A living body detection method and system based on multi-modal imaging integration analysis

Abstract

The application provides a kind of based on multi-modal imaging integration analysis living body detection method and system, it is related to multi-modal fusion technical field, the application obtains the multi-modal imaging data of multiple organs of living object;According to multi-modal imaging data, respectively calculate fluorescence intensity distribution characteristics, tissue density value characteristics and proton density signal characteristics;Fluorescence intensity distribution characteristics and proton density signal characteristics are respectively associated with tissue density value characteristics and are processed cross-modality, obtain first association relationship, and second association relationship;First association relationship and second association relationship are fused with parameters, obtain cross-modality feature matching data set, and it is parameter calibrated, obtain calibrated cross-modality feature matching data set;The cross-modality characteristics of each organ are analyzed, and the living body detection result is obtained;Realize the collaborative processing of the physiological and structural characteristics of multiple organs of living object, improve the accuracy and reliability of cross-modality biological feature recognition.

Description

Technical Field

[0001] This application relates to the field of multimodal fusion technology, and in particular to a liveness detection method and system based on multimodal imaging integration analysis. Background Technology

[0002] In modern medical and biological research, there is a growing demand for dynamic functional analysis of multiple organs in living subjects. To accurately assess the synergistic effects between different organs and their response mechanisms to changes in the internal and external environment, a method capable of integrating information from multiple imaging modalities is urgently needed to provide comprehensive and accurate data support.

[0003] Currently, mainstream solutions integrate data collected by multiple imaging devices and perform preliminary fusion of this information from different modalities to generate comprehensive image data. However, existing solutions have some significant drawbacks. For example, due to the lack of an effective cross-modal feature matching mechanism, it is difficult to achieve true synchronization and alignment of data between different imaging modalities, affecting the accuracy of subsequent analysis results. In application scenarios with high real-time requirements, existing data fusion strategies cannot guarantee sufficient processing speed, resulting in time delays. Summary of the Invention

[0004] The purpose of this application is to provide a liveness detection method and system based on multimodal imaging integration analysis, in order to solve the problems in the prior art, such as difficulty in achieving synchronization and alignment between different modal data, insufficient accuracy of analysis results, and insufficient real-time processing speed of data fusion strategies.

[0005] To address the aforementioned technical problems, in a first aspect, this application provides a liveness detection method based on multimodal imaging integrated analysis, comprising:

[0006] Acquire multimodal imaging data of multiple organs of a living subject, including original fluorescence imaging images, original tissue density imaging images, and original proton density imaging images;

[0007] Based on the original fluorescence imaging image, the original tissue density imaging image, and the original proton density imaging image, the fluorescence intensity distribution characteristics, tissue density value characteristics, and proton density signal characteristics are calculated respectively.

[0008] The fluorescence intensity distribution feature and proton density signal feature are respectively subjected to cross-modal correlation processing with the tissue density value feature to obtain the first correlation relationship between the fluorescence intensity distribution feature and the tissue density value feature, and the second correlation relationship between the proton density signal feature and the tissue density value feature;

[0009] The first and second association relationships are fused to obtain a cross-modal feature matching dataset, and the cross-modal feature matching dataset is then calibrated to obtain a calibrated cross-modal feature matching dataset.

[0010] The cross-modal features of each organ in the calibrated cross-modal feature matching dataset are analyzed in a linked manner to obtain the liveness detection results.

[0011] Optionally, based on the original fluorescence imaging image, the original tissue density imaging image, and the original proton density imaging image, fluorescence intensity distribution characteristics, tissue density value characteristics, and proton density signal characteristics are calculated, including:

[0012] In the original fluorescence imaging image, the organ regions where each organ is located are marked, the fluorescence signal intensity values ​​of the pixels in each organ region are counted, and the variation range, distribution range, and intensity difference between pixels of each fluorescence signal intensity value are calculated. The variation range, distribution range, and intensity difference corresponding to each organ are summarized to obtain the fluorescence intensity distribution characteristics.

[0013] The original tissue density imaging image is subjected to tomographic correlation processing to determine the tomographic regions of each organ to obtain the tomographic region set of each organ. The average density, extreme density, and density dispersion of each tomographic region set are calculated. The average density, extreme density, and density dispersion of each organ are summarized to obtain the tissue density value features.

[0014] The proton signal regions of each organ are marked in the original proton density imaging image, and the average signal intensity, peak signal intensity, and signal coverage of each proton signal region are calculated. The average signal intensity, peak signal intensity, and signal coverage of each organ are then summarized to obtain the proton density signal features.

[0015] Optionally, the fluorescence intensity distribution feature and the proton density signal feature are respectively subjected to cross-modal correlation processing with the tissue density value feature to obtain a first correlation relationship between the fluorescence intensity distribution feature and the tissue density value feature, and a second correlation relationship between the proton density signal feature and the tissue density value feature, including:

[0016] Based on the organ region and tomographic region set, spatial coordinate matching is performed on the original fluorescence imaging image and the original tissue density imaging image. Under the matched spatial coordinates, the variation range, distribution range, and intensity difference between each pixel in the fluorescence intensity distribution feature of the same organ are respectively correlated with the average density, extreme density, and density dispersion in the tissue density feature. This yields the first fluorescence density coordinate parameter pair, the second fluorescence density coordinate parameter pair, and the third fluorescence density coordinate parameter pair for each organ, forming a first correlation relationship that includes spatial coordinate information and parameter correlation coefficients.

[0017] Based on the proton signal region and the tomographic region set, spatial coordinate calibration is performed on the original proton density imaging image and the original tissue density imaging image. Within the calibrated spatial range, the average signal intensity, peak signal intensity, and signal coverage of the same organ in the proton density signal features are respectively correlated at the regional level with the average density value, extreme density value, and density dispersion. This yields the first proton density coordinate parameter pair, the second proton density coordinate parameter pair, and the third proton density coordinate parameter pair for each organ, forming a second correlation relationship that includes tomographic region information and parameter clustering features.

[0018] Optionally, the average signal intensity, peak signal intensity, and signal coverage of the same organ in the proton density signal features are correlated at the regional level with the average density, extreme density, and density dispersion, respectively, to obtain a first pair of proton density coordinate parameters, a second pair of proton density coordinate parameters, and a third pair of proton density coordinate parameters for each organ, thereby forming a second correlation relationship that includes tomographic region information and parameter clustering features, including:

[0019] The overlapping areas of the proton signal regions and tomographic regions of each organ are marked as proton density-correlated regions.

[0020] The coordinate boundary of the proton density associated region is used as a spatial identifier. The spatial identifier is associated with the average signal intensity and average density of the same organ to obtain a first proton density coordinate parameter pair. The spatial identifier is associated with the signal peak and density extreme value of the same organ to obtain a second proton density coordinate parameter pair. The spatial identifier is associated with the signal coverage and density dispersion of the same organ to obtain a third proton density coordinate parameter pair.

[0021] By integrating the first, second, and third proton density coordinate parameter pairs of the same organ, a proton density region association set is obtained.

[0022] Based on the fault region numbering, the proton density region association sets of all organs are classified to obtain multiple sets of association data. All association data are then regionally aggregated to obtain the second association relationship.

[0023] Optionally, the first association relationship and the second association relationship are fused to obtain a cross-modal feature matching dataset, and the cross-modal feature matching dataset is then parameter-calibrated to obtain a calibrated cross-modal feature matching dataset, including:

[0024] The first fluorescence density coordinate parameter pair, the second fluorescence density coordinate parameter pair, and the third fluorescence density coordinate parameter pair in the first association relationship of the same organ are integrated with the first proton density coordinate parameter pair, the second proton density coordinate parameter pair, and the third proton density coordinate parameter pair in the second association relationship to obtain the first cross-modal sub-feature set, the second cross-modal sub-feature set, and the third cross-modal sub-feature set, so as to generate a cross-modal feature matching dataset.

[0025] Based on the cross-modal feature matching dataset, the actual range of fluorescence signal intensity, the actual range of tissue density attenuation coefficient, and the actual range of proton signal resonance frequency are determined.

[0026] The first difference between the actual range of fluorescence signal intensity and the preset standard wavelength range, the second difference between the actual range of tissue density attenuation coefficient and the preset standard attenuation coefficient range, and the third difference between the actual range of proton signal resonance frequency and the preset standard resonance frequency range are calculated to perform multimodal parameter calibration on the cross-modal feature matching dataset, thereby obtaining the calibrated cross-modal feature matching dataset.

[0027] Optionally, a linkage analysis is performed on the cross-modal features of each organ in the calibrated cross-modal feature matching dataset to obtain liveness detection results, including:

[0028] Cross-modal feature parameters of each organ are extracted from the calibrated cross-modal feature matching dataset to determine the feature association dimensions between multiple organs. The feature association dimensions include the inter-organ same parameter difference dimension, the intra-organ heterogeneous parameter ratio dimension, and the inter-organ heterogeneous parameter trend dimension.

[0029] The first set of correlation parameters for each organ under the dimension of the same parameter difference between organs, the second set of correlation parameters under the dimension of different parameter ratio within organs, and the third set of correlation parameters under the dimension of different parameter trends between organs are calculated to form a set of organ linkage features.

[0030] Based on organ category and feature dimension, the set of organ linkage features is classified and labeled to obtain a feature association map;

[0031] The cross-modal feature parameters and correlation parameters of each organ in the feature association map are integrated to obtain the liveness detection results.

[0032] Optionally, the following parameters are calculated: a first set of correlation parameters for each organ under the inter-organ same parameter difference category dimension, a second set of correlation parameters under the intra-organ heterogeneous parameter ratio category dimension, and a third set of correlation parameters under the inter-organ heterogeneous parameter trend category dimension, including:

[0033] Based on the inter-organ parameter difference class dimension, the maximum wavelength difference, the maximum density difference, and the maximum proton difference in the actual range of fluorescence signal intensity, tissue density attenuation coefficient, and proton signal resonance frequency range of any two organs are calculated. The maximum wavelength difference, maximum density difference, and maximum proton difference are then summarized to obtain the first set of associated parameters.

[0034] Based on the organ-specific parameter ratio category, the median fluorescence value of the actual range of fluorescence signal intensity, the median density value of the actual range of tissue density attenuation coefficient, and the median proton value of the actual range of proton signal resonance frequency are calculated for each organ to determine the fluorescence density ratio and density-proton ratio. The fluorescence density ratio and density-proton ratio of each organ are then summarized to obtain the second set of associated parameters.

[0035] Based on the inter-organ heterogeneous parameter trend class dimension, the first change direction of the fluorescence median value, the second change direction of the density median value, and the third change direction of the proton median value in two adjacent organs in the anatomical position sequence are determined. The first change direction, the second change direction, and the third change direction are summarized to obtain the third set of associated parameters.

[0036] Secondly, this application provides a liveness detection system based on multimodal imaging integrated analysis, comprising:

[0037] The acquisition module is used to acquire multimodal imaging data of multiple organs of a living object, including original fluorescence imaging images, original tissue density imaging images, and original proton density imaging images.

[0038] The calculation module is used to calculate the fluorescence intensity distribution characteristics, tissue density value characteristics, and proton density signal characteristics based on the original fluorescence imaging image, the original tissue density imaging image, and the original proton density imaging image, respectively.

[0039] The correlation module is used to perform cross-modal correlation processing on the fluorescence intensity distribution feature and the proton density signal feature with the tissue density value feature, respectively, to obtain a first correlation relationship between the fluorescence intensity distribution feature and the tissue density value feature, and a second correlation relationship between the proton density signal feature and the tissue density value feature;

[0040] The calibration module is used to fuse the first association relationship and the second association relationship into parameters to obtain a cross-modal feature matching dataset, and to calibrate the parameters of the cross-modal feature matching dataset to obtain a calibrated cross-modal feature matching dataset.

[0041] The analysis module is used to perform linked analysis on the cross-modal features of each organ in the calibrated cross-modal feature matching dataset to obtain the liveness detection results.

[0042] Thirdly, this application provides an electronic device, comprising:

[0043] Memory, used to store computer programs;

[0044] A processor, configured to execute the computer program to implement the steps of a liveness detection method based on multimodal imaging integrated analysis as described in the first aspect above.

[0045] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the steps of a liveness detection method based on multimodal imaging integrated analysis as described in the first aspect above.

[0046] This application provides a liveness detection method based on multimodal imaging integration analysis. The method acquires multimodal imaging data of multiple organs of a live object, including original fluorescence imaging images, original tissue density imaging images, and original proton density imaging images. Based on these images, fluorescence intensity distribution features, tissue density value features, and proton density signal features are calculated. The fluorescence intensity distribution features and proton density signal features are then correlated with the tissue density value features across modalities to obtain a first correlation between the fluorescence intensity distribution features and the tissue density value features, and a second correlation between the proton density signal features and the tissue density value features. The first and second correlations are then fused to obtain a cross-modal feature matching dataset. This dataset is then calibrated to obtain a calibrated cross-modal feature matching dataset. Finally, the cross-modal features of each organ in the calibrated dataset are analyzed in conjunction to obtain the liveness detection result.

[0047] This application offers the following advantages: By acquiring multimodal imaging data of multiple organs from a living subject, it provides a comprehensive data foundation for subsequent multidimensional analysis, overcoming the information limitations of single-modal imaging in characterizing the complex properties of biological tissues. It achieves the transformation from raw imaging data to quantifiable and comparable feature parameters, improving the resolvability and structured representation capabilities of different modal data. It enhances the spatial and functional correspondence of features across modalities, resolving feature misalignment and semantic disconnect caused by differences in imaging principles. It improves the consistency and reliability of multi-source feature data, suppressing systematic errors caused by equipment differences, environmental noise, or imaging timing deviations, thus ensuring the construction of a high-precision integrated analysis model. It enables the collaborative assessment of the functional states of multiple organs, dynamically capturing the correlation patterns of physiological responses between organs, thereby improving the overall accuracy of in vivo detection and the ability to determine physiological correlations. Furthermore, by performing spatial coordinate matching and calibration based on organ and tomographic regions for the original fluorescence imaging and tissue density imaging images, and the original proton density imaging and tissue density imaging images, respectively, feature parameters of each modality were correlated item by item within a unified spatial framework, forming a first correlation relationship and a second correlation relationship. This solves the analytical bias caused by spatial mismatch and coarse feature matching in traditional fusion methods, and improves the localization accuracy and physiological reliability of multi-organ cross-modal analysis. Attached Figure Description

[0048] To more clearly illustrate the technical solutions of the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 A schematic flowchart of a liveness detection method based on multimodal imaging integrated analysis provided in this application embodiment;

[0050] Figure 2 A schematic flowchart of another liveness detection method based on multimodal imaging integrated analysis provided in this application embodiment;

[0051] Figure 3 A schematic diagram of the structure of a liveness detection system based on multimodal imaging integrated analysis provided in this application embodiment; Detailed Implementation

[0052] To address the need for in vivo multi-organ cross-modal functional linkage detection, existing technologies lack sophisticated cross-modal feature matching mechanisms, often leading to feature misalignment, information redundancy, or conflicts, thus affecting the accurate assessment of organ synergistic functional status. This application proposes a live detection method and system based on multimodal imaging integration analysis. Through a dual-path association strategy at the feature level and region level, semantic alignment of multimodal physiological parameters is achieved within a unified anatomical framework. Based on this, parameter fusion and systematic calibration are performed to construct a highly consistent cross-modal feature matching dataset. Finally, through linkage analysis of the calibrated data, the functional correlation of multiple organs in dynamic physiological processes is comprehensively captured. This improves integration accuracy while enhancing the physiological reliability and analytical depth of live detection, overcoming the key technical barriers of existing technologies in terms of cross-modal alignment accuracy and functional integration effectiveness.

[0053] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are merely some embodiments of the present application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0054] The core of this application is to provide a liveness detection method based on multimodal imaging integrated analysis, and a flowchart of one specific implementation is shown below. Figure 1 As shown, the method includes:

[0055] Step 101: Acquire multimodal imaging data of multiple organs of a living subject, including original fluorescence imaging images, original tissue density imaging images, and original proton density imaging images.

[0056] In this step, the living subject refers to a biological individual with vital activities, and multimodal imaging analysis is performed based on its physiological structure and functional characteristics. Multimodal imaging data refers to the comprehensive imaging information obtained by detecting the living subject through multiple imaging methods, including raw tissue density imaging images reflecting organ density characteristics and raw proton density imaging images reflecting organ proton distribution characteristics. Raw fluorescence imaging images refer to unprocessed images obtained through fluorescence imaging technology, used to extract fluorescence signal-related features of organs. Raw tissue density imaging images refer to unprocessed images obtained through tissue density imaging technology, used to extract density-related features of organs. Raw proton density imaging images refer to unprocessed images obtained through proton density imaging technology, used to extract proton signal-related features of organs.

[0057] In this embodiment of the application, a living object is scanned by a fluorescence imaging device, a tissue density imaging device, and a proton density imaging device to obtain imaging data of multiple organs in the living object. This imaging data is multimodal imaging data. The unprocessed image generated by the fluorescence imaging device is the original fluorescence imaging image, the unprocessed image generated by the tissue density imaging device is the original tissue density imaging image, and the unprocessed image generated by the proton density imaging device is the original proton density imaging image.

[0058] Step 102: Based on the original fluorescence imaging image, the original tissue density imaging image, and the original proton density imaging image, calculate the fluorescence intensity distribution characteristics, tissue density value characteristics, and proton density signal characteristics, respectively.

[0059] In this step, fluorescence intensity distribution features refer to the set of features calculated based on the original fluorescence imaging image that reflects the variation law of organ fluorescence signal intensity, used to represent the distribution of organ fluorescence characteristics. Tissue density value features refer to the set of features calculated based on the original tissue density imaging image that reflects the variation law of organ density values, including the average density, density extreme values, and density dispersion, used to represent the distribution of organ density characteristics. Proton density signal features refer to the set of features calculated based on the original proton density imaging image that reflects the variation law of organ proton signal, used to represent the distribution of organ proton signal characteristics.

[0060] Step 103: Perform cross-modal correlation processing on the fluorescence intensity distribution feature and the proton density signal feature with the tissue density value feature, respectively, to obtain the first correlation relationship between the fluorescence intensity distribution feature and the tissue density value feature, and the second correlation relationship between the proton density signal feature and the tissue density value feature.

[0061] In this step, the first correlation refers to the set of relationships obtained after cross-modal correlation processing of fluorescence intensity distribution characteristics and tissue density value characteristics, which reflects the correlation between fluorescence characteristics and density characteristics. The second correlation refers to the set of relationships obtained after cross-modal correlation processing of proton density signal characteristics and tissue density value characteristics, which reflects the correlation between proton signal characteristics and density characteristics.

[0062] Step 104: Perform parameter fusion on the first association relationship and the second association relationship to obtain a cross-modal feature matching dataset, and perform parameter calibration on the cross-modal feature matching dataset to obtain a calibrated cross-modal feature matching dataset.

[0063] In this step, the cross-modal feature matching dataset refers to the dataset obtained by parameter fusion of the first and second association relationships, used to integrate feature association information of fluorescence, density, and proton modes. The calibrated cross-modal feature matching dataset refers to the dataset obtained after parameter calibration of the cross-modal feature matching dataset, used to improve the consistency and accuracy of multimodal features.

[0064] Step 105: Perform a linkage analysis on the cross-modal features of each organ in the calibrated cross-modal feature matching dataset to obtain the liveness detection results.

[0065] In this step, cross-modal features refer to organ features extracted from the calibrated cross-modal feature matching dataset that integrate multiple modal information such as fluorescence, density, and protons, and are used for multi-organ linkage analysis. Live detection results refer to the results obtained after linkage analysis of the cross-modal features of multiple organs in the calibrated cross-modal feature matching dataset, used to reflect the physiological or pathological state of the living subject.

[0066] The embodiments of this application realize the deep integration and correlation analysis of multimodal information, improve the comprehensiveness and accuracy of live detection, and provide a more reliable basis for the assessment of the physiological or pathological state of live objects.

[0067] This application provides a specific embodiment. Step 102 involves calculating the fluorescence intensity distribution characteristics, tissue density value characteristics, and proton density signal characteristics based on the original fluorescence imaging image, the original tissue density imaging image, and the original proton density imaging image, respectively. This specifically includes the following steps:

[0068] Step 201: Mark the organ regions where each organ is located in the original fluorescence imaging image, count the fluorescence signal intensity values ​​of the pixels in each organ region, and calculate the variation range, distribution range, and intensity difference between pixels of each fluorescence signal intensity value. Summarize the variation range, distribution range, and intensity difference corresponding to each organ to obtain the fluorescence intensity distribution characteristics.

[0069] In this step, the organ region refers to the image region corresponding to a specific organ within a living subject, marked by an image segmentation algorithm in the original fluorescence imaging image, used for subsequent extraction of fluorescence signal-related parameters of that organ. A pixel is the basic unit constituting the original fluorescence imaging image, used to quantify the intensity of fluorescence signals at different locations in the image. The fluorescence signal intensity value refers to the numerical value carried by each pixel in the original fluorescence imaging image, reflecting the fluorescence radiation intensity at that location, used to calculate fluorescence-related feature parameters. The variation range refers to the maximum difference in fluorescence signal intensity values ​​within the same organ region, reflecting the fluctuation range of the fluorescence signal in that organ. The distribution range refers to the interval from the minimum to the maximum value of the fluorescence signal intensity values ​​within the same organ region, used to represent the overall distribution span of the fluorescence signal within that organ. The intensity difference between pixels refers to the average of the differences in fluorescence signal intensity values ​​between any two pixels within the same organ region, reflecting the uniformity of the fluorescence signal within that organ.

[0070] In this embodiment, the original fluorescence imaging image is processed. By identifying the grayscale differences and contour features of different organs in the image, the organ regions where each organ is located are marked. Then, the fluorescence signal intensity values ​​of all pixels in each organ region are statistically analyzed using pixel extraction technology. The variation range, distribution range, and intensity difference between pixels are calculated. The calculation formula is: variation range = maximum fluorescence signal intensity - minimum fluorescence signal intensity; intensity difference between pixels = sum of the differences in fluorescence signal intensity values ​​between all pairs of pixels ÷ number of pixels. Finally, the variation range, distribution range, and intensity difference between pixels corresponding to each organ are integrated according to organ category to obtain fluorescence intensity distribution characteristics.

[0071] Step 202: Perform tomographic correlation processing on the original tissue density imaging image to determine the tomographic regions of each organ, so as to obtain the tomographic region set of each organ, and calculate the average density, extreme density, and density dispersion of each tomographic region set. Summarize the average density, extreme density, and density dispersion of each organ to obtain the tissue density value features.

[0072] In this step, a tomographic region refers to an image region of the same organ within a specific tomographic plane, determined through tomographic correlation processing in the original tissue density imaging image. It reflects the density distribution of the organ within that tomographic plane. A tomographic region set is a collection formed by integrating tomographic regions of the same organ across all tomographic planes, used for comprehensive analysis of the organ's density characteristics. The density mean is the arithmetic mean of all density values ​​within the same tomographic region set, reflecting the average density level of the organ across multiple tomographic planes. Density extremes refer to the maximum and minimum density values ​​within the same tomographic region set, used to represent extreme cases of organ density. Density dispersion refers to the degree of dispersion of density values ​​within the same tomographic region set, reflecting the uniformity of the organ's density distribution.

[0073] In this embodiment, the original tissue density imaging image is processed by examining images of different sections one by one, identifying the iconic structures and contour curves of the same organ in each section, and determining the area occupied by the organ in each section as a section region by comparing the positions of these landmarks and contours in different sections. All section regions of the same organ are integrated to form a section region set for each organ. Then, the density data in each section region set is analyzed to calculate the average density value. The calculation formula is: Average density value = Sum of all density values ​​in the section region set ÷ Number of density values, density extreme values ​​(including maximum and minimum density values), and density dispersion. The calculation formula is: Density dispersion = √[(Sum of each density value - average density value)² ÷ Number of density values]. Finally, the average density value, density extreme value, and density dispersion corresponding to each organ are summarized by organ category to obtain tissue density value features.

[0074] Step 203: Mark the proton signal regions where each organ is located from the original proton density imaging image, and calculate the average signal intensity, peak signal intensity, and signal coverage of each proton signal region. Summarize the average signal intensity, peak signal intensity, and signal coverage of each organ to obtain the proton density signal features.

[0075] In this step, the proton signal region refers to the proton signal distribution area corresponding to a specific organ, marked by a signal region recognition algorithm in the original proton density imaging image, used to extract proton signal-related parameters for that organ. The average signal intensity is the arithmetic mean of all proton signal values ​​within the same proton signal region, reflecting the overall strength of the proton signal in that organ. The signal peak value is the maximum proton signal value within the same proton signal region, used to represent the strongest point of the proton signal in that organ. The signal coverage area refers to the proportion of pixels containing proton signals within the same proton signal region to the total number of pixels in that region, reflecting the breadth of proton signal coverage within that organ.

[0076] In this embodiment, from the original proton density imaging image, the background region and the proton signal region are distinguished based on the difference in proton signal intensity. Then, combined with the anatomical location and morphological characteristics of the organs, the corresponding range of each organ is marked as the proton signal region. The signal values ​​of all proton signal pixels in the region are extracted, and these values ​​are added together to obtain the sum. The calculation formula is: average signal intensity = sum of all proton signals in the region ÷ number of proton signals. The average signal intensity is calculated, and the largest value is selected as the signal peak value. The total number of pixels in the proton signal region of the organ and the number of pixels containing proton signals are counted. The calculation formula is: signal coverage = number of pixels containing proton signals ÷ total number of pixels in the proton signal region of the organ × 100%. The signal coverage is calculated. Finally, the average signal intensity, signal peak value, and signal coverage are combined and merged according to organ category to obtain the proton density signal features.

[0077] This application embodiment performs targeted processing on three types of original imaging images to accurately mark organ-related regions, calculates and summarizes key parameters reflecting the fluorescence, density, and proton signal characteristics of each organ, provides high-quality feature data for subsequent cross-modal correlation processing, ensures the accuracy and completeness of multimodal features, and lays the foundation for the reliability of in vivo detection results.

[0078] For example, in the multimodal imaging analysis of experimental animal A, the liver region in the original fluorescence imaging image was first marked. This region contains 5 pixels, and the fluorescence signal intensity values ​​of each pixel were statistically obtained as 70, 90, 110, 130, and 150. The variation range was calculated as 150 - 70 = 80. The distribution range is from the minimum to the maximum value of all fluorescence signal intensity values, i.e., 70 to 150. When calculating the intensity difference between pixels, the differences between each pair of pixels were first calculated: 90 - 70 = 20, 110 - 70 = 40, and 130 = 150. 0-70=60, 150-70=80, 110-90=20, 130-90=40, 150-90=60, 130-110=20, 150-110=40, 150-130=20, the sum is: 20+40+60+80+20+40+60+20+40+20=400, the number of pixel pairs is 10, the intensity difference between pixels = 400÷10=25, these parameters are summarized to obtain the fluorescence intensity distribution characteristics of the liver; then the original tissue density imaging image is subjected to tomographic correlation processing. The liver was located in the fault regions of layers 3 to 8, a total of 6 layers, forming a set of fault regions for the liver. The density values ​​of each fault region were 0.9 g / cm³, 1.0 g / cm³, 1.1 g / cm³, 1.2 g / cm³, 1.3 g / cm³, and 1.1 g / cm³, respectively. When calculating the average density, the total density was 0.9 + 1.0 + 1.1 + 1.2 + 1.3 + 1.1 = 6.6 g / cm³, and the average density was 6.6 ÷ 6 = 1.1 g / cm³. The extreme density value was the minimum of all density values, 0.9 g / cm³. m³ and the maximum value is 1.3 g / cm³; when calculating the density dispersion, first calculate the square of the difference between each density value and the average value ((0.9-1.1)²=0.04, (1.0-1.1)²=0.01, (1.1-1.1)²=0, (1.2-1.1)²=0.01, (1.3-1.1)²=0.04, (1.1-1.1)²=0), the sum is 0.04+0.01+0+0.01+0.04+0=0.1, density dispersion =√(0.1÷6)≈√0.0167≈0.13. Summarize these parameters to obtain the tissue density characteristics of the liver; finally, mark the proton signal region of the liver from the original proton density imaging image. This region has a total of 1000 pixels, of which 900 pixels contain proton signals; when calculating the average signal intensity, take 10 pixels with signals, whose signal values ​​are 80, 85, 90, 95, 90, 85, 95, 100, 80, and 90, respectively, and the sum is 80+85+90+95+90+85+95+100+80+90=890. The average signal intensity = 890÷10=89≈90; the signal peak value is the maximum value of all signal values, 100; the signal coverage = 900÷1000×100%=90%.

[0079] This application provides a specific embodiment, such as Figure 2 As shown, step 103 involves performing cross-modal correlation processing on the fluorescence intensity distribution feature and the proton density signal feature with the tissue density value feature, respectively, to obtain a first correlation relationship between the fluorescence intensity distribution feature and the tissue density value feature, and a second correlation relationship between the proton density signal feature and the tissue density value feature. This specifically includes the following steps:

[0080] Step 301: Based on the organ region and tomographic region set, perform spatial coordinate matching on the original fluorescence imaging image and the original tissue density imaging image. Under the matched spatial coordinates, the variation range, distribution range, and intensity difference between pixels of the fluorescence signal intensity value of the same organ in the fluorescence intensity distribution feature are respectively correlated with the average density value, extreme density value, and density dispersion in the tissue density value feature. This yields the first fluorescence density coordinate parameter pair, the second fluorescence density coordinate parameter pair, and the third fluorescence density coordinate parameter pair for each organ, forming a first correlation relationship that includes spatial coordinate information and parameter correlation coefficients.

[0081] In this step, the matched spatial coordinates refer to the unified coordinate system obtained after matching the spatial coordinates of the original fluorescence imaging image and the original tissue density imaging image, used to ensure a one-to-one correspondence between the spatial positions of the same organ in the two images. The first fluorescence density coordinate parameter pair refers to the parameter combination formed by feature-level correlation between the variation range of fluorescence signal intensity values ​​of the same organ and the average density value in the tissue density feature under the matched spatial coordinates, used to reflect the correlation between the fluorescence signal fluctuation range and the average tissue density. The second fluorescence density coordinate parameter pair refers to the parameter combination formed by feature-level correlation between the distribution range of fluorescence signal intensity values ​​of the same organ and the density extreme values ​​in the tissue density feature under the matched spatial coordinates, used to reflect the correlation between the fluorescence signal distribution span and the extreme values ​​of tissue density. The third fluorescence density coordinate parameter pair refers to the parameter combination formed by feature-level correlation between the intensity difference between pixels of fluorescence signal intensity values ​​of the same organ and the density dispersion in the tissue density feature under the matched spatial coordinates, used to reflect the correlation between fluorescence signal uniformity and tissue density uniformity.

[0082] In this embodiment, based on the organ regions marked in the original fluorescence imaging image and the set of tomographic regions obtained from the original tissue density imaging image, common anatomical landmarks of the same organ in the two images are selected. The coordinate system of the fluorescence imaging image is aligned with the coordinate system of the tissue density imaging image. By adjusting the scaling and rotation angle of the fluorescence imaging image, the coordinates of the landmarks in the two images are made consistent, resulting in matched spatial coordinates. Under these matched spatial coordinates, the variation range of fluorescence signal intensity values ​​in the fluorescence intensity distribution characteristics of the same organ is found, along with the average density value in the tissue density value characteristics of the organ. The corresponding spatial coordinates and values ​​are recorded together to form a first fluorescence density coordinate parameter pair. Similarly, the distribution range of fluorescence signal intensity values ​​is recorded in correspondence with density extremes to form a second fluorescence density coordinate parameter pair. The intensity differences between each pixel are recorded in correspondence with the density dispersion to form a third fluorescence density coordinate parameter pair. According to organ category, these three parameter pairs for each organ are integrated into a first correlation relationship, which reflects the matched spatial coordinate information and the correlation coefficients between the parameters.

[0083] Step 302: Based on the proton signal region and the tomographic region set, perform spatial coordinate calibration on the original proton density imaging image and the original tissue density imaging image. Within the calibrated spatial range, correlate the average signal intensity, peak signal intensity, and signal coverage of the same organ in the proton density signal features with the average density value, extreme density value, and density dispersion at the regional level, respectively, to obtain the first proton density coordinate parameter pair, the second proton density coordinate parameter pair, and the third proton density coordinate parameter pair for each organ, so as to form a second correlation relationship containing tomographic region information and parameter clustering features.

[0084] In this step, the calibrated spatial range refers to the overlapping spatial region obtained after performing spatial coordinate calibration on the original proton density imaging and tissue density imaging images. This ensures that the spatial ranges of the same organ in the two images completely correspond. The first proton density coordinate parameter pair refers to a parameter combination formed by regionally correlating the average signal intensity in the proton density signal characteristics of the same organ with the average density value in the tissue density value characteristics within the calibrated spatial range. This reflects the correlation between the overall strength of the proton signal and the average tissue density. The second proton density coordinate parameter pair refers to a parameter combination formed by regionally correlating the signal peak value in the proton density signal characteristics of the same organ with the density extreme value in the tissue density value characteristics within the calibrated spatial range. This reflects the correlation between the strongest point of the proton signal and the extreme value of the tissue density. The third proton density coordinate parameter pair refers to a parameter combination formed by regionally correlating the signal coverage area in the proton density signal characteristics of the same organ with the density dispersion in the tissue density value characteristics within the calibrated spatial range. This reflects the correlation between the breadth of proton signal coverage and the uniformity of tissue density.

[0085] In this embodiment, based on the proton signal region marked in the original proton density imaging image and the set of tomographic regions obtained from the original tissue density imaging image, the difference in the contour size of the same organ in the two images is measured. The coordinate scale of the original proton density imaging image is adjusted according to the difference so that the boundaries of the proton signal region and the set of tomographic regions in three-dimensional space completely coincide, resulting in a calibrated spatial range. Within the calibrated spatial range, the average signal intensity in the proton density signal features of the same organ is extracted and compared with the average density in the tissue density value features of the organ. The spatial range information and values ​​corresponding to the two are recorded together to form a first proton density coordinate parameter pair. Similarly, the signal peak value and the density extreme value are recorded to form a second proton density coordinate parameter pair. The signal coverage area and the density dispersion are recorded to form a third proton density coordinate parameter pair. According to the organ category, these three parameter pairs for each organ are integrated into a second correlation relationship, which reflects the tomographic region information and the clustering characteristics between the parameters.

[0086] This application embodiment achieves precise correlation between fluorescence, proton features and tissue density features by performing spatial coordinate matching and calibration on imaging images of different modalities. This forms a correlation relationship that includes spatial information and parameter relationships, laying a spatially consistent and closely correlated foundation for the deep integration of multimodal features and improving the accuracy and reliability of multimodal analysis.

[0087] This application provides a specific embodiment. Step 302 involves performing a region-level correlation between the average signal intensity, peak signal intensity, and signal coverage of the same organ in the proton density signal features and the average density value, extreme density value, and density dispersion, respectively, to obtain a first proton density coordinate parameter pair, a second proton density coordinate parameter pair, and a third proton density coordinate parameter pair for each organ, thereby forming a second correlation relationship that includes tomographic region information and parameter clustering features. Specifically, this includes the following steps:

[0088] Step 311: Mark the overlapping areas of the proton signal regions and tomographic regions of each organ as proton density correlation regions.

[0089] In this step, the proton density correlation region refers to the three-dimensional region where the proton signal region of each organ overlaps with the tomographic region. It is used to define the spatial range of the regional correlation and ensure that the correlation parameters come from the same spatial region.

[0090] In this embodiment of the application, the proton signal region and tomographic region of each organ are obtained, the coordinate range of the two regions are superimposed and displayed in a three-dimensional coordinate system, and the pixels in the two regions are compared point by point to see if they exist in both regions at the same time. The region composed of all the pixels that exist in the same region is counted. This region is the three-dimensional region where the two overlap. This region is marked in the image with a special color or mark and named the proton density associated region.

[0091] Step 312: Use the coordinate boundary of the proton density associated region as a spatial identifier. Associate the spatial identifier with the average signal intensity and average density of the same organ to obtain a first proton density coordinate parameter pair. Associate the spatial identifier with the signal peak value and density extreme value of the same organ to obtain a second proton density coordinate parameter pair. Associate the spatial identifier with the signal coverage and density dispersion of the same organ to obtain a third proton density coordinate parameter pair.

[0092] In this step, the spatial identifier refers to the three-dimensional coordinate boundary of the proton density correlation region, which is used to uniquely identify the spatial location of the correlation parameter, so that different parameters can be correlated through the same spatial identifier.

[0093] In this embodiment, the three-dimensional coordinate data of the proton density associated region is read, and the minimum and maximum coordinate values ​​of the region on the coordinate axes are determined. These coordinate values ​​are combined into a three-dimensional coordinate boundary as a spatial identifier. In the data table, the spatial identifier is used as an index for a row, and the average signal intensity value and average density value of the same organ are filled in the same row to form a complete record as the first proton density coordinate parameter pair. Similarly, using the spatial identifier as an index, the signal peak value and density extreme value are filled in the same row to form the second proton density coordinate parameter pair. Using the spatial identifier as an index, the signal coverage range value and density dispersion value are filled in the same row to form the third proton density coordinate parameter pair.

[0094] Step 313: Integrate the first proton density coordinate parameter pair, the second proton density coordinate parameter pair, and the third proton density coordinate parameter pair of the same organ to obtain the proton density region association set.

[0095] In this step, the proton density region association set refers to the integrated set of the first proton density coordinate parameter pair, the second proton density coordinate parameter pair, and the third proton density coordinate parameter pair of the same organ, which is used to centrally reflect the correlation between the proton signal and density characteristics of the organ.

[0096] In this embodiment of the application, a first proton density coordinate parameter pair, a second proton density coordinate parameter pair, and a third proton density coordinate parameter pair of the same organ are collected. These parameter pairs are sorted according to spatial identifiers so that parameter pairs corresponding to the same spatial identifier are arranged adjacently. Then, the sorted parameter pairs are grouped according to parameter type, and parameter pairs belonging to the same parameter type are placed together to form an ordered list containing all regional correlation parameters of the organ. This list is the proton density regional correlation set.

[0097] Step 314: Based on the fault region number, classify the proton density region association sets of all organs to obtain multiple sets of association data, and then perform regional aggregation of all association data to obtain the second association relationship.

[0098] In this step, the fault region number refers to a unique identifier assigned to each fault region, based on the order and location information of the faults, used to distinguish different fault regions and provide a basis for the classification of associated data. Associated data refers to the set of proton density region associations for all organs under the same fault region number after classification, used to reflect the multi-organ association characteristics within a specific fault region.

[0099] In this embodiment of the application, based on the inherent numbering information of the fault region, the proton density region association sets of all organs are grouped according to the number of the fault region to which they belong, so that the association sets in the fault region corresponding to the same number are grouped together to obtain multiple sets of association data; according to the positional order of the spatial identifier, each set of association data is arranged, and duplicate information is merged to finally form a second association relationship containing the association information of all fault regions.

[0100] This application embodiment accurately marks the proton density-related regions, uses spatial identifiers as a link to associate proton signals and density feature parameters, and then integrates, classifies, and aggregates them regionally to form a structured second association relationship. This achieves the accuracy and systematic nature of regional-level association, providing high-quality association data for the deep fusion of multimodal features.

[0101] This application provides a specific embodiment. Step 104 involves fusing the first association relationship and the second association relationship to obtain a cross-modal feature matching dataset, and then calibrating the parameters of the cross-modal feature matching dataset to obtain a calibrated cross-modal feature matching dataset. The specific steps include:

[0102] Step 401: Integrate the first fluorescence density coordinate parameter pair, the second fluorescence density coordinate parameter pair, and the third fluorescence density coordinate parameter pair in the first association relationship of the same organ with the first proton density coordinate parameter pair, the second proton density coordinate parameter pair, and the third proton density coordinate parameter pair in the second association relationship to obtain the first cross-modal sub-feature set, the second cross-modal sub-feature set, and the third cross-modal sub-feature set, so as to generate a cross-modal feature matching dataset.

[0103] In this step, the first cross-modal sub-feature set refers to the subset of data formed by integrating the first fluorescence density coordinate parameter pair and the first proton density coordinate parameter pair of the same organ, used to reflect the cross-modal correlation between fluorescence and proton signals at the level of basic features. The second cross-modal sub-feature set refers to the subset of data formed by integrating the second fluorescence density coordinate parameter pair and the second proton density coordinate parameter pair of the same organ, used to reflect the cross-modal correlation between fluorescence and proton signals at the level of extreme features. The third cross-modal sub-feature set refers to the subset of data formed by integrating the third fluorescence density coordinate parameter pair and the third proton density coordinate parameter pair of the same organ, used to reflect the cross-modal correlation between fluorescence and proton signals at the level of distribution features.

[0104] In this embodiment, the first fluorescence density coordinate parameter pair in the first association relationship and the first proton density coordinate parameter pair in the second association relationship of the same organ are compared. The spatial identifiers of the two are compared to find parameter pairs with the same identifier. All parameter values ​​and coordinate information contained in these two parameter pairs are merged into a new data structure, which is the first cross-modal sub-feature set. Using the same method, the second fluorescence density coordinate parameter pair and the second proton density coordinate parameter pair are aligned and merged according to their spatial identifiers to obtain the second cross-modal sub-feature set. The third fluorescence density coordinate parameter pair and the third proton density coordinate parameter pair are aligned and merged according to their spatial identifiers to obtain the third cross-modal sub-feature set. The first, second and third cross-modal sub-feature sets of all organs are sequentially placed into the same database table to generate a cross-modal feature matching dataset.

[0105] Step 402: Based on the cross-modal feature matching dataset, determine the actual range of fluorescence signal intensity, the actual range of tissue density attenuation coefficient, and the actual range of proton signal resonance frequency.

[0106] In this step, the actual range of fluorescence signal intensity refers to the overall fluctuation range of fluorescence signal intensity values ​​for all organs, extracted based on the cross-modal feature matching dataset, and is used to reflect the overall intensity span of the fluorescence signal. The actual range of tissue density attenuation coefficient refers to the overall fluctuation range of tissue density attenuation coefficients for all organs, extracted based on the cross-modal feature matching dataset, and is used to reflect the overall magnitude of tissue density attenuation. The actual range of proton signal resonance frequency refers to the overall fluctuation range of proton signal resonance frequencies for all organs, extracted based on the cross-modal feature matching dataset, and is used to reflect the overall range of proton signal resonance frequencies.

[0107] In this embodiment, the cross-modal feature matching dataset is analyzed, and the fluorescence signal intensity related values ​​of all organs are extracted from the first cross-modal sub-feature set, the second cross-modal sub-feature set, and the third cross-modal sub-feature set. The minimum and maximum values ​​are determined to form the actual range of fluorescence signal intensity. The attenuation coefficient related to tissue density is extracted, and the minimum and maximum values ​​are determined to form the actual range of tissue density attenuation coefficient. The resonant frequency related values ​​of proton signal are extracted, and the minimum and maximum values ​​are determined to form the actual range of proton signal resonant frequency.

[0108] Step 403: Calculate the first difference between the actual range of fluorescence signal intensity and the preset standard wavelength range, the second difference between the actual range of tissue density attenuation coefficient and the preset standard attenuation coefficient range, and the third difference between the actual range of proton signal resonance frequency and the preset standard resonance frequency range, so as to perform multimodal parameter calibration on the cross-modal feature matching dataset and obtain the calibrated cross-modal feature matching dataset.

[0109] In this step, the preset standard wavelength range refers to a pre-defined standard wavelength range corresponding to the fluorescence signal intensity under normal physiological conditions, used as a reference benchmark for fluorescence parameter calibration. The first difference refers to the difference between the actual fluorescence signal intensity range and the preset standard wavelength range, used to quantify the deviation of the fluorescence parameters from the standard. The preset standard attenuation coefficient range refers to a pre-defined standard range corresponding to the tissue density attenuation coefficient under normal physiological conditions, used as a reference benchmark for density parameter calibration. The second difference refers to the difference between the actual tissue density attenuation coefficient range and the preset standard attenuation coefficient range, used to quantify the deviation of the density parameters from the standard. The preset standard resonance frequency range refers to a pre-defined standard range corresponding to the proton signal resonance frequency under normal physiological conditions, used as a reference benchmark for proton parameter calibration. The third difference refers to the difference between the actual proton signal resonance frequency range and the preset standard resonance frequency range, used to quantify the deviation of the proton parameters from the standard.

[0110] In this embodiment, the first difference between the actual range of fluorescence signal intensity and the preset standard wavelength range is calculated using the formula: First difference = Upper limit of actual fluorescence signal intensity range - Upper limit of standard wavelength range, and First difference = Lower limit of actual fluorescence signal intensity range - Lower limit of standard wavelength range; the second difference between the actual range of tissue density attenuation coefficient and the preset standard attenuation coefficient range is calculated using the formula: Second difference = Upper limit of actual tissue density attenuation coefficient range - Upper limit of standard attenuation coefficient range, and Second difference = Lower limit of actual tissue density attenuation coefficient range - Lower limit of standard attenuation coefficient range; the second difference between the actual range of proton signal resonance frequency and the preset standard attenuation coefficient range is calculated using the formula: The third difference in the standard resonance frequency range is calculated using the following formulas: Third difference = Upper limit of the actual resonance frequency range of the proton signal - Upper limit of the standard resonance frequency range, and Third difference = Lower limit of the actual resonance frequency range of the proton signal - Lower limit of the standard resonance frequency range. Based on these three differences, the corresponding first difference is subtracted from the relevant values ​​of each fluorescence signal intensity in the cross-modal feature matching dataset for correction; the corresponding second difference is subtracted from the value of each tissue density attenuation coefficient for correction; and the corresponding third difference is subtracted from the value of each proton signal resonance frequency for correction. All the corrected data are then recombined to obtain the calibrated cross-modal feature matching dataset.

[0111] This application embodiment generates a matching dataset by fusing parameters with different correlations across modalities, determines the actual feature range and compares it with the standard interval to calculate the difference, and then completes parameter calibration. This achieves deep integration and standardized adjustment of multimodal features, improves data consistency and reliability, and provides high-quality basic data for subsequent linkage analysis.

[0112] This application provides a specific embodiment. Step 105 involves performing a linkage analysis on the cross-modal features of each organ in the calibrated cross-modal feature matching dataset to obtain the liveness detection result. This specifically includes the following steps:

[0113] Step 501: Extract cross-modal feature parameters of each organ from the calibrated cross-modal feature matching dataset to determine the feature association dimensions among multiple organs. The feature association dimensions include the inter-organ same parameter difference dimension, the intra-organ heterogeneous parameter ratio dimension, and the inter-organ heterogeneous parameter trend dimension.

[0114] In this step, cross-modal feature parameters refer to organ feature parameters extracted from the calibrated cross-modal feature matching dataset, which integrate multimodal information such as fluorescence, tissue density, and proton signals, and are used to reflect the comprehensive attributes of organs in multimodal features. Feature correlation dimension refers to the angle or type used to analyze the correlation between cross-modal feature parameters of multiple organs, revealing the correlation patterns of organ features from different levels. Inter-organ parameter difference dimension refers to the analytical dimension used to compare the numerical differences of the same type of cross-modal feature parameters between different organs, reflecting the distributional differences of the same parameter in different organs. Intra-organ heterogeneous parameter ratio dimension refers to the analytical dimension used to analyze the proportional relationship between different types of cross-modal feature parameters within the same organ, reflecting the coordinated relationship between different features within the same organ. Inter-organ heterogeneous parameter trend dimension refers to the analytical dimension used to observe whether the changing directions of different types of cross-modal feature parameters between different organs are consistent, reflecting the coordinated change patterns of multiple parameters in different organs.

[0115] In this embodiment, cross-modal feature parameters of each organ are extracted from the calibrated cross-modal feature matching dataset. These parameters are classified by organ, and the same parameter in different organs is compared to observe the numerical differences and determine the inter-organ parameter difference class dimension. Different parameters of the same organ are analyzed to calculate their proportional relationship and determine the intra-organ heterogeneous parameter ratio class dimension. Different parameters of different organs are compared to observe the direction of their change with organ position and determine the inter-organ heterogeneous parameter trend class dimension. The inter-organ parameter difference class dimension, the intra-organ heterogeneous parameter ratio class dimension, and the inter-organ heterogeneous parameter trend class dimension are summarized to obtain the feature association dimension.

[0116] Step 502: Calculate the first set of correlation parameters for each organ under the dimension of the same parameter difference between organs, the second set of correlation parameters under the dimension of different parameter ratios within organs, and the third set of correlation parameters under the dimension of different parameter trends between organs, so as to form a set of organ linkage features.

[0117] In this step, the first set of related parameters refers to the set of related parameters obtained under the dimension of inter-organ parameter differences, used to quantify the degree of difference of the same parameter in different organs. The second set of related parameters refers to the set of related parameters obtained under the dimension of intra-organ heterogeneous parameter ratios, used to quantify the proportional relationship of different parameters in the same organ. The third set of related parameters refers to the set of related parameters obtained under the dimension of inter-organ heterogeneous parameter trends, used to reflect the synergy of parameter changes. The organ linkage feature set refers to the comprehensive dataset formed by integrating the first, second, and third sets of related parameters, used to systematically present the linkage relationship of organ features.

[0118] In this embodiment, for the dimension of inter-organ same parameter difference, the numerical difference of corresponding parameters of any two organs is calculated to form a first associated parameter set; for the dimension of intra-organ heterogeneous parameter ratio, the ratio of different parameters of the same organ is calculated to form a second associated parameter set; for the dimension of inter-organ heterogeneous parameter trend, the consistency of the change direction of parameters of different organs is statistically analyzed to form a third associated parameter set; these three associated parameter sets are integrated to obtain an organ linkage feature set containing multi-dimensional association information of all organs.

[0119] Step 503: Classify and label the organ linkage feature set according to organ category and feature dimension to obtain feature association map.

[0120] In this step, organ category refers to the classification of organs within a living organism, such as liver, heart, and kidneys, used to distinguish different organs in the feature association map. Feature dimension refers to the specific type of feature association dimension, namely, inter-organ parameter difference, intra-organ heterogeneous parameter ratio, and inter-organ heterogeneous parameter trend, used to label the association parameters in the feature association map according to the analysis perspective. The feature association map is a visual map obtained after classifying and labeling the set of organ linkage features according to organ category and feature dimension, used to intuitively display the feature linkage of multiple organs and multiple dimensions.

[0121] In this embodiment of the application, regions are divided according to organ categories, such as liver and heart, and feature dimensions. Data in the organ linkage feature set are marked with different symbols according to feature dimensions to obtain a feature association map that intuitively displays the correlation between parameters of each organ.

[0122] Step 504: Integrate the cross-modal feature parameters and correlation parameters of each organ in the feature association map to obtain the liveness detection result.

[0123] In this step, the association parameter refers to the numerical value or label calculated under the feature association dimension to reflect the parameter association relationship, and is used to quantify or describe the degree of association between features.

[0124] In this embodiment of the application, the cross-modal feature parameters and correlation parameters of each organ marked in the feature correlation map are systematically sorted and summarized, and the scattered correlation information is integrated into a comprehensive conclusion reflecting the overall state of the living body, so as to obtain the living body detection result.

[0125] This application embodiment extracts cross-modal feature parameters and determines multi-dimensional correlations, calculates and integrates correlation parameters to form a set of linked features, generates a map, and integrates the results to obtain the detection results. This realizes a systematic linked analysis of multi-organ and multi-modal features, improves the comprehensiveness and depth of live detection, and provides strong support for accurately assessing the live status.

[0126] This application provides a specific embodiment. Step 502 involves calculating the first set of correlation parameters for each organ under the dimension of inter-organ same parameter difference, the second set of correlation parameters under the dimension of intra-organ heterogeneous parameter ratio, and the third set of correlation parameters under the dimension of inter-organ heterogeneous parameter trend. Specifically, this includes the following steps:

[0127] Step 511: Based on the inter-organ parameter difference class dimension, calculate the maximum wavelength difference in the actual range of fluorescence signal intensity, the maximum density difference in the actual range of tissue density attenuation coefficient, and the maximum proton difference in the actual range of proton signal resonance frequency for any two organs. Summarize the maximum wavelength difference, maximum density difference, and maximum proton difference to obtain the first set of associated parameters.

[0128] In this step, the maximum wavelength difference refers to the maximum numerical difference between the actual ranges of fluorescence signal intensity of any two organs within the same parameter difference category, reflecting the maximum degree of difference in fluorescence signal intensity range between organs. The maximum density difference refers to the maximum numerical difference between the actual ranges of tissue density attenuation coefficients of any two organs within the same parameter difference category, reflecting the maximum degree of difference in tissue density attenuation coefficient range between organs. The maximum proton difference refers to the maximum numerical difference between the actual ranges of proton signal resonance frequencies of any two organs within the same parameter difference category, reflecting the maximum degree of difference in proton signal resonance frequency range between organs.

[0129] In this embodiment, based on the inter-organ parameter difference category, the actual range of fluorescence signal intensity of any two organs is compared to calculate the maximum difference, i.e., the maximum wavelength difference. The calculation formula is: Maximum wavelength difference = the maximum value of the upper limit difference or the maximum value of the lower limit difference of the actual range of fluorescence signal intensity of any two organs. The actual attenuation coefficient range of tissue density is compared to calculate the maximum difference, i.e., the maximum density difference. The calculation formula is: Maximum density difference = the maximum value of the upper limit difference or the maximum value of the lower limit difference of the actual attenuation coefficient range of tissue density of any two organs. The actual resonance frequency range of proton signal is compared to calculate the maximum difference, i.e., the maximum proton difference. The calculation formula is: Maximum proton difference = the maximum value of the upper limit difference or the maximum value of the lower limit difference of the actual resonance frequency range of proton signal of any two organs. These three maximum differences are summarized according to parameter type to obtain the first associated parameter set.

[0130] Step 512: Based on the organ-specific parameter ratio class dimension, calculate the median fluorescence value of the actual range of fluorescence signal intensity, the median density value of the actual range of tissue density attenuation coefficient, and the median proton value of the actual range of proton signal resonance frequency in each organ to determine the fluorescence density ratio and density-proton ratio. Summarize the fluorescence density ratio and density-proton ratio of each organ to obtain the second set of associated parameters.

[0131] In this step, the fluorescence median value refers to the median value of the actual range of fluorescence signal intensity for a single organ, representing the typical level of fluorescence signal intensity for that organ. The density median value refers to the median value of the actual range of tissue density attenuation coefficients for a single organ, representing the typical level of tissue density attenuation coefficients for that organ. The proton median value refers to the median value of the actual range of proton signal resonance frequencies for a single organ, representing the typical level of proton signal resonance frequencies for that organ. The fluorescence density ratio refers to the ratio of the fluorescence median value to the density median value for a single organ, reflecting the relative relationship between the organ's fluorescence and density characteristics. The density-proton ratio refers to the ratio of the density median value to the proton median value for a single organ, reflecting the relative relationship between the organ's density and proton signal characteristics.

[0132] In this embodiment, based on the organ-specific heterogeneous parameter ratio dimension, the median fluorescence value of the actual range of fluorescence signal intensity for each organ is determined using the following formula: Median fluorescence value = (Upper limit of actual range of fluorescence signal intensity + Lower limit of actual range of fluorescence signal intensity) ÷ 2. The median density value of the actual attenuation coefficient range of tissue density is also determined using the following formula: Median density value = (Upper limit of actual range of tissue density attenuation coefficient + Lower limit of actual range of tissue density attenuation coefficient) ÷ 2. Finally, the median proton value of the actual resonance frequency range of proton signals is determined using the following formula: Median proton value = (Upper limit of actual range of tissue density attenuation coefficient + Lower limit of actual range of tissue density attenuation coefficient) ÷ 2. The ratio of the fluorescence density ratio is calculated as follows: (upper limit of the actual resonance frequency range of the sub-signal + lower limit of the actual resonance frequency range of the proton signal) ÷ ​​2. The fluorescence density ratio is calculated as follows: fluorescence density ratio = fluorescence density ratio ÷ density density ratio. The density density ratio is calculated as follows: density density ratio = density density ratio ÷ proton density ratio. The fluorescence density ratio and density density ratio of each organ are summarized to obtain the second set of correlation parameters.

[0133] Step 513: Based on the inter-organ heterogeneous parameter trend class dimension, determine the first change direction of the fluorescence median value, the second change direction of the density median value, and the third change direction of the proton median value in two adjacent organs in the anatomical position sequence. Summarize the first change direction, the second change direction, and the third change direction to obtain the third set of associated parameters.

[0134] In this step, the anatomical positional order refers to the spatial arrangement of organs within a living subject, determined based on anatomical norms, such as the organ order from head to abdomen. This order clarifies the adjacency relationships between organs and provides a positional benchmark for trend analysis. The first direction of change refers to the trend of the median fluorescence value between two adjacent organs within the anatomical positional order, reflecting the variation pattern of fluorescence characteristics between adjacent organs. The second direction of change refers to the trend of the median density value between two adjacent organs within the anatomical positional order, reflecting the variation pattern of density characteristics between adjacent organs. The third direction of change refers to the trend of the median proton value between two adjacent organs within the anatomical positional order, reflecting the variation pattern of proton signal characteristics between adjacent organs.

[0135] In this embodiment, based on the inter-organ heterogeneous parameter trend dimension, anatomical data is consulted to determine the order of the organs in the living subject according to their anatomical location. Two adjacent organs are selected sequentially, and the median fluorescence value of the preceding organ and the median fluorescence value of the following organ are extracted. The values ​​of the two are compared; if the value of the following organ is greater than that of the preceding organ, it is marked as increased; otherwise, it is marked as decreased, thus obtaining the first direction of change. Using the same method, the median density values ​​of two adjacent organs are compared to obtain the second direction of change. The median proton values ​​of two adjacent organs are compared to obtain the third direction of change. The first, second, and third directions of change for each pair of adjacent organs are recorded in a table, and the records of all adjacent organ pairs are summarized to obtain the third set of associated parameters.

[0136] This application embodiment constructs a multi-dimensional set of associated parameters to comprehensively capture the differences, proportions, and trend correlations of organ features, thereby improving the depth and accuracy of liveness detection.

[0137] Figure 3 This is a schematic diagram of a specific embodiment of an imaging integration and analysis liveness detection system provided in this application, with reference to... Figure 3 The system may include:

[0138] The acquisition module 21 is used to acquire multimodal imaging data of multiple organs of a living object, including original fluorescence imaging images, original tissue density imaging images, and original proton density imaging images.

[0139] The calculation module 22 is used to calculate the fluorescence intensity distribution characteristics, tissue density value characteristics, and proton density signal characteristics based on the original fluorescence imaging image, the original tissue density imaging image, and the original proton density imaging image, respectively.

[0140] The association module 23 is used to perform cross-modal association processing on the fluorescence intensity distribution feature and the proton density signal feature with the tissue density value feature, respectively, to obtain a first association relationship between the fluorescence intensity distribution feature and the tissue density value feature, and a second association relationship between the proton density signal feature and the tissue density value feature.

[0141] The calibration module 24 is used to fuse the first association relationship and the second association relationship into parameters to obtain a cross-modal feature matching dataset, and to calibrate the parameters of the cross-modal feature matching dataset to obtain a calibrated cross-modal feature matching dataset.

[0142] Analysis module 25 is used to perform linkage analysis on the cross-modal features of each organ in the calibrated cross-modal feature matching dataset to obtain the live detection results.

[0143] This application provides a liveness detection system based on multimodal imaging integrated analysis to implement the aforementioned liveness detection method based on multimodal imaging integrated analysis. Therefore, the specific implementation of the liveness detection system based on multimodal imaging integrated analysis can be found in the embodiment section of the liveness detection method based on multimodal imaging integrated analysis above. The specific implementation can be referred to the description of the corresponding embodiments, which will not be repeated here.

[0144] This application also provides an electronic device, comprising: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of the liveness detection method based on multimodal imaging integrated analysis described above.

[0145] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of any of the above-described methods for liveness detection based on multimodal imaging integrated analysis.

[0146] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as USB flash drives, read-only memory, random access memory, portable hard drives, magnetic disks, or optical disks.

[0147] Embodiments of the present invention also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the above embodiments of the liveness detection method based on multimodal imaging integrated analysis.

[0148] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0149] The present application provides a detailed description of a liveness detection method and system based on multimodal imaging integration analysis. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and its core ideas. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from its principles, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A liveness detection method based on multimodal imaging integrated analysis, characterized in that, include: Acquire multimodal imaging data of multiple organs of a living subject, including original fluorescence imaging images, original tissue density imaging images, and original proton density imaging images; Based on the original fluorescence imaging image, the original tissue density imaging image, and the original proton density imaging image, the fluorescence intensity distribution characteristics, tissue density value characteristics, and proton density signal characteristics are calculated respectively. The fluorescence intensity distribution feature and proton density signal feature are respectively subjected to cross-modal correlation processing with the tissue density value feature to obtain the first correlation relationship between the fluorescence intensity distribution feature and the tissue density value feature, and the second correlation relationship between the proton density signal feature and the tissue density value feature; The first and second association relationships are fused to obtain a cross-modal feature matching dataset, and the cross-modal feature matching dataset is then calibrated to obtain a calibrated cross-modal feature matching dataset. The cross-modal features of each organ in the calibrated cross-modal feature matching dataset are analyzed in a linked manner to obtain the liveness detection results; The fluorescence intensity distribution feature and proton density signal feature are respectively subjected to cross-modal correlation processing with the tissue density value feature to obtain a first correlation relationship between the fluorescence intensity distribution feature and the tissue density value feature, and a second correlation relationship between the proton density signal feature and the tissue density value feature, including: Based on the organ region and tomographic region set, spatial coordinate matching is performed on the original fluorescence imaging image and the original tissue density imaging image. Under the matched spatial coordinates, the variation range, distribution range, and intensity difference between each pixel in the fluorescence intensity distribution feature of the same organ are respectively correlated with the average density, extreme density, and density dispersion in the tissue density feature. This yields the first fluorescence density coordinate parameter pair, the second fluorescence density coordinate parameter pair, and the third fluorescence density coordinate parameter pair for each organ, forming a first correlation relationship that includes spatial coordinate information and parameter correlation coefficients. Based on the proton signal region and the tomographic region set, spatial coordinate calibration is performed on the original proton density imaging image and the original tissue density imaging image. Within the calibrated spatial range, the average signal intensity, peak signal intensity, and signal coverage of the same organ in the proton density signal features are respectively correlated with the average density value, extreme density value, and density dispersion at the regional level to obtain the first proton density coordinate parameter pair, the second proton density coordinate parameter pair, and the third proton density coordinate parameter pair for each organ, so as to form a second correlation relationship that includes tomographic region information and parameter clustering features. For the average signal intensity, peak signal intensity, and signal coverage of the same organ in the proton density signal features, regional correlations are performed with the average density value, extreme density value, and density dispersion, respectively, to obtain a first pair of proton density coordinate parameters, a second pair of proton density coordinate parameters, and a third pair of proton density coordinate parameters for each organ, thus forming a second correlation relationship that includes tomographic region information and parameter clustering features, including: The overlapping areas of the proton signal regions and tomographic regions of each organ are marked as proton density correlation regions; The coordinate boundary of the proton density associated region is used as a spatial identifier. The spatial identifier is associated with the average signal intensity and average density of the same organ to obtain a first proton density coordinate parameter pair. The spatial identifier is associated with the signal peak and density extreme value of the same organ to obtain a second proton density coordinate parameter pair. The spatial identifier is associated with the signal coverage and density dispersion of the same organ to obtain a third proton density coordinate parameter pair. By integrating the first, second, and third proton density coordinate parameter pairs of the same organ, a proton density region association set is obtained. Based on the fault region numbering, the proton density region association sets of all organs are classified to obtain multiple sets of association data. All association data are then regionally aggregated to obtain the second association relationship.

2. The method according to claim 1, characterized in that, Based on the original fluorescence imaging image, the original tissue density imaging image, and the original proton density imaging image, the fluorescence intensity distribution characteristics, tissue density value characteristics, and proton density signal characteristics are calculated, including: In the original fluorescence imaging image, the organ regions where each organ is located are marked, the fluorescence signal intensity values ​​of the pixels in each organ region are counted, and the variation range, distribution range, and intensity difference between pixels of each fluorescence signal intensity value are calculated. The variation range, distribution range, and intensity difference corresponding to each organ are summarized to obtain the fluorescence intensity distribution characteristics. The original tissue density imaging image is subjected to tomographic correlation processing to determine the tomographic regions of each organ to obtain the tomographic region set of each organ. The average density, extreme density, and density dispersion of each tomographic region set are calculated. The average density, extreme density, and density dispersion of each organ are summarized to obtain the tissue density value features. The proton signal regions of each organ are marked in the original proton density imaging image, and the average signal intensity, peak signal intensity, and signal coverage of each proton signal region are calculated. The average signal intensity, peak signal intensity, and signal coverage of each organ are then summarized to obtain the proton density signal features.

3. The method according to claim 1, characterized in that, The first and second association relationships are fused to obtain a cross-modal feature matching dataset. The cross-modal feature matching dataset is then calibrated to obtain a calibrated cross-modal feature matching dataset, including: The first fluorescence density coordinate parameter pair, the second fluorescence density coordinate parameter pair, and the third fluorescence density coordinate parameter pair in the first association relationship of the same organ are integrated with the first proton density coordinate parameter pair, the second proton density coordinate parameter pair, and the third proton density coordinate parameter pair in the second association relationship to obtain the first cross-modal sub-feature set, the second cross-modal sub-feature set, and the third cross-modal sub-feature set, so as to generate a cross-modal feature matching dataset. Based on the cross-modal feature matching dataset, the actual range of fluorescence signal intensity, the actual range of tissue density attenuation coefficient, and the actual range of proton signal resonance frequency are determined. The first difference between the actual range of fluorescence signal intensity and the preset standard wavelength range, the second difference between the actual range of tissue density attenuation coefficient and the preset standard attenuation coefficient range, and the third difference between the actual range of proton signal resonance frequency and the preset standard resonance frequency range are calculated to perform multimodal parameter calibration on the cross-modal feature matching dataset, thereby obtaining the calibrated cross-modal feature matching dataset.

4. The method according to claim 1, characterized in that, A linked analysis is performed on the cross-modal features of each organ in the calibrated cross-modal feature matching dataset to obtain liveness detection results, including: Cross-modal feature parameters of each organ are extracted from the calibrated cross-modal feature matching dataset to determine the feature association dimensions between multiple organs. The feature association dimensions include the inter-organ same parameter difference dimension, the intra-organ heterogeneous parameter ratio dimension, and the inter-organ heterogeneous parameter trend dimension. Calculate the first set of correlation parameters for each organ under the dimension of the same parameter difference between organs, the second set of correlation parameters under the dimension of different parameter ratio within organs, and the third set of correlation parameters under the dimension of different parameter trends between organs, so as to form a set of organ linkage features. Based on organ category and feature dimension, the set of organ linkage features is classified and labeled to obtain a feature association map; The cross-modal feature parameters and correlation parameters of each organ in the feature association map are integrated to obtain the liveness detection results.

5. The method according to claim 4, characterized in that, Calculate the first set of correlation parameters for each organ under the inter-organ same parameter difference category dimension, the second set of correlation parameters under the intra-organ heterogeneous parameter ratio category dimension, and the third set of correlation parameters under the inter-organ heterogeneous parameter trend category dimension, including: Based on the inter-organ parameter difference class dimension, the maximum wavelength difference, the maximum density difference, and the maximum proton difference in the actual range of fluorescence signal intensity, tissue density attenuation coefficient, and proton signal resonance frequency range of any two organs are calculated. The maximum wavelength difference, maximum density difference, and maximum proton difference are then summarized to obtain the first set of associated parameters. Based on the organ-specific parameter ratio category, the median fluorescence value of the actual range of fluorescence signal intensity, the median density value of the actual range of tissue density attenuation coefficient, and the median proton value of the actual range of proton signal resonance frequency are calculated for each organ to determine the fluorescence density ratio and density-proton ratio. The fluorescence density ratio and density-proton ratio of each organ are then summarized to obtain the second set of associated parameters. Based on the inter-organ heterogeneous parameter trend class dimension, the first change direction of the fluorescence median value, the second change direction of the density median value, and the third change direction of the proton median value in two adjacent organs in the anatomical position sequence are determined. The first change direction, the second change direction, and the third change direction are summarized to obtain the third set of associated parameters.

6. A liveness detection system based on multimodal imaging integrated analysis, characterized in that, include: The acquisition module is used to acquire multimodal imaging data of multiple organs of a living object, including original fluorescence imaging images, original tissue density imaging images, and original proton density imaging images. The calculation module is used to calculate the fluorescence intensity distribution characteristics, tissue density value characteristics, and proton density signal characteristics based on the original fluorescence imaging image, the original tissue density imaging image, and the original proton density imaging image, respectively. The correlation module is used to perform cross-modal correlation processing on the fluorescence intensity distribution feature and the proton density signal feature with the tissue density value feature, respectively, to obtain a first correlation relationship between the fluorescence intensity distribution feature and the tissue density value feature, and a second correlation relationship between the proton density signal feature and the tissue density value feature; The calibration module is used to fuse the first association relationship and the second association relationship into parameters to obtain a cross-modal feature matching dataset, and to calibrate the parameters of the cross-modal feature matching dataset to obtain a calibrated cross-modal feature matching dataset. The analysis module is used to perform a linked analysis of the cross-modal features of each organ in the calibrated cross-modal feature matching dataset to obtain the liveness detection results. The fluorescence intensity distribution feature and proton density signal feature are respectively subjected to cross-modal correlation processing with the tissue density value feature to obtain a first correlation relationship between the fluorescence intensity distribution feature and the tissue density value feature, and a second correlation relationship between the proton density signal feature and the tissue density value feature, including: Based on the organ region and tomographic region set, spatial coordinate matching is performed on the original fluorescence imaging image and the original tissue density imaging image. Under the matched spatial coordinates, the variation range, distribution range, and intensity difference between each pixel in the fluorescence intensity distribution feature of the same organ are respectively correlated with the average density, extreme density, and density dispersion in the tissue density feature. This yields the first fluorescence density coordinate parameter pair, the second fluorescence density coordinate parameter pair, and the third fluorescence density coordinate parameter pair for each organ, forming a first correlation relationship that includes spatial coordinate information and parameter correlation coefficients. Based on the proton signal region and the tomographic region set, spatial coordinate calibration is performed on the original proton density imaging image and the original tissue density imaging image. Within the calibrated spatial range, the average signal intensity, peak signal intensity, and signal coverage of the same organ in the proton density signal features are respectively correlated with the average density value, extreme density value, and density dispersion at the regional level to obtain the first proton density coordinate parameter pair, the second proton density coordinate parameter pair, and the third proton density coordinate parameter pair for each organ, so as to form a second correlation relationship that includes tomographic region information and parameter clustering features. For the average signal intensity, peak signal intensity, and signal coverage of the same organ in the proton density signal features, regional correlations are performed with the average density value, extreme density value, and density dispersion, respectively, to obtain a first pair of proton density coordinate parameters, a second pair of proton density coordinate parameters, and a third pair of proton density coordinate parameters for each organ, thus forming a second correlation relationship that includes tomographic region information and parameter clustering features, including: The overlapping areas of the proton signal regions and tomographic regions of each organ are marked as proton density correlation regions; The coordinate boundary of the proton density associated region is used as a spatial identifier. The spatial identifier is associated with the average signal intensity and average density of the same organ to obtain a first proton density coordinate parameter pair. The spatial identifier is associated with the signal peak and density extreme value of the same organ to obtain a second proton density coordinate parameter pair. The spatial identifier is associated with the signal coverage and density dispersion of the same organ to obtain a third proton density coordinate parameter pair. By integrating the first, second, and third proton density coordinate parameter pairs of the same organ, a proton density region association set is obtained. Based on the fault region numbering, the proton density region association sets of all organs are classified to obtain multiple sets of association data. All association data are then regionally aggregated to obtain the second association relationship.

7. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor, configured to execute the computer program to implement the steps of a liveness detection method based on multimodal imaging integrated analysis as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, enables a liveness detection method based on multimodal imaging integrated analysis as described in any one of claims 1 to 5.

Images