tissues. Since different types of optical information (spectral, spatial resolution, temporal resolution) differ in their sensitivity to thin-film interference and their mechanisms of reflecting deep tissue characteristics, inconsistencies among preliminary physiological characteristics typically indicate that the initial thin-film optical parameters fail to accurately reflect the actual thin-film properties, resulting in ineffective suppression of thin-film interference. Therefore, comparing the consistency among these preliminary physiological characteristics can serve as an effective indicator of the accuracy of the current thin-film optical parameters. When inconsistencies are detected, the system iteratively adjusts the thin-film optical parameters and re-processes and re-estimates the features until the preliminary physiological characteristics from different sources achieve a high degree of consistency. This process essentially utilizes the inherent correlation between multimodal optical information for cross-validation and parameter optimization, enabling the adjusted thin-film optical parameters to more accurately characterize the actual thin-film properties, thereby achieving more precise suppression of thin-film interference.

[0052] Through the above technical solution, this application can adaptively optimize the optical parameters of the thin film, thereby significantly improving the suppression effect on thin film interference when the optical parameters are unknown or dynamically changing. As a result, the accuracy of deep biological feature extraction is significantly improved, avoiding misjudgments or information distortion caused by thin film interference. This makes the evaluation results of antibacterial drug efficacy more reliable and accurate, providing clinicians with more accurate treatment decision-making basis and contributing to the realization of precision medicine.

[0053] As one embodiment of the present invention, the step of fusing deep biological features to form a description of the state of deep infected tissue includes: The ratio of optical signal intensity of receiving units at different distances in spatially resolved information and the distribution of early arriving photons in the photon arrival time spectrum in temporal resolved information are continuously monitored to obtain local optical property indicators of the thin film. It should be noted that the ratio of optical signal intensity at different distances in the spatially resolved information can be understood as the ratio of the captured optical signal intensity measured by placing optical receivers at different distances and calculating these ratios. This ratio reflects the scattering and absorption characteristics of light in thin films and shallow tissues. The early arrival photon distribution in the time-resolved information refers to the analysis of the time distribution of photons from incident to received by the detector using time-resolved detection technology. Early-arriving photons typically have shallower penetration depths and are more sensitive to thin film properties. By comprehensively analyzing this information, optical properties such as the local scattering coefficient and absorption coefficient of the thin film can be quantified.

[0054] Identify changes in local optical characteristic indicators. Changes in local optical characteristic indicators refer to the rate or magnitude of change of local optical characteristic indicators within a preset time window exceeding a preset threshold. It should be noted that identifying changes in local optical properties means that when the rate or magnitude of change of a local optical property exceeds a preset threshold within a preset time window, the thin film properties are considered to have changed significantly. For example, when the scattering coefficient of a thin film increases rapidly in a short period of time, or when the absolute value of its absorption coefficient changes by more than a certain percentage, the identification mechanism will be triggered. Its purpose is to promptly detect dynamic changes in the thin film state, providing a basis for evaluating the accuracy of subsequent feature extraction.

[0055] To assess the impact of changes in local optical properties on the accuracy of extracting various deep biological features from spectral, spatial, and temporal information; It should be noted that the purpose of this step is to quantify the differentiated impact of thin film variations on the data quality of different optical modes. For example, enhanced scattering of the thin film may have a greater impact on the depth penetration capability of spatially resolved information, while having a smaller impact on specific absorption peaks of spectral information. Through this assessment, it is possible to identify which sources of optical information are more reliable for deep biological signatures under the current thin film condition.

[0056] Based on the degree of influence, dynamic weights are assigned to the source channels of each deep biological feature; It should be noted that dynamic weights refer to weight values ​​that are adjusted in real time based on the evaluation results. Channels with less influence (i.e., higher extraction accuracy) have larger weight values; conversely, channels with greater influence (i.e., lower extraction accuracy) have smaller weight values. The purpose is to prioritize deep biological features that are less affected by thin film interference and have higher extraction accuracy in the subsequent fusion process.

[0057] Based on dynamic weights, the deep biological characteristics reflected by spectral information, spatial resolution information and temporal resolution information are weighted and fused to form a description of the state of deep infected tissues.

[0058] It should be noted that weighted fusion is a method that combines information from multiple data sources. By assigning different weights to each data source, the final fusion result can better reflect the true situation. Its purpose is to generate a comprehensive and reliable description of the state of deep-seated infections, providing accurate input for evaluating the efficacy of antimicrobial drugs.

[0059] The proposed solution continuously monitors the local optical properties of the thin film and identifies its changes, enabling real-time understanding of the film's impact on optical signal transmission. This allows for the assessment of the degree to which these changes affect the accuracy of deep biomarker extraction from different optical information sources (spectral, spatially resolved, and temporally resolved information). Based on this degree of impact, dynamic weights are assigned to the source channels of each deep biomarker, ensuring that features from channels less affected by the film or with higher extraction accuracy receive higher weights during the fusion process. This effectively suppresses interference from dynamic film changes and improves the accuracy and reliability of describing the state of deeply infected tissues.

[0060] Through the above technical solution, this application can dynamically adapt to changes in the optical properties of the thin film on the surface of the infected area, avoiding the problem of inaccurate extraction of deep biological features caused by thin film interference. By dynamically weighting and fusing deep biological features from different optical information sources, the robustness and accuracy of describing the state of deep infected tissues are significantly improved, thereby making the evaluation results of antimicrobial drug efficacy more accurate and reliable, and providing a more solid data foundation for clinical decision-making.

[0061] As one embodiment of the present invention, the step of evaluating the impact of changes on the accuracy of extraction of various deep biological features from spectral information, spatial resolution information, and temporal resolution information includes: Monitor the instantaneous fluctuation amplitude of various deep biological characteristics reflected by spectral information, spatial resolution information and temporal resolution information during the period of change; Calculate the cross-channel consistency deviation of the same deep biological feature reflected by spectral, spatially resolved, and temporally resolved information during the period of change; The impact of quantization changes on the accuracy of extracting various deep biological features from spectral, spatial, and temporal information is determined based on instantaneous fluctuation amplitude and cross-channel consistency deviation.

[0062] Specifically, instantaneous fluctuation amplitude refers to the range or standard deviation of numerical changes in each deep biofeature extracted from spectral, spatially resolved, and temporally resolved information within a very short time window during a change in the local optical properties of the thin film. Its purpose is to directly reflect the immediate impact of thin film changes on feature stability. For example, it can be obtained by calculating the difference between the maximum and minimum values ​​of consecutive samples of a certain deep biofeature within a specific time window, or its standard deviation. Cross-channel consistency deviation can be understood as the degree of difference in the same deep biofeature (e.g., tissue oxygen saturation, blood perfusion, etc.) reflected by spectral, spatially resolved, and temporally resolved information across different optical channels during a change in the local optical properties of the thin film. Its purpose is to assess the differential impact of thin film interference on different optical information channels. In practical applications, this deviation can be quantified by calculating the relative error, correlation coefficient, or root mean square error between the extraction results of the same deep biofeature in different channels. Furthermore, based on instantaneous fluctuation amplitude and cross-channel consistency deviation, the degree of influence of changes on the extraction accuracy of each deep biofeature can be comprehensively quantified. For example, a weighted function or machine learning model can be designed to use instantaneous fluctuation amplitude as an indicator of feature stability and cross-channel consistency deviation as an indicator of feature reliability. A combined score of these two metrics can then be output to determine the overall impact. A higher score indicates a greater impact of thin-film variations on the accuracy of feature extraction.

[0063] This application's approach, by introducing monitoring of the instantaneous fluctuation amplitude of deep biomarkers, can directly capture the immediate impact of changes in the local optical properties of the thin film on feature stability. Simultaneously, by calculating the cross-channel consistency deviation of the same deep biomarker across different optical information channels, it can effectively identify the differentiated impact of thin film interference on different channels, thereby more comprehensively evaluating the reliability of feature extraction. It is precisely this multi-dimensional and refined evaluation mechanism that makes the quantification of the degree of influence of thin film changes more accurate and robust, providing a solid foundation for subsequent dynamic weight allocation.

[0064] As one embodiment of the present invention, the step of continuously monitoring the ratio of optical signal intensity of receiving units at different distances in spatially resolved information, and the distribution of early-arriving photons in the photon arrival time spectrum in temporally resolved information, to obtain local optical property indicators of the thin film includes: By using multiple spatially separated optical receivers, the ratio of the light signal intensity captured by different optical receivers is continuously calculated to obtain the ratio of the light signal intensity of receiving units at different distances in the spatial resolution information; It should be noted that spatially separated photodetectors refer to multiple photodetectors arranged at different distances or angles on or near the surface of the infected area. These detectors can capture light signals scattered or transmitted from the infected area. Due to the attenuation and scattering of light as it propagates through the medium, the intensity of the light signals captured by detectors at different distances reflects the optical properties of the medium (including thin films and deep tissues). The light signal intensity ratio refers to the ratio between the light signal intensities captured by different spatially separated photodetectors. This ratio can effectively eliminate the influence of common-mode noise such as light source intensity fluctuations and is more sensitive to the scattering and absorption characteristics of thin films, especially in the near-infrared or visible light bands.

[0065] By using a time-resolved detector, the photon arrival time spectrum is continuously analyzed to extract the number of early-arriving photons or their distribution characteristics within a preset time window, so as to obtain the distribution of early-arriving photons in the photon arrival time spectrum in the time-resolved information. It should be noted that a time-resolved detector is a detector capable of measuring the time distribution of photons arrival, such as a single-photon avalanche diode (SPAD) array or a streak camera. It records the time required for each photon to travel from the source to the detector, thus forming a photon arrival time spectrum. The photon arrival time spectrum describes the time distribution of photons as they travel through the medium and reach the detector. Early-arriving photons typically experience fewer scattering events and carry more information about the shallow medium (such as a thin film). The number of early-arriving photons or their distribution characteristics refer to the number of photons arriving within a pre-defined short time window in the photon arrival time spectrum, or their temporal distribution pattern. These early photons are significantly influenced by the optical properties of the thin film and can be used to characterize the scattering and absorption effects of the thin film.

[0066] Based on the ratio of optical signal intensity and the number or distribution characteristics of early-arriving photons, the local scattering and absorption effects of the thin film are comprehensively quantified to obtain the local optical properties of the thin film.

[0067] It should be noted that the local scattering and absorption effects of thin films refer to the ability of the thin film on the surface of the infected region to scatter and absorb incident light. Scattering effects describe the degree to which photons change their propagation direction, while absorption effects describe the degree to which photon energy is absorbed by the medium. These effects directly affect the transmission path and intensity of light in the thin film. Local optical property indices are quantitative representations of the local scattering and absorption effects of thin films, such as scattering coefficient, absorption coefficient, and reduced scattering coefficient. These indices are key parameters for assessing the degree of interference of thin films in the extraction of deep biological features. Comprehensive quantification means combining spatially resolved information (the ratio of light signal intensity) and temporally resolved information (the number of early-arriving photons or their distribution characteristics) from two different dimensions of data, and jointly inferring the local optical property indices of the thin film by establishing a physical model or a data-driven model. This multimodal combination can improve the accuracy and robustness of the quantification results.

[0068] This application's scheme achieves precise acquisition of local optical property indicators of thin films by combining spatially and temporally resolved information. Specifically, by utilizing multiple spatially separated photodetectors, the spatial intensity distribution of light after propagation in the thin film and shallow tissues can be captured. By analyzing the ratio of light signal intensity captured by detectors at different distances, the scattering and absorption characteristics of the thin film can be effectively reflected. Simultaneously, analyzing the photon arrival time spectrum using a temporally resolved detector, particularly extracting the number or distribution characteristics of early-arriving photons, provides more direct information on photon propagation paths and scattering events within the thin film. Early-arriving photons, due to their shorter propagation paths, are less affected by deeper tissues and are therefore more sensitive to the local optical properties of the thin film. By comprehensively quantifying these two different but complementary optical information dimensions, the local scattering and absorption effects of the thin film can be characterized more comprehensively and accurately, thus obtaining reliable indicators of the thin film's local optical properties. This multimodal monitoring method effectively overcomes the limitations of single optical information in characterizing thin film properties in complex biological tissues, laying the foundation for the accurate extraction and fusion of subsequent deep biological features.

[0069] As one embodiment of the present invention, the step of comprehensively quantifying the local scattering and absorption effects of the thin film based on the ratio of optical signal intensity and the number or distribution characteristics of early-arriving photons to obtain the local optical property index of the thin film includes: Based on the pre-defined physical properties of thin-film optical transmission, a correspondence is established between the ratio of optical signal intensity and the number or distribution characteristics of early-arriving photons and the local scattering and absorption effects of thin films. It should be noted that the above step refers to using known physical or empirical models to describe the propagation behavior of light in a thin-film medium. The physical properties of thin-film optical transmission can include parameters such as the film's thickness, refractive index, scattering coefficient, and absorption coefficient. The corresponding relationship can be a mathematical function, a lookup table, or a machine learning model, whose inputs are the ratio of light signal intensity and the number or distribution characteristics of early-arriving photons, and whose output is the local scattering and absorption effects of the thin film. For example, methods such as diffusion approximation theory, Monte Carlo simulation, or radiative transfer equations can be used to establish such a physical model to predict the light signal response at different spatial distances and time windows under given thin-film optical properties.

[0070] By adjusting the parameters of local scattering and absorption effects of the thin film, the correspondence is optimally matched with the ratio of the measured optical signal intensity and the number or distribution characteristics of early-arriving photons. It should be noted that the above step refers to treating the local scattering and absorption effects of the thin film as adjustable parameters in the established physical or empirical model. These parameters are continuously adjusted using iterative optimization algorithms, such as least squares, gradient descent, or genetic algorithms, to minimize the error between the model's predicted light signal intensity ratio and the number or distribution characteristics of early-arriving photons and the actual measured values. "Best match" can be understood as the difference between the model's predicted values ​​and the actual measured values ​​reaching a preset minimum threshold, or achieving a statistically optimal fit.

[0071] Based on the correspondence that achieves the best match, the local scattering and absorption effects of the thin film are comprehensively quantified to obtain the local optical properties of the thin film.

[0072] It should be noted that the above step refers to using the parameter values ​​of the local scattering and absorption effects of the thin film in the model at this point, after the parameters have been adjusted and optimally matched, as the final quantification result. These parameter values ​​represent the local optical properties of the thin film in the current infected region, such as the effective scattering coefficient and effective absorption coefficient of the thin film.

[0073] This application's solution effectively addresses the limitations of traditional comprehensive quantization methods in accurately capturing the complex changes in thin-film optical properties by introducing a model-building and parameter optimization process based on pre-defined thin-film optical transmission physical characteristics. Specifically, firstly, by establishing a physical correspondence between the light signal intensity ratio, the distribution of early-arriving photons, and the local scattering and absorption effects of the thin film, a solid theoretical foundation is provided for the quantization process. This ensures that the interpretation of the light signal is no longer a simple empirical correlation but based on the intrinsic physical laws of light propagation in the medium. Secondly, by adjusting the parameters of the local scattering and absorption effects of the thin film and achieving optimal matching with actual measurement data, this solution can adaptively correct the deviation between the model and the actual situation, thereby overcoming the challenges posed by inhomogeneous or dynamically changing thin-film structures. This iterative optimization process ensures that the acquired local optical property indicators of the thin film can more accurately reflect the true state of the thin film in the infected area, thus laying the foundation for the accurate extraction of subsequent deep biological features.

[0074] like Figure 2 The system shown is an antimicrobial drug efficacy evaluation system, which includes: The information acquisition module 201 is used to acquire various optical information from the infected area, including spectral information, spatial resolution information and temporal resolution information. The preprocessing module 202 is used to preprocess various optical information to suppress thin film interference and extract deep biological features; Feature fusion module 203 is used to fuse deep biological features to form a description of the state of deep infected tissue; The verification and evaluation module 204 is used to perform redundancy verification on the description of deep infection tissue state and evaluate the reliability of the description of deep infection tissue state in order to obtain the evaluation results of antimicrobial drug efficacy.

[0075] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A method for evaluating the efficacy of an antibacterial drug, characterized by, The method comprises the following steps: Collecting multiple optical information from the infected area; Preprocessing the multiple optical information to suppress the interference of the thin film and extract deep biological features; Fusing the deep biological features to form a deep infected tissue state description; Performing a redundancy check on the deep infected tissue state description and evaluating the reliability of the deep infected tissue state description to obtain an antibacterial drug efficacy evaluation result.

2. The method for evaluating the therapeutic efficacy of an antibacterial drug according to claim 1, wherein The multiple optical information includes spectral information, spatial resolution information, and time resolution information.

3. The method for evaluating the therapeutic efficacy of an antibacterial drug according to claim 2, wherein The step of performing a redundancy check on the deep infected tissue state description and evaluating the reliability of the deep infected tissue state description comprises: Applying mechanical vibration to the surface of the infected area and capturing optical signal responses synchronized with the mechanical vibration to identify whether there is a micron-level physical interface disturbance between the thin film and the deep tissue; When the micron-level physical interface disturbance is identified, recording the distortion mode of the spectral information, the spatial resolution information, and the time resolution information caused by the micron-level physical interface disturbance; Comparing and quantifying the consistency deviation between the deep biological features reflected by the spectral information, the spatial resolution information, and the time resolution information to perform a redundancy check on the deep infected tissue state description.

4. The method for evaluating the therapeutic efficacy of an antibacterial drug according to claim 3, wherein The step of evaluating the reliability of the deep infected tissue state description to obtain an antibacterial drug efficacy evaluation result comprises: When there is a consistency deviation between the deep biological features, combining the micron-level physical interface disturbance characteristics and the distortion mode to determine the root cause of the consistency deviation; According to the consistency degree of the deep biological features and the micron-level physical interface disturbance processing effect, calculating the reliability of the deep infected tissue state description to obtain an antibacterial drug efficacy evaluation result.

5. The method for evaluating the therapeutic efficacy of an antibacterial drug according to claim 2, wherein The step of preprocessing the multiple optical information to suppress the interference of the thin film and extract deep biological features comprises: Based on the preset initial thin film optical parameters, performing preliminary processing on the spectral information, the spatial resolution information, and the time resolution information to estimate the preliminary physiological characteristics of the deep infected tissue; Comparing the consistency between the preliminary physiological characteristics; When there is a difference between the preliminary physiological characteristics, adjusting the thin film optical parameters to reduce the difference between the preliminary physiological characteristics; Using the adjusted thin film optical parameters to process the multiple optical information to extract the deep biological features.

6. The method for evaluating the therapeutic efficacy of an antibacterial drug according to claim 2, wherein The step of fusing the deep biological features to form a deep infected tissue state description comprises: Continuously monitoring the light signal intensity ratio of different distance receiving units in the spatial resolution information and the early arrival photon distribution of the photon arrival time spectrum in the time resolution information to obtain the local optical characteristic indicators of the thin film; Identifying changes in the local optical characteristic indicators, which refers to the change rate or amplitude of the local optical characteristic indicators within a preset time window exceeding a preset threshold; The impact of changes in the local optical properties on the accuracy of extracting various deep biological features from the spectral information, the spatial resolution information, and the temporal resolution information was evaluated. Based on the degree of influence, a dynamic weight is assigned to the source channel of each deep biological feature; Based on the dynamic weights, the deep biological characteristics reflected by the spectral information, the spatial resolution information, and the temporal resolution information are weighted and fused to form a description of the deep infected tissue state.

7. The method for evaluating the therapeutic efficacy of an antibacterial drug according to claim 6, wherein The step of assessing the impact of the changes on the accuracy of extracting each deep biological feature from the spectral information, the spatial resolution information, and the temporal resolution information includes: Monitor the instantaneous fluctuation amplitude of each deep biological characteristic reflected by the spectral information, the spatial resolution information and the temporal resolution information during the period of change; Calculate the cross-channel consistency deviation of the same deep biological feature reflected by the spectral information, the spatial resolution information, and the temporal resolution information during the period of change; Based on the instantaneous fluctuation amplitude and the cross-channel consistency deviation, the degree of impact of the change on the accuracy of extraction of each deep biological feature in the spectral information, the spatial resolution information and the temporal resolution information is quantified.

8. The method for evaluating the therapeutic efficacy of an antibacterial drug according to claim 6, wherein The step of continuously monitoring the ratio of optical signal intensity of receiving units at different distances in the spatially resolved information, and the distribution of early-arriving photons in the photon arrival time spectrum in the temporally resolved information, to obtain the local optical property indicators of the thin film includes: By using multiple spatially separated optical receivers, the ratio of light signal intensity captured by different optical receivers is continuously calculated to obtain the ratio of light signal intensity of receiving units at different distances in the spatial resolution information; Using a time-resolved detector, the photon arrival time spectrum is continuously analyzed to extract the number of early-arriving photons or their distribution characteristics within a preset time window, so as to obtain the distribution of early-arriving photons in the photon arrival time spectrum of the time-resolved information. Based on the ratio of optical signal intensity and the number or distribution characteristics of early-arriving photons, the local scattering and absorption effects of the thin film are comprehensively quantified to obtain the local optical properties of the thin film.

9. The method for evaluating the therapeutic efficacy of an antibacterial drug according to claim 8, wherein The method is based on the ratio of optical signal intensity and the number or distribution characteristics of early-arriving photons. The steps for comprehensively quantifying the local scattering and absorption effects of the thin film to obtain local optical property indicators of the thin film include: Based on the preset physical properties of thin-film optical transmission, a correspondence is established between the ratio of optical signal intensity and the number or distribution characteristics of early-arriving photons and the local scattering and absorption effects of the thin film. By adjusting the parameters of the local scattering and absorption effect of the thin film, the correspondence is optimally matched with the ratio of the measured optical signal intensity and the number or distribution characteristics of early-arriving photons. Based on the correspondence that achieves the best match, the local scattering and absorption effects of the thin film are comprehensively quantified to obtain the local optical properties of the thin film.

10. An antibacterial drug efficacy evaluation system for performing an antibacterial drug efficacy evaluation method according to any one of claims 1 to 9, characterized by, The system includes: The information acquisition module is used to acquire various optical information from the infected area, including spectral information, spatial resolution information, and temporal resolution information. A preprocessing module is used to preprocess the various optical information to suppress thin-film interference and extract deep biological features; The feature fusion module is used to fuse the deep biological features to form a description of the state of deep infected tissue; The verification and evaluation module is used to perform redundancy verification on the description of the deep infection tissue state and evaluate the reliability of the description of the deep infection tissue state to obtain the evaluation results of the antimicrobial drug efficacy.

Images