Optimization method and system for multi-light source illumination of brain gray matter vascular photoacoustic imaging and storage medium
Patent Information
- Application Number
- CN202611385048.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-08
- Publication Date
- 2026-10-13
AI Technical Summary
[0006]本发明的目的在于提供一种面向脑灰质血管光声成像的多光源照明配置优化方法、系统及存储介质，以解决现有技术中光源配置多依赖规则模型或固定照明方式、难以评价多光源组合在真实解剖结构下适用性的问题
[0020]第一，在单光源和双光源总输入能量一致的条件下评价不同照明配置，避免将双光源响应增强误认为输入能量增加的结果；
Smart Images
Figure CN122904548A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the technical fields of photoacoustic imaging, medical image simulation and light source layout optimization, specifically involving a method, system and storage medium for optimizing multi-source illumination configuration for photoacoustic imaging of brain gray matter vessels. Background Technology
[0002] Photoacoustic imaging is an imaging technique that combines optical absorption contrast with the penetrating power of ultrasound. When biological tissue absorbs pulsed light energy, it undergoes transient thermoelastic expansion, which excites an ultrasound signal. By detecting this ultrasound signal, the distribution of light absorption within the tissue can be indirectly reflected. Because hemoglobin exhibits significant light absorption properties, cerebral blood vessels are one of the important targets for photoacoustic imaging.
[0003] In photoacoustic imaging of cerebral vessels, incident light needs to penetrate multiple layers of tissue, including the scalp, skull, cerebrospinal fluid, and brain parenchyma. The different absorption, scattering, and refraction characteristics of these tissues cause the light flux to attenuate, diffuse, and redistribute during propagation, resulting in weaker effective light deposition at deep gray matter vascular targets. Simultaneously, superficial tissues may generate strong background photoacoustic sources, thus affecting the relative response of deep targets.
[0004] Existing light source placement methods are mostly based on single-source lighting, fixed incident geometry, or regular layered brain tissue models. While regular layered models facilitate rapid calculations and parametric scanning, real brain tissue exhibits irregular boundaries, local thickness variations, and spatial heterogeneity. Therefore, optimal illumination configurations obtained in regular models may not be directly applicable to anatomical brain models. Furthermore, existing methods often lack a unified evaluation of the stability of multi-source combinations, incident angles, incident directions, and model migration under conditions where the total input energy of a single or dual-source light source is consistent.
[0005] Therefore, there is a need for a multi-source illumination configuration optimization method that can generate and evaluate multiple light source configurations under uniform energy conditions, convert light propagation results into initial photoacoustic sound pressure source terms, and perform transfer verification using simplified models, different target depths, and anatomical brain models. Summary of the Invention
[0006] The purpose of this invention is to provide a method, system, and storage medium for optimizing multi-source illumination configuration in photoacoustic imaging of brain gray matter vessels, in order to solve the problems in the prior art where the light source configuration often depends on rule models or fixed illumination methods, and it is difficult to evaluate the applicability of multi-source combinations in real anatomical structures.
[0007] To achieve the above objectives, the present invention provides the following technical solution.
[0008] Construct an optical simulation model of brain tissue and a target region for gray matter vessels. The optical simulation model of brain tissue includes a regular layered brain tissue model and an anatomical brain tissue model; the regular layered model may include tissue layers such as scalp, skull, cerebrospinal fluid, and gray matter, while the anatomical brain tissue model can be obtained from brain tissue probabilistic maps or medical image segmentation.
[0009] Generate a set of candidate lighting configurations. Candidate light source types include 16 types such as hyperbolic Gaussian light sources, pencil array light sources, collimated Gaussian light sources, and surface light sources. Candidate lighting configurations include single-source vertical incidence, single-source oblique incidence, dual-source symmetrical incidence, and dual-source symmetrical incidence, with the ability to further set different incidence angles and directions, and also to place a larger number of light sources.
[0010] Monte Carlo optical transmission simulations were performed under the condition of consistent total input energy. The weight of the single-source configuration was set to 1, and the two beams in the dual-source configuration were simulated independently and then linearly superimposed with weights of 0.5 and 0.5 respectively, so that the total input energy of the dual-source configuration was consistent with that of the single-source configuration.
[0011] The relative photoacoustic initial sound pressure source term is calculated based on the luminous flux distribution obtained from Monte Carlo simulation. The absorbed energy density satisfies:
[0012] The relative photoacoustic initial sound pressure source term satisfies:
[0013] in Let Γ be the absorption coefficient, Φ(r) be the luminous flux, and Γ(r) be the Grünnesson parameter.
[0014] Calculate the evaluation metrics. The evaluation metrics include the average initial sound pressure in the target area, the proportion of the initial sound pressure in the target area to the total initial sound pressure in the entire model, the proportion of the initial sound pressure in the gray area to the total initial sound pressure in the entire model, the contrast-to-noise ratio of the initial sound pressure in the target area to the background area, the Michelson contrast ratio of the target area to the background area, the low-threshold penetration depth, and the full width at half maximum (FWHM) of the light field at the target depth.
[0015] Construct a comprehensive score. Normalize the indicators that indicate better performance with larger values, and perform inverse minimum-maximum normalization on the half-height and full-width indicators. Then, sum the normalized indicators according to preset weights to obtain the comprehensive score for each candidate lighting configuration.
[0016] Target depth validation and model transfer validation are performed. The stability of candidate configurations is validated at multiple target depths; further, the candidate configurations are transferred to an anatomical brain tissue model, gray matter vascular targets are constructed at different local locations, and comprehensive scores and cross-location stability are calculated to output recommended multi-source illumination configurations.
[0017] The present invention also provides a system for implementing the above method, comprising a model building module, a candidate configuration generation module, a Monte Carlo simulation module, an initial sound pressure calculation module, an index calculation module, a comprehensive scoring module, and a configuration output module. Each module can be implemented by a computer program executed in a processor.
[0018] The present invention also provides an electronic device and a computer-readable storage medium, the electronic device including a processor and a memory, wherein a computer program stored in the memory implements the above-described method when executed by the processor; and a computer program stored in the computer-readable storage medium implements the above-described method when executed by the processor.
[0019] Compared with the prior art, the present invention has at least the following beneficial effects:
[0020] First, evaluate different lighting configurations under the condition that the total input energy of single and dual light sources is the same, so as to avoid mistaking the enhanced response of dual light sources as the result of increased input energy.
[0021] Second, the Monte Carlo optical transmission results are transformed into relative photoacoustic initial sound pressure source terms, so that the light source optimization is directly geared towards the generation of photoacoustic source terms;
[0022] Third, the comprehensive score is based on multi-index normalization, which simultaneously considers target enhancement, gray matter deposition, background contrast, penetration ability, and light field width.
[0023] Fourth, by using brain models with different target depths and anatomical structures for transfer verification, we can improve the ability to judge the applicability of candidate lighting configurations under real brain structure conditions.
[0024] Fifth, it can provide parameter basis for light source layout, light flux compensation, shallow background suppression and subsequent image reconstruction in cerebral vascular photoacoustic imaging systems. Attached Figure Description
[0025] Figure 1 This is a flowchart illustrating the multi-source lighting configuration optimization method described in this invention.
[0026] Figure 2 This is a schematic diagram of the process for generating the candidate lighting configuration set according to the present invention.
[0027] Figure 3 This is a schematic diagram of the module structure of the multi-source lighting configuration optimization system described in this invention.
[0028] Figure 4 This is a schematic diagram illustrating the simplified model selection, target depth verification, and anatomical model transfer verification process described in this invention. Detailed Implementation
[0029] The specific embodiments of the present invention will now be described with reference to the accompanying drawings. It should be understood that the following embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.
[0030] like Figure 1 As shown, the method of the present invention first obtains or constructs an optical simulation model of brain tissue. The model can be a regular layered brain tissue model, and the total physical size and voxel size of the model are set under the premise that photons do not overflow, and then the corresponding thickness is assigned to each layer of tissue; alternatively, an anatomical brain tissue model obtained from brain tissue probability maps or medical image segmentation results can be used. Such models are generally directly imported from external sources.
[0031] Each voxel is assigned a tissue label, and the absorption coefficient, scattering coefficient, anisotropy factor, refractive index, and Grunerson parameter are set according to the tissue label. Gray matter blood vessel targets can be set as cylindrical or approximately cylindrical target regions extending along a predetermined direction.
[0032] like Figure 2 As shown, the candidate illumination configuration set is determined by the light source type, incident angle, incident direction, and combination method. Light source types may include hyperbolic Gaussian sources, pencil array sources, collimated Gaussian sources, and surface sources. Combination methods include single-source perpendicular incidence, single-source oblique incidence, dual-source symmetrical incidence, and dual-source symmetrical incidence. For a dual-source configuration, the two beams are located on opposite sides of the target and both point towards the center of the blood vessel target.
[0033] In one embodiment, the weight of the single-source configuration is 1; the two beams of the dual-source configuration are subjected to Monte Carlo simulation independently to obtain the luminous flux distributions Φ1(r) and Φ2(r), and the total luminous flux distribution is calculated according to the following formula: Φ total (r) = 0.5Φ1(r) + 0.5Φ2(r)
[0034] Therefore, the total input energy of the dual-source configuration remains the same as that of the single-source configuration, and the subsequent differences are mainly determined by the source type, incident angle, incident direction, and tissue propagation path.
[0035] In one embodiment, the initial sound pressure calculation module calculates the absorbed energy density based on the luminous flux distribution and further calculates the initial photoacoustic sound pressure. In this invention, the initial photoacoustic sound pressure is used for relative comparisons between different candidate illumination configurations and should not be interpreted as an absolute sound pressure in actual Pa units.
[0036] In one embodiment, the index calculation module extracts the vascular target region, the gray matter region, and the background region. The average initial sound pressure of the target region is used to represent the average photoacoustic source intensity inside the vascular target; the initial sound pressure ratio of the target region is used to represent the contribution of the vascular target to the initial sound pressure source term of the entire model; the initial sound pressure ratio of the gray matter region is used to represent the overall deposition of the tissue layer where the target is located; the contrast-to-noise ratio of the initial sound pressure of the target to the background is used to represent the distinguishability between the target and the background; the penetration depth is used to represent the low-threshold propagation capability of the light field; and the full width at half maximum (FWHM) is used to represent the lateral coverage of the light field at the depth of the target.
[0037] In one embodiment, the comprehensive scoring module normalizes multiple indicators. For indicators where larger values indicate better performance, min-max normalization is used; for indicators with full width at half maximum (FWHM) and full width at half maximum (WWHM), inverse min-max normalization is used. After normalization, the indicators are weighted and summed according to preset weights to obtain the comprehensive score. The preset weights can be adjusted according to the needs of target enhancement, penetration priority, contrast priority, or balance.
[0038] like Figure 4 As shown, model transfer validation can include three stages: first, large-scale candidate configuration screening is performed in a regular layered brain tissue model; second, the stability of the candidate configurations is verified at multiple target depths; finally, representative candidate configurations are applied to the anatomical brain tissue model, and gray matter blood vessel targets are constructed at different local locations, comparing the comprehensive scores and candidate configuration rankings within each location. The final output is a recommended multi-source illumination configuration and its corresponding evaluation results.
[0039] In the anatomical brain tissue model, the method for constructing gray matter vessels is as follows: First, the initial light source position is set with an incident direction of [-1, 0, -1], allowing it to propagate into the gray matter region along the XZ plane at a 45° oblique incidence path. Subsequently, vessel implantation points are determined within the gray matter region of the XZ plane corresponding to this incident path, and vessel targets extending along the Y direction are constructed. The vessels have a circular cross-section in the XZ plane and automatically fill a continuous gray matter region along the Y direction, approximating an overall cylindrical shape. After completing the vessel label construction, the geometric center of the entire vessel region is calculated and used as the unified focusing target for each representative light source configuration.
[0040] like Figure 3 As shown, the system of the present invention can be deployed in a computer device with a graphics processor. The model building module, candidate configuration generation module, Monte Carlo simulation module, initial sound pressure calculation module, index calculation module, comprehensive scoring module, and configuration output module can be implemented by software programs, hardware circuits, or a combination of software and hardware.
[0041] The above embodiments are merely preferred embodiments of the present invention. Those skilled in the art can adjust or replace the model type, light source type, index weight, target depth, and verification position without departing from the principle of the present invention, and such adjustments or replacements should all fall within the protection scope of the present invention.
Claims
1. A method for optimizing multi-source illumination configuration for photoacoustic imaging of brain gray matter vessels, characterized in that, Includes the following steps: Construct an optical simulation model of brain tissue and a target region of gray matter blood vessels. The optical simulation model of brain tissue includes a regular layered brain tissue model and an anatomical brain tissue model. Generate a candidate lighting configuration set, which includes a single light source configuration, a dual isomorphic light source symmetrical incidence configuration, and a dual hybrid light source symmetrical incidence configuration. Under the condition that the total input energy of single-source and dual-source configurations is the same, Monte Carlo light transmission simulation is performed on each lighting configuration in the candidate lighting configuration set to obtain the corresponding luminous flux distribution. The relative photoacoustic initial sound pressure source term is calculated based on the tissue absorption coefficient, light flux distribution, and Grunessen parameter (Γ). Evaluation metrics are constructed based on the average initial sound pressure of the target region, the proportion of the initial sound pressure of the gray region to the total initial sound pressure of the whole model, the contrast noise ratio between the initial sound pressure signals of the target region and the background region, the penetration depth when the normalized light flux drops to 0.1%, and the full width at half maximum (FWHM) of the light field at the target depth. The evaluation indicators are dimensionless and weighted to obtain a comprehensive score; based on the comprehensive score and the results of individual indicators, a multi-source lighting configuration is recommended.
2. The method according to claim 1, characterized in that, The regular layered brain tissue model includes a scalp layer, a skull layer, a cerebrospinal fluid layer, and a gray matter layer arranged along the depth direction, and a vascular target region extending in a predetermined direction is set in the gray matter layer; the anatomical brain tissue model is obtained by brain tissue probability mapping or medical image segmentation results, and gray matter, white matter, cerebrospinal fluid, skull, scalp, and implanted vascular target are represented in the form of tissue labels.
3. The method according to claim 1, characterized in that, The light source types in the candidate lighting configuration set include at least two of the following: pencil light source, isotropic light source, conical light source, arcsine light source, line light source, slit light source, collimated Gaussian light source, angular Gaussian light source, hyperboloid Gaussian light source, planar light source, disk light source, ring light source, pencil array light source, spatial frequency Fourier light source, one-dimensional Fourier light source, and two-dimensional Fourier light source; the candidate lighting configuration set further includes lighting configurations formed by combinations of different incident angles, different incident directions, and different light source types.
4. The method according to claim 1, characterized in that, The weight of the single light source configuration is set to 1; the two beams in the dual identical light source symmetrical incident configuration and the dual hybrid light source symmetrical incident configuration are simulated independently and then linearly superimposed according to their weights, and the weights of the two beams are set to 0.5 respectively, so that the total input energy of the dual light source configuration is consistent with that of the single light source configuration.
5. The method according to claim 1, characterized in that, The relative photoacoustic initial sound pressure source term is determined according to equation (1): in, Let be the absorption coefficient at the r-th voxel. Let Γ(r) be the luminous flux at the r-th voxel, Γ(r) be the Grünerson parameter at the r-th voxel, and P0(r) be the relative photoacoustic initial sound pressure source term at the r-th voxel.
6. The method according to claim 1, characterized in that, The evaluation indicators include: The average initial sound pressure in the target area is denoted as TRMIP, which reflects the average initial sound pressure intensity of a unit voxel inside the vascular target. The average initial sound pressure in the target area is determined according to equation (2): Where ROIt represents the target vascular region, and Nt represents the voxel count of the target vascular region. The initial photoacoustic pressure at the r-th voxel; The percentage of the initial sound pressure in the target region to the total initial sound pressure of the entire model is denoted as TRIPP, which reflects the contribution of the vascular target to the background initial sound pressure of the entire region. The percentage of the initial sound pressure in the target region to the total initial sound pressure of the entire model is determined according to equation (3): in, For the entire organizational model area; The proportion of the initial sound pressure in the gray matter region to the total initial sound pressure of the entire model is denoted as GMIPP. It is used to reflect the contribution ratio of the gray matter region to the source term of the initial sound pressure of the entire model and to evaluate whether the light energy is effectively deposited to the tissue layer where the target is located. The proportion of the initial sound pressure in the gray matter region to the total initial sound pressure of the entire model is determined according to Equation (4): in, This is a gray region; The initial sound pressure signal contrast-to-noise ratio (IPCNR) between the target region and the background region is denoted as IPCNR. It is used to reflect the magnitude of the difference between the target region and the background region relative to the fluctuation within the region. The larger the value, the easier it is for the target to be distinguished from the background in the initial pressure distribution. The initial sound pressure signal contrast-to-noise ratio between the target region and the background region is determined according to equation (5): in, The average initial sound pressure level in the target vascular region. The average initial sound pressure level in the background region. The number of voxels in the background region. The background area is selected as the non-vascular tissue area surrounding the vascular target, specifically the gray matter ring background area within a certain distance outside the blood vessel. The Michelson contrast ratio between the target region and the background region is denoted as IPMC, which represents the relative initial pressure difference between the target region and the background region of the blood vessel. The Michelson contrast ratio between the target region and the background region is determined according to equation (6): when Greater than When the value is higher, the higher the contrast of the target relative to the background; The penetration depth at which the normalized luminous flux drops to 0.1% is denoted as PD0.1%, and the penetration depth at which the normalized luminous flux drops to 0.1% is determined according to equation (7): in, The depth-normalized luminous flux curve; The full width at half maximum (FWHM) of the light field at the target depth is denoted as The full width at half maximum (FWHM) of the light field at the target depth is used to reflect the horizontal distribution characteristics of the illumination light field at the target depth, and is determined according to equation (8): Where x1 and x2 are the two positions corresponding to the lateral luminous flux profile at the depth of the target center dropping to half of the peak value; To uniformly rank different light source configurations, a comprehensive score is constructed. The score comprehensively considers five normalized indicators (TRMIP, GMIPP, PD0.1%, FWHM, and IPCNR), and the comprehensive score is in the form of formula (9): Where N represents the normalization of each indicator to the range of 0–1; w1, w2, w3, w4, and w5 are the weight values of each item; The overall score can be calculated by weighting any of the seven indicators above according to actual needs.
7. The method according to claim 1, characterized in that, The dimensionless processing includes normalizing the indicators with larger values that indicate better performance, performing inverse normalization on the full width at half maximum (FWHM) of the light field at the target depth, and then weighting and summing the normalized indicators according to preset weights to obtain a comprehensive score.
8. The method according to claim 1, characterized in that, It also includes a depth validation step and a model transfer validation step; the depth validation step is used to calculate the evaluation index and comprehensive score of candidate lighting configurations at multiple target depths; the model transfer validation step is used to apply the candidate lighting configurations selected from the regular layered brain tissue model to the anatomical brain tissue model, and determine the final recommended configuration based on the comprehensive score and cross-location stability within the anatomical brain tissue model.
9. A multi-source illumination configuration optimization system for photoacoustic imaging of brain gray matter vessels, characterized in that, Includes the following modules: The model building module is used to build optical simulation models of brain tissue and target regions of gray matter blood vessels; The candidate configuration generation module is used to generate candidate lighting configurations for single light source, dual identical light source, and dual hybrid light source; The Monte Carlo simulation module is used to obtain the luminous flux distribution corresponding to each lighting configuration. The initial sound pressure calculation module is used to calculate the relative photoacoustic initial sound pressure source term; The index calculation module is used to calculate the initial sound pressure, contrast, penetration depth, and light field width of the region. The comprehensive scoring module is used to normalize the indicators and perform weighted summation; Configure the output module to output the recommended lighting configuration.
10. An electronic device or computer-readable storage medium, characterized in that, The electronic device includes a processor and a memory, wherein the memory or the computer-readable storage medium stores a computer program that, when executed by the processor, implements the method according to any one of claims 1 to 8.