Diabetic foot complex wound necrotic tissue recognition method and system based on multispectral polarization imaging and microcirculation blood oxygen spectrum

Abstract

The application relates to the technical field of medical image processing, and discloses a diabetic foot complex wound necrotic tissue recognition method and system based on multispectral polarization imaging and microcirculation blood oxygen spectrum, wherein the method comprises the following steps: acquiring wound collection data and inverting exudate optical parameters; performing surface layer reflection deduction based on the exudate optical parameters; extracting a necrotic tissue feature map; and performing microcirculation blood oxygen correction and necrosis probability segmentation. Compared with the prior art which mainly depends on a single spectral parameter to judge necrosis, especially under the condition that the diabetic foot wound is covered by exudate with large changes in viscosity and component concentration, the technical problem that deep tissue reflection characteristics cannot be extracted cannot be solved. Since the exudate optical parameter inversion, Stokes polarization decomposition and dynamic waveband selection are adopted, the cooperative recognition of the necrotic tissue spectral characteristics and the microcirculation blood oxygen information under the interference of the exudate is realized, and the accuracy of the diabetic foot complex wound necrotic tissue recognition is improved.

Description

Technical Field

[0001] This invention relates to the field of medical image processing technology, and in particular to a method and system for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation blood oxygenation. Background Technology

[0002] Diabetic foot is a serious chronic complication of diabetes, characterized by complex wounds often accompanied by tissue necrosis, infection, and copious exudate. Accurate identification and segmentation of necrotic tissue are crucial for effective debridement and subsequent treatment. However, in current clinical practice, physicians primarily rely on visual observation, palpation, or simple monospectral imaging equipment for assessment.

[0003] These methods have significant shortcomings. For example, visual observation and palpation are highly subjective, heavily influenced by the physician's experience, and struggle to accurately define necrotic boundaries, especially when covered by exudate and with severe tissue edema. Traditional single-spectrum or RGB imaging techniques cannot effectively penetrate turbid exudate layers, often obtaining only surface reflection information and failing to reflect the true state of deeper tissues. Furthermore, necrotic tissue and hypoxic viable tissue caused by microcirculatory disturbances may exhibit similar appearances under ordinary imaging, leading to misdiagnosis. Existing technologies cannot fully meet the need for objective, accurate, and non-invasive automatic identification and quantitative analysis of necrotic tissue under complex wound conditions. Therefore, there is an urgent need for a technical solution that can penetrate surface interference, accurately distinguish between necrotic and hypoxic viable tissue, and provide clear demarcation boundaries even in the presence of large amounts of exudate and microcirculatory abnormalities, in order to improve the accuracy of debridement surgery and patient prognosis. Summary of the Invention

[0004] To address the aforementioned technical shortcomings, the purpose of this invention is to propose a method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation. This method aims to solve the technical problem that existing technologies often rely on a single spectral parameter for necrosis assessment, especially when the diabetic foot wound is covered with exudate of varying viscosity and concentration, making it impossible to extract deep tissue reflection features.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: The present invention provides a method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation blood oxygenation spectrum.

[0006] The method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation includes:

[0007] Step S10: Acquire wound data of complex diabetic foot wounds, and perform exudate optical parameter inversion based on the wound data to obtain an exudate optical parameter set;

[0008] Step S20: Based on the exudate optical parameter set, surface reflection is subtracted using the Stokes polarization decomposition and Fresnel reflection correction method to obtain a deep tissue polarization image set;

[0009] Step S30: Based on the deep tissue polarization image set, extract the necrosis-related reflection features using a dynamic band selection method to obtain a necrosis tissue feature map;

[0010] Step S40: Based on the necrotic tissue feature map, the Beer-Lambert reflectance compensation method is used to correct the microcirculation blood oxygenation, and the corrected blood oxygen saturation map is obtained.

[0011] Step S50: Perform necrosis probability mapping and threshold segmentation based on the corrected blood oxygen saturation map to obtain the necrotic tissue identification and segmentation results.

[0012] Preferably, step S10, which involves acquiring wound data of complex diabetic foot wounds and performing exudate optical parameter inversion based on the wound data to obtain an exudate optical parameter set, specifically includes:

[0013] Step S101: Acquire multispectral polarization images of complex diabetic foot wounds and simultaneously acquire wound exudate samples to obtain raw wound data;

[0014] Step S102: The viscosity and component concentration of the wound exudate sample are measured to obtain exudate state data;

[0015] Step S103: Calculate the effective refractive index, absorption coefficient, and scattering coefficient at each wavelength based on the exudate state data, and combine the effective refractive index, absorption coefficient, and scattering coefficient into an exudate optical parameter set.

[0016] Preferably, step S20, which involves using the Stokes polarization decomposition and Fresnel reflection correction method to subtract surface reflections based on the exudate optical parameter set to obtain a deep tissue polarization image set, specifically includes:

[0017] Step S201: Calculate the Stokes parameters of each pixel position based on the multispectral polarization image in step S10 to obtain the polarization state parameter map;

[0018] Step S202: Based on the polarization state parameter map and the exudate optical parameter set, the Fresnel reflection correction method is used to calculate the surface reflection component of the exudate to obtain the surface reflection estimation map;

[0019] Step S203: Subtract the surface reflection estimation map from the multispectral polarization image and retain the polarization response formed by deep tissue scattering to obtain a set of deep tissue polarization images.

[0020] Preferably, in step S202, the surface reflection estimation map is determined by the exudate reflection suppression weight, and the exudate reflection suppression weight satisfies:

[0021]

[0022] in, Indicates wavelength Viscosity of exudate and polarization angle The weight of exudate reflection inhibition; Indicates the basic reflection suppression weight; Indicates reference viscosity; Indicates wavelength The corresponding phase offset; This represents the wavelength correction amount determined by the scattering coefficient of the exudate.

[0023] Preferably, step S30, which involves extracting necrosis-related reflection features based on the deep tissue polarization image set using a dynamic band selection method to obtain a necrosis tissue feature map, specifically includes:

[0024] Step S301: Perform pixel-level reflectance normalization on the deep tissue polarization image set to obtain a deep tissue reflectance sequence;

[0025] Step S302: Based on the deep tissue reflectance sequence and the exudate viscosity in step S10, calculate the dynamic necrosis difference index under different band combinations to obtain the candidate band combination feature value.

[0026] Step S303: Select the maximum response value among the candidate band combination feature values ​​as the necrotic tissue feature value at the corresponding pixel position, and generate a necrotic tissue feature map;

[0027] Wherein, the dynamic necrosis difference index satisfies:

[0028]

[0029] in, y is the x-coordinate of the pixel, and y is the y-coordinate of the pixel. Indicates pixel position At the i-th and j-th wavelengths and Dynamic necrosis difference index under combination; This indicates the pixel location in the deep tissue polarization image output in step S20. At wavelength Normalized reflectance; Indicates pixel position At wavelength Normalized reflectance; Indicates the viscosity of the exudate; Indicates reference viscosity; Indicates the viscosity correction factor; This represents a constant used to prevent the denominator from being zero.

[0030] Preferably, step S40, which involves using the Beer-Lambert reflectance compensation method to correct microcirculation oxygenation based on the necrotic tissue feature map to obtain a corrected oxygen saturation map, specifically includes:

[0031] Step S401: Determine the suspected necrotic area and adjacent reference area using the necrotic tissue feature map to obtain a blood oxygen correction area map;

[0032] Step S402: Estimate the exudate coverage thickness based on the blood oxygen correction region map, and compensate the narrowband spectral reflectance by combining the absorption coefficient in step S10 to obtain the corrected reflectance map;

[0033] Step S403: Based on the corrected reflectivity map, calculate the blood oxygen saturation at each pixel location using the modified Beer-Lambert law to obtain the corrected blood oxygen saturation map.

[0034] Preferably, step S50, which involves performing necrosis probability mapping and threshold segmentation based on the corrected blood oxygen saturation map to obtain the necrotic tissue identification and segmentation results, specifically includes:

[0035] Step S501: Perform pixel-level registration between the necrotic tissue feature map and the corrected blood oxygen saturation map to obtain a fused feature map;

[0036] Step S502: Based on the fused feature map, the Sigmoid probability mapping method is used to calculate the probability that each pixel position belongs to necrotic tissue, and a necrotic tissue probability map is obtained;

[0037] Step S503: Based on the necrotic tissue probability map, the Otsu threshold segmentation method is used to determine the boundary of the necrotic tissue, and the necrotic tissue identification and segmentation results are obtained.

[0038] This invention also provides a system for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation, comprising:

[0039] The optical parameter inversion module is used to acquire wound data of complex diabetic foot wounds, and perform exudate optical parameter inversion based on the wound data to obtain an exudate optical parameter set.

[0040] The polarization reflection subtraction module is used to perform surface reflection subtraction based on the exudate optical parameter set using the Stokes polarization decomposition and Fresnel reflection correction method to obtain a deep tissue polarization image set.

[0041] The necrosis feature extraction module is used to extract necrosis-related reflection features based on the deep tissue polarization image set using a dynamic band selection method, so as to obtain a necrosis tissue feature map.

[0042] The blood oxygen correction module is used to perform microcirculation blood oxygen correction based on the necrotic tissue feature map using the Beer-Lambert reflectance compensation method to obtain a corrected blood oxygen saturation map.

[0043] The probability segmentation module is used to perform necrosis probability mapping and threshold segmentation based on the corrected blood oxygen saturation map to obtain the necrotic tissue identification and segmentation results.

[0044] The present invention also provides a device for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation. The device includes a memory, a processor, and a program for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation stored in the memory and executable on the processor. When the program for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation is executed by the processor, the above method is implemented.

[0045] The present invention also provides a computer program product, the computer program product including a program for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation spectrum, wherein the program for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation spectrum implements the above method when executed by a processor.

[0046] The beneficial effects of this invention are as follows:

[0047] 1. This invention integrates multispectral polarization imaging technology with microcirculation blood oxygenation analysis to construct a complete technical chain from surface interference subtraction to deep feature extraction and then to blood oxygenation artifact correction. First, by utilizing Stokes polarization decomposition and combining it with Fresnel reflection correction based on the optical parameters of exudate, the strong specular reflection and scattering interference generated by wound exudate are effectively suppressed, allowing the true polarization scattering information of deep tissues to be highlighted, providing a high-quality data foundation for subsequent feature extraction.

[0048] 2. This invention further utilizes a dynamic band selection mechanism to adaptively extract the most sensitive spectral features for necrotic tissue based on the state of exudate and differences in tissue reflectance at different bands, generating a necrotic tissue feature map. Subsequently, addressing the microcirculatory disturbances often associated with necrotic areas, the Beer-Lambert reflectance compensation method is employed to correct blood oxygen saturation, eliminating artifacts that might be misjudged as necrosis due to low blood oxygen content. Finally, by fusing optical features with the corrected blood oxygen information through probabilistic mapping, high-precision identification and automatic segmentation of necrotic tissue are achieved. This method significantly improves the objectivity, accuracy, and robustness of necrotic tissue identification in the complex wound environment of diabetic foot. Attached Figure Description

[0049] Figure 1 This is a flowchart illustrating the first embodiment of the method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation maps according to the present invention.

[0050] Figure 2 This is a schematic diagram of the original wound multi-channel acquisition matrix, representing the first embodiment of the method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation blood oxygenation spectrum according to the present invention.

[0051] Figure 3 This is a schematic diagram of high-quality image fragment extraction after alignment, based on the first embodiment of the method for identifying necrotic tissue in complex diabetic foot wounds according to multispectral polarization imaging and microcirculation blood oxygenation spectrum of the present invention.

[0052] Figure 4 This is a schematic diagram of the sampling segment ingress threshold screening curve of the first embodiment of the method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation blood oxygenation spectrum of the present invention.

[0053] Figure 5 This is a schematic diagram of the wound feature response distribution under unadjusted color gamut in the first embodiment of the method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation blood oxygenation spectrum of the present invention.

[0054] Figure 6 This is a schematic diagram of the extraction of candidate necrotic tissue after purification, based on the first embodiment of the method for identifying necrotic tissue in complex diabetic foot wounds according to multispectral polarization imaging and microcirculation blood oxygenation spectrum of the present invention.

[0055] Figure 7 This is a schematic diagram of strong anomaly display under normalized fusion features in the first embodiment of the method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation blood oxygenation spectrum of the present invention.

[0056] Figure 8This is an illustration of the enhanced display of weakly abnormal regions under logarithmic normalization in the first embodiment of the present invention, which is based on multispectral polarization imaging and microcirculation blood oxygenation spectrum for the identification of necrotic tissue in complex diabetic foot wounds.

[0057] Figure 9 This is a schematic diagram illustrating the generation of an abnormality level mask for necrotic tissue in complex diabetic foot wounds, based on a first embodiment of the present invention, which utilizes multispectral polarization imaging and microcirculation oxygenation maps.

[0058] Figure 10 This is a schematic diagram of the residual distribution of the model under normalization in the first embodiment of the method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation blood oxygenation spectrum of the present invention.

[0059] Figure 11 This is a schematic diagram of small residual enhancement under symmetric log normalization in the first embodiment of the method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation blood oxygenation spectrum of the present invention. Detailed Implementation

[0060] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0061] Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0062] Example 1: As Figure 1 The diagram shown is a flowchart of the first embodiment of the method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation spectrum according to the present invention. The first embodiment of the method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation spectrum according to the present invention is presented.

[0063] In the first embodiment, the method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation includes:

[0064] Step S10: Acquire wound data of complex diabetic foot wounds, and perform exudate optical parameter inversion based on the wound data to obtain an exudate optical parameter set;

[0065] The "wound acquisition data" in this step specifically refers to the multispectral polarization image sequence of the diabetic foot wound acquired simultaneously using a multispectral polarization imaging device, as well as wound exudate samples obtained through auxiliary means (such as micro-sampling). "Exudate optical parameter inversion" refers to calculating the key optical characteristic parameters of the exudate at different imaging wavelengths, including effective refractive index, absorption coefficient, and scattering coefficient, using an optical model based on laboratory analysis of the exudate samples (such as viscosity measurement and component concentration detection). This "exudate optical parameter set" constitutes the quantitative basis for accurate optical modeling and interference subtraction in subsequent steps.

[0066] This step provides crucial prior physical parameters for the entire method. By acquiring and reversing the set of optical parameters of the exudate, subsequent optical correction and feature extraction processes are no longer based on experience or assumptions, but rather on a quantitative physical model reflecting the actual state of the current wound. This ensures the physical interpretability of the entire identification process and its adaptability to different patients and different wound conditions.

[0067] Compared to traditional methods that rely solely on raw image data acquired by imaging equipment or fixed, universal tissue optical parameters, this invention achieves personalized modeling of the surface media properties of a wound by simultaneously acquiring and inverting the optical parameters of the exudate from a specific wound. This effectively overcomes the problem of inaccurate universal model correction caused by significant individual differences in the composition, concentration, and viscosity of the exudate, laying a solid foundation for subsequent high-precision reflection subtraction and deep information extraction. For example, in a diabetic foot patient, the wound exudate may exhibit different viscosities and protein contents depending on the degree of infection. Through this step, the viscosity η of the patient's exudate sample can be measured to be relatively high, and the scattering coefficients at wavelengths of 660nm and 850nm can be calculated to be significantly different. This personalized parameter set will be used in the next step to guide the algorithm on how to effectively subtract surface reflection light for this high-viscosity, high-scattering exudate, thereby avoiding under- or over-correction caused by using universal parameters.

[0068] For example, such as Figure 2 As shown in the figure, the horizontal axis can represent a sampling time window or a spatial sliding window, and the vertical axis can represent different spectral channels, different polarization channels, or combined channels. The color intensity represents the response intensity of the corresponding channel within the corresponding window. Figure 2It can be seen that the original acquisition matrix contains multiple high-response and low-response regions, indicating that the response of complex wounds is not uniform across different channels. However, these high-response regions cannot be directly identified as necrotic tissue areas, nor can low-response regions be directly identified as normal tissue areas. For example, a relatively smooth exudate surface may create strong specular reflection, causing the area to exhibit a high response; a high blood content in the exudate may enhance absorption in specific wavelength bands, causing the area to exhibit a low response. Figure 2 This indicates that the original acquisition matrix contains a large number of optical disturbances that are unrelated to necrosis but can affect the judgment. Therefore, it is necessary to establish an optical parameter set for the exudate in step S10 to provide a basis for subsequently removing surface interference from the mixed signal.

[0069] For example, the selected area represents the effective analysis region after channel alignment, sampling window alignment, or local quality screening. Since multispectral polarization imaging typically involves multiple acquisitions at different wavelengths and polarization angles, slight local movement of the wound, changes in probe angle, patient foot tremors, or exudate flow can all cause spatial offsets between channels. Without alignment, pixels corresponding to the same tissue location at different bands may be misaligned, leading to the problem of subtracting the reflection component from one location to another when calculating Stokes parameters or subtracting surface reflection. Figure 3 The high-quality fragment extraction shown can improve the positional consistency between data from different channels, enabling subsequent polarization decomposition and reflection subtraction processes to be based on stable pixel correspondences.

[0070] For example, the solid line represents the average quality, average response, or correlation index of the segments within the continuous sampling window, while the dashed line represents the ingress threshold. Segments above the ingress threshold are considered valid segments for subsequent processing, while segments below the ingress threshold are considered significantly affected by image blur, local occlusion, acquisition jitter, excessive reflection, or low signal-to-noise ratio. This screening process removes low-quality data that is detrimental to the inversion of exudate optical parameters, making the exudate optical parameter set more reflective of the actual wound condition. Therefore, Figure 4 This demonstrates that this step involves not only simple image acquisition but also constraints on the reliability of the acquired data, thereby improving the stability of subsequent recognition processes.

[0071] Step S20: Based on the exudate optical parameter set, surface reflection is subtracted using the Stokes polarization decomposition and Fresnel reflection correction method to obtain a deep tissue polarization image set;

[0072] In this step, "Stokes polarization decomposition" refers to processing the multispectral polarization image acquired in step S10 to calculate the Stokes parameters (S0, S1, S2) for each pixel, thereby fully describing the polarization state of the light at that point (such as linear polarization degree and polarization angle). "Fresnel reflection correction" is a reflection calculation based on a physical optics model. Here, it specifically refers to using the set of optical parameters of the exudate obtained in step S10 (especially the effective refractive index), combined with the Fresnel formula and polarization information, to quantitatively estimate the specular reflection intensity component generated by the wound exudate-air interface, i.e., the "surface reflection estimation map". The "deep tissue polarization image set" is the image set obtained by subtracting the surface reflection estimation value at the corresponding position from each pixel of the original multispectral polarization image. It mainly retains the polarized light component that returns after light penetrates the exudate layer, is scattered with deep tissues (such as dermis and subcutaneous tissue), and carries tissue information.

[0073] This step effectively removes strong reflection interference from the surface of the wound. By combining physical models and measured parameters, specular reflection light mixed in with the original signal that does not carry information about deep tissue is separated and subtracted, allowing subsequent processing to focus on deep scattered light signals that truly reflect the pathological state of the tissue (such as necrosis), thus improving the signal-to-noise ratio and the reliability of feature extraction.

[0074] Compared to existing technologies that commonly use simple polarization difference or fixed threshold-based reflection subtraction methods, this invention incorporates personalized optical parameters of the exudate into the Fresnel reflection model, resulting in more accurate estimation of surface reflection. In particular, by introducing an "exudate reflection suppression weight k" related to wavelength λ, viscosity η, and polarization angle θ, the subtraction intensity can be dynamically adjusted to adapt to different exudate states and imaging geometries. This allows for a more thorough and adaptive elimination of surface interference, avoiding the problems of insufficient subtraction (residual artifacts) or excessive subtraction (loss of useful signal) under complex conditions common in traditional methods.

[0075] For example, in the central region of a wound, due to the thick accumulation of exudate and the relatively smooth surface, strong specular reflections may occur. Traditional methods might smooth this area simply through image processing filtering, but this would blur details. This invention, however, calculates and determines that at a specific wavelength and imaging angle in this region, the exudate reflection suppression weight k value is high, indicating a significant contribution from surface reflection. Based on this, the algorithm generates a high-brightness "surface reflection estimation map" and accurately subtracts it from the original image. After subtraction, the deep tissue texture (potentially containing features of necrotic tissue) that was originally obscured by strong reflections in this area becomes visible.

[0076] For example, such as Figure 5 As shown, Figure 5Although local high-response areas exist, the boundary between these areas and the background is unstable, and some weakly anomalous areas are easily masked by surrounding textures and low-frequency background responses. This figure illustrates that, without sufficient surface reflection subtraction and feature enhancement, necrotic tissue, granulation tissue, fibrous exudate, and hemorrhagic exudate in wound images are prone to response aliasing. If directly based on... Figure 5 The results shown may lead to the misidentification of necrotic areas by identifying reflective areas of exudate, or the failure to detect early necrotic tissue due to insufficient contrast in weak abnormal areas.

[0077] Step S30: Based on the deep tissue polarization image set, extract the necrosis-related reflection features using a dynamic band selection method to obtain a necrosis tissue feature map;

[0078] The "deep tissue polarization image set" in this step is the output of step S20. It is multi-band image data that has had surface reflection subtracted and mainly reflects the scattering characteristics of deep tissue. The "dynamic band selection method" means that instead of using a fixed set of two or more bands, the "dynamic necrosis difference index (DNI)" is dynamically calculated for all possible band combinations (λi, λj) based on the normalized reflectance value of each pixel in different bands and the exudate viscosity information obtained in step S10. This index comprehensively considers the influence of the difference in reflectance between the two bands (reflecting tissue composition) and the viscosity of the exudate. Each pixel value in the "necrotic tissue feature map" is determined by selecting the maximum DNI value calculated from all candidate band combinations. This means that for regions in different locations and states in the image, the algorithm automatically selects the optimal band combination that best highlights their necrosis characteristics for representation.

[0079] This step enables intelligent extraction of specific spectral reflectance features of necrotic tissue. Through a dynamic band selection mechanism, this method adaptively mines the information most relevant to necrosis from multispectral data, generating a high-contrast feature map in which the feature values ​​of necrotic areas are significantly enhanced, while the feature values ​​of normal or edematous areas are relatively suppressed, thus providing clear clues for subsequent identification.

[0080] Compared to traditional multispectral analysis methods that use a fixed set of preset bands (such as hemoglobin characteristic bands) for ratio or difference calculations, the dynamic selection mechanism of this invention has stronger adaptability and specificity. It does not rely on prior, fixed assumptions about the spectral curves of necrotic tissue, but rather uses a data-driven approach to find the band combination that best distinguishes each pixel from its surrounding tissue or background. Simultaneously, the viscosity correction term introduced in the formula allows the feature extraction process to be fine-tuned according to the physical state of the exudate, improving the robustness of the method under wound environments with varying viscosity.

[0081] For example, in pixel region A, the reflectance difference of deep tissue is greatest at wavelengths λ1=550nm and λ2=650nm. Combined with the high viscosity η of the current wound surface, the calculated DNI value is the highest. Therefore, the necrotic tissue characteristic value of this pixel is determined by this wavelength combination. In the adjacent region B, the DNI value may be highest at the combination of wavelengths λ3=750nm and λ4=850nm. This means that for potential necrotic regions with different spectral responses due to depth, composition, or blood oxygenation, this method can find its most specific spectral feature expression and uniformly map them onto the feature map to form a coherent suspected necrotic region.

[0082] For example, such as Figure 6 As shown, the highlighted areas and contour areas represent suspected necrotic areas formed after surface reflection subtraction, dynamic band selection, and local response enhancement. Compared to Figure 5 , Figure 6 The abnormal responses are more concentrated, the background area is suppressed, and the boundaries of the candidate region are clearer. Figure 6 This demonstrates that by executing steps S20 and S30 consecutively, more stable candidate necrosis regions can be extracted from complex images affected by exudate coverage and surface reflection interference. These candidate regions are not the final necrosis determination results, but rather provide a spatial prior for blood oxygen correction in step S40, enabling subsequent blood oxygen saturation calculations to focus on suspected necrosis regions and their adjacent reference regions.

[0083] Step S40: Based on the necrotic tissue feature map, the Beer-Lambert reflectance compensation method is used to correct the microcirculation blood oxygenation, and the corrected blood oxygen saturation map is obtained.

[0084] The "necrotic tissue feature map" in this step comes from step S30, which initially identifies areas suspected of necrosis. The "Beer-Lambert reflectance compensation method" is a law in optics describing the attenuation of light in an absorbing medium. When applied here, it targets the "suspected necrotic area" and its "adjacent reference area" (usually a non-necrotic but adjacent area on the feature map is selected for comparison). First, the thickness of the exudate layer covering this area is estimated based on contextual information. Then, combined with the "absorption coefficient" of the exudate in the corresponding wavelength band obtained in step S10, the narrow-band spectral reflectance of the deep tissue obtained from step S20 is compensated to eliminate the influence of the exudate layer itself on light absorption, thus obtaining a "corrected reflectance map" of the tissue that more closely approximates the "no exudate coverage" assumption. Finally, based on the corrected reflectance, the relative concentrations of oxyhemoglobin and deoxyhemoglobin in the tissue are calculated using the modified Beer-Lambert law, thereby obtaining the "corrected oxygen saturation map."

[0085] The core technical effect of this step is the correction of microcirculatory oxygenation information. Necrotic tissue is usually accompanied by microvascular occlusion and interruption of oxygen supply, resulting in extremely low oxygen saturation. However, severe tissue edema or simple microcirculatory disturbances can also lead to localized hypoxia, forming a low-signal area (artifact) resembling necrosis on the oximetry map. This step, by compensating for the influence of exudate absorption and accurately calculating oxygen saturation, aims to obtain information that more accurately reflects the tissue's microcirculatory oxygenation status, thereby distinguishing true necrosis (high value on the characteristic map + extremely low value on the oximetry map) from simply hypoxic living tissue (potentially low value on the characteristic map + low value on the oximetry map).

[0086] Compared to traditional methods that directly apply a blood oxygenation calculation model to the raw reflectance image or use only a single blood oxygenation threshold for judgment, this invention adds a reflectance compensation step based on the optical parameters of the exudate before blood oxygenation calculation. This takes into account the additional absorption of specific wavelengths of light by the exudate layer (especially exudate containing blood components), avoiding the artificial underestimation of blood oxygen saturation values ​​due to exudate absorption. This reduces the possibility of misjudging living tissue covered with a thick layer of bloody exudate as necrosis, and improves the reliability of blood oxygenation parameters as a basis for judgment.

[0087] For example, in a suspected necrotic area identified by the feature map, the original deep reflectance is low in the hemoglobin absorption peak band (e.g., 660nm). Without compensation, direct calculation might show extremely low blood oxygen saturation in this area, easily leading to a diagnosis of necrosis. However, through this step, the algorithm estimates that the exudate in this area is relatively thick, and that the exudate also has some absorption at 660nm. Through compensation calculation, after deducting the influence of exudate absorption, it is found that the corrected reflectance of this area is not as low as expected. Although the recalculated blood oxygen saturation is still low, it does not reach the necrosis threshold. Combined with the fact that its necrotic tissue characteristic values ​​may also be low, it can ultimately be excluded from the necrotic area, thus avoiding a false diagnosis.

[0088] Step S50: Perform necrosis probability mapping and threshold segmentation based on the corrected blood oxygen saturation map to obtain the necrotic tissue identification and segmentation results.

[0089] The "corrected blood oxygen saturation map" in this step comes from step S40 and reflects the corrected tissue microcirculation oxygenation level; the "necrotic tissue feature map" comes from step S30 and reflects the probability of necrosis based on spectral reflectance characteristics. "Pixel-level registration" ensures that the two images are perfectly aligned in space. The "fusion feature map" combines the information from these two dimensions at the pixel level (e.g., by stitching or weighting) to form a multi-dimensional feature vector for each pixel. The "Sigmoid probability mapping method" is a method that maps the fused feature values ​​to probability values ​​between 0 and 1 using the Sigmoid function; here, it is used to calculate the "probability of each pixel belonging to necrotic tissue," outputting a "necrotic tissue probability map." The "Otsu thresholding method" is an image segmentation algorithm that automatically calculates the optimal threshold to distinguish between foreground (necrotic) and background (non-necrotic). After being applied to the probability map, a clear "necrotic tissue identification and segmentation result" is obtained, i.e., a binarized mask image, where the white area represents the identified necrotic tissue.

[0090] This step enables the final decision-making and visualization output. By fusing optical reflectance features and physiological blood oxygenation information, and using a probabilistic model for comprehensive judgment, this method not only provides a binary judgment of whether each location is necrotic, but also gives its confidence level (probability). Otsu threshold segmentation automatically determines the segmentation boundary based on the probability distribution, ultimately outputting an intuitive, quantitative, and directly applicable segmentation map of necrotic tissue regions that can be used to guide debridement, completing the transformation from multimodal data to clinical decision-making information.

[0091] Compared to simple threshold segmentation relying on a single feature map or methods that depend on manual sketching by doctors, this invention uses a probabilistic fusion model to integrate spectral features reflecting tissue morphology / composition and blood oxygenation features reflecting tissue metabolism / function, making the decision-making basis more comprehensive and reliable. The Sigmoid mapping transforms continuous features into probabilities, which better aligns with the concept of uncertainty in medical diagnosis. The automatic threshold determination using the Otsu method ensures the objectivity and repeatability of the segmentation results, avoiding subjective biases from manually setting thresholds, ultimately achieving automated, high-precision identification and quantitative segmentation of necrotic tissue.

[0092] For example, in a certain area at the edge of a wound, the necrotic tissue characteristic value is high (suggesting possible necrosis), but the corrected blood oxygen saturation is not extremely low (suggesting that there may still be a weak blood supply). Traditional single threshold methods may struggle to handle this contradictory situation. This step, however, integrates both pieces of information through fusion and probability mapping: hyperspectral features increase the probability of necrosis, while extremely low blood oxygen values ​​appropriately decrease this probability. The final calculated probability of necrosis may be a moderate value (e.g., 0.6). On the probability map, this area appears as a medium grayscale. The Otsu algorithm automatically determines an optimal threshold (e.g., 0.55) based on the grayscale distribution of the entire probability map. Since the probability of 0.6 for this area is higher than the threshold of 0.55, it is still segmented as necrotic tissue. This process simulates the thinking of a doctor who integrates multiple signs to make a judgment, but it is more quantitative and consistent.

[0093] For example, such as Figure 7 As shown in the figure, strong anomaly regions occupy the majority of the dynamic range, while weak anomaly regions are not clearly displayed. This figure illustrates that in fused features, regions with different degrees of anomaly can have large numerical ranges. If ordinary normalization is used, strong necrosis regions will be significantly highlighted, while weak anomaly regions, such as necrosis edges, early necrosis areas, or those partially obscured by exudate, may be compressed into a narrower display range. Therefore, when performing necrosis probability mapping, it is necessary to consider the dynamic range distribution of the fused features to avoid strong anomaly regions becoming too prominent and obscuring weak anomaly regions.

[0094] like Figure 8 As shown, with Figure 7 compared to, Figure 8 Not only does it preserve the central region of strong abnormalities, but it also reveals surrounding transitional and weakly abnormal regions through contour lines and brightness variations. This figure demonstrates that logarithmic normalization or similar dynamic range compression can improve the visibility of weakly abnormal regions without weakening the ability to identify strong abnormalities. For diabetic foot wounds, weakly abnormal regions often correspond to necrotic edges, areas of decreased tissue activity, or the forefront of necrotic expansion; these regions are particularly important for determining debridement boundaries.

[0095] like Figure 9 As shown, the black areas represent abnormal tissue regions obtained based on fusion features and probability mapping, while the boxed areas represent local areas of focus or boundary correction areas. This figure illustrates that this step ultimately transforms a continuous probability map into an intuitive necrotic tissue mask, clearly expressing the location, shape, and area of ​​the necrotic region. Compared to doctors manually drawing boundaries based solely on visual experience, the anomaly level mask offers higher consistency and repeatability; compared to simply outputting a heatmap, the mask results are more convenient for area statistics, boundary marking, and debridement path planning.

[0096] like Figure 10As shown, the red and blue areas represent the positive and negative residuals, respectively, and the residual magnitude reflects the degree of deviation between the processed result and the expected corrected state. It is understandable that after surface reflection subtraction, dynamic band selection, and blood oxygen compensation, some local areas may still experience undercompensation or overcompensation. For example, if the estimated exudate coverage thickness in a certain area is too low, blood oxygen compensation may be insufficient; if the reflection subtraction weight is too high, it may weaken some effective deep tissue signals.

[0097] For example, such as Figure 11 As shown, Figure 11 This diagram illustrates the enhancement of small residuals under symmetric logarithmic normalization. Regions with weak residuals are further magnified, and the selected areas represent local areas that are not easily noticeable in ordinary residual maps but may still affect boundary judgment. It should be noted that small residuals are not necessarily invalid noise. In complex diabetic foot wounds, small residuals often appear at necrotic edges, areas of abrupt changes in exudate thickness, areas of transitional tissue texture, or areas of gradual changes in blood oxygenation. Enhancing small residuals through symmetric logarithmic normalization allows for clearer observation of local deviations at boundaries, providing a more definitive basis for subsequent smoothing, correction, or manual verification of segmentation boundaries.

[0098] Example 2: Furthermore, the necrotic tissue identification system for complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation mapping provided by this invention employs the necrotic tissue identification method for complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation mapping described in the above embodiments, and can solve the technical problem of identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation mapping. The beneficial effects of the necrotic tissue identification system for complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation mapping provided by this invention are the same as those of the necrotic tissue identification method for complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation mapping provided in the above embodiments, and other technical features of the necrotic tissue identification system for complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation mapping are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0099] Example 3: This invention provides a device for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation mapping. The device includes: at least one processor; and a memory communicatively connected to the at least one processor. The memory stores instructions executable by the at least one processor, which are executed by the at least one processor to enable the at least one processor to perform the method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation mapping described in Example 1 above. The device for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation mapping in this invention can include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. The device for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation is merely an example and should not limit the functionality or scope of the embodiments of the present invention. The device may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) that can perform various appropriate actions and processes according to a program stored in a read-only memory or a program loaded from a storage device into a random access memory. The random access memory also stores various programs and data required for the operation of the device. The processing unit, read-only memory, and random access memory are interconnected via a bus. An I / O interface is also connected to the bus. Typically, the following systems can be connected to the I / O interface: input devices including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices including, for example, magnetic tapes, hard disks, etc.; and communication devices. The communication device allows the identification device for necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation to communicate wirelessly or wiredly with other devices to exchange data. Although the figures show identification devices for necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation with various systems, it should be understood that implementation or possession of all the systems shown is not required.It can be implemented alternatively or with more or fewer systems.

[0100] Example 4: This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation mapping as described above. The computer program product provided by this invention can solve the technical problem of identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation mapping. Compared with the prior art, the beneficial effects of the computer program product provided by this invention are the same as those of the method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation mapping provided in the above embodiments, and will not be repeated here.

[0101] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device, or installed from a read-only memory. When the computer program is executed by a processing device, it performs the functions defined in the methods of the embodiments disclosed in this invention.

[0102] It should be understood that the various parts disclosed in this invention can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0103] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the present invention and its equivalents, the present invention also intends to include these modifications and variations.

Claims

1. A method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation, characterized in that, The methods include: Step S10: Acquire wound data of complex diabetic foot wounds, and perform exudate optical parameter inversion based on the wound data to obtain an exudate optical parameter set; Step S20: Based on the exudate optical parameter set, surface reflection is subtracted using the Stokes polarization decomposition and Fresnel reflection correction method to obtain a deep tissue polarization image set; Step S30: Based on the deep tissue polarization image set, extract the necrosis-related reflection features using a dynamic band selection method to obtain a necrosis tissue feature map; Step S40: Based on the necrotic tissue feature map, the Beer-Lambert reflectance compensation method is used to correct the microcirculation blood oxygenation, and the corrected blood oxygen saturation map is obtained. Step S50: Perform necrosis probability mapping and threshold segmentation based on the corrected blood oxygen saturation map to obtain the necrotic tissue identification and segmentation results.

2. The method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation as described in claim 1, characterized in that, Step S10, which involves acquiring wound data of complex diabetic foot wounds and performing exudate optical parameter inversion based on the wound data to obtain an exudate optical parameter set, specifically includes: Step S101: Acquire multispectral polarization images of complex diabetic foot wounds and simultaneously acquire wound exudate samples to obtain raw wound data; Step S102: The viscosity and component concentration of the wound exudate sample are measured to obtain exudate state data; Step S103: Calculate the effective refractive index, absorption coefficient, and scattering coefficient at each wavelength based on the exudate state data, and combine the effective refractive index, absorption coefficient, and scattering coefficient into an exudate optical parameter set.

3. The method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation as described in claim 1, characterized in that, Step S20, which involves using the Stokes polarization decomposition and Fresnel reflection correction method to subtract surface reflections based on the exudate optical parameter set to obtain a deep tissue polarization image set, specifically includes: Step S201: Calculate the Stokes parameters of each pixel position based on the multispectral polarization image in step S10 to obtain the polarization state parameter map; Step S202: Based on the polarization state parameter map and the exudate optical parameter set, the Fresnel reflection correction method is used to calculate the surface reflection component of the exudate to obtain the surface reflection estimation map; Step S203: Subtract the surface reflection estimation map from the multispectral polarization image and retain the polarization response formed by deep tissue scattering to obtain a set of deep tissue polarization images.

4. The method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation as described in claim 3, characterized in that, In step S202, the surface reflection estimation map is determined by the exudate reflection suppression weight, which satisfies the following: in, Indicates wavelength Viscosity of exudate and polarization angle The weight of exudate reflection inhibition; Indicates the basic reflection suppression weight; Indicates reference viscosity; Indicates wavelength The corresponding phase offset; This represents the wavelength correction amount determined by the scattering coefficient of the exudate.

5. The method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation as described in claim 1, characterized in that, Step S30, which involves extracting necrosis-related reflection features based on the deep tissue polarization image set using a dynamic band selection method to obtain a necrosis tissue feature map, specifically includes: Step S301: Perform pixel-level reflectance normalization on the deep tissue polarization image set to obtain a deep tissue reflectance sequence; Step S302: Based on the deep tissue reflectance sequence and the exudate viscosity in step S10, calculate the dynamic necrosis difference index under different band combinations to obtain the candidate band combination feature value. Step S303: Select the maximum response value among the candidate band combination feature values ​​as the necrotic tissue feature value at the corresponding pixel location, and generate a necrotic tissue feature map; Wherein, the dynamic necrosis difference index satisfies: in, y is the x-coordinate of the pixel, and y is the y-coordinate of the pixel. Indicates pixel position At the i-th and j-th wavelengths and Dynamic necrosis difference index under combination; This indicates the pixel location in the deep tissue polarization image output in step S20. At wavelength Normalized reflectance; Indicates pixel position At wavelength Normalized reflectance; Indicates the viscosity of the exudate; Indicates reference viscosity; Indicates the viscosity correction factor; This represents a constant used to prevent the denominator from being zero.

6. The method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation as described in claim 1, characterized in that, Step S40, which involves using the Beer-Lambert reflectance compensation method to correct microcirculation oxygenation based on the necrotic tissue feature map to obtain the corrected oxygen saturation map, specifically includes: Step S401: Determine the suspected necrotic area and adjacent reference area using the necrotic tissue feature map to obtain a blood oxygen correction area map; Step S402: Estimate the exudate coverage thickness based on the blood oxygen correction region map, and compensate the narrowband spectral reflectance by combining the absorption coefficient in step S10 to obtain the corrected reflectance map; Step S403: Based on the corrected reflectivity map, calculate the blood oxygen saturation at each pixel location using the modified Beer-Lambert law to obtain the corrected blood oxygen saturation map.

7. The method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation as described in claim 1, characterized in that, Step S50, which involves performing necrosis probability mapping and threshold segmentation based on the corrected blood oxygen saturation map to obtain the necrotic tissue identification and segmentation results, specifically includes: Step S501: Perform pixel-level registration between the necrotic tissue feature map and the corrected blood oxygen saturation map to obtain a fused feature map; Step S502: Based on the fused feature map, the Sigmoid probability mapping method is used to calculate the probability that each pixel position belongs to necrotic tissue, and a necrotic tissue probability map is obtained; Step S503: Based on the necrotic tissue probability map, the Otsu threshold segmentation method is used to determine the boundary of the necrotic tissue, and the necrotic tissue identification and segmentation results are obtained.

8. A system for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation, applied to the method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation as described in any one of claims 1 to 7, characterized in that, The system for identifying necrotic tissue in complex diabetic foot wounds includes: The optical parameter inversion module is used to acquire wound data of complex diabetic foot wounds, and to perform exudate optical parameter inversion based on the wound data to obtain an exudate optical parameter set. The polarization reflection subtraction module is used to perform surface reflection subtraction based on the exudate optical parameter set using the Stokes polarization decomposition and Fresnel reflection correction method to obtain a deep tissue polarization image set. The necrosis feature extraction module is used to extract necrosis-related reflection features based on the deep tissue polarization image set using a dynamic band selection method to obtain a necrosis tissue feature map. The blood oxygen correction module is used to perform microcirculation blood oxygen correction based on the necrotic tissue feature map using the Beer-Lambert reflectance compensation method to obtain the corrected blood oxygen saturation map. The probability segmentation module is used to perform necrosis probability mapping and threshold segmentation based on the corrected blood oxygen saturation map to obtain the necrotic tissue identification and segmentation results.

9. A device for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation, characterized in that, The device for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation includes: a memory, a processor, and a program for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation stored in the memory and executable on the processor. When the program for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation is executed by the processor, it implements the method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation as described in any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a program for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation. When the program for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation is executed by a processor, it implements the method for identifying necrotic tissue in complex diabetic foot wounds based on multispectral polarization imaging and microcirculation oxygenation as described in any one of claims 1 to 7.

Images