A method, system and storage medium for label-free imaging of blood vessels in response

Abstract

The application discloses a kind of response blood vessel label-free imaging method, system and storage medium, it is related to biomedical optical imaging and hemodynamics detection technical field.The method is under the condition of no exogenous optical contrast agent, the label-free optical image sequence formed during the target region of target living body under the action of predetermined modulation condition is collected, and the image acquisition time and modulation time reference are obtained;According to the image sequence, the reflection or scattering intensity characteristic sequence of each pixel position and / or speckle dynamic blood flow characteristic sequence are calculated;According to modulation time reference, determine target frequency and phase reference, carry out complex exponential projection or orthogonal demodulation at target frequency to the optical characteristic sequence of each pixel position, obtain pixel-level response coefficient.The application can extract the blood vessel that responds from static blood vessel, non-response blood vessel and non-target frequency background, and is suitable for reflection imaging, laser speckle imaging and fusion imaging.

Description

Technical Field

[0001] This invention relates to the fields of biomedical optical imaging and hemodynamic detection technology, specifically to a method, system, and storage medium for responding to label-free vascular imaging. Background Technology

[0002] Parameters such as vascular morphology, blood flow velocity, blood volume, blood oxygenation status, and vessel diameter can reflect the perfusion status and physiological function changes of living tissues. Under periodic stimulation, task induction, physiological modulation, drug effects, or other predetermined modulation conditions, some vessels within a target area may exhibit hemodynamic responses related to the modulation conditions. For example, during neurovascular coupling induced by neural activity, local blood flow velocity, blood volume, oxyhemoglobin concentration, and deoxyhemoglobin concentration may change; when peripheral tissues are subjected to electrical, mechanical, tactile, thermal, or other stimuli, the blood vessels in the corresponding areas may also show changes in blood flow status or optical characteristics. Therefore, imaging the specific vessels that respond under the modulation conditions helps in observing and analyzing tissue functional activity, blood flow regulation, and vascular response processes.

[0003] Existing vascular imaging technologies mainly include laser speckle imaging, optical coherence tomography (OCT), fluorescence imaging, two-photon imaging, confocal imaging, photoacoustic imaging, ultrasound imaging, magnetic resonance imaging (MRI), and computed tomography (CT). These technologies can obtain information on vascular structure, blood flow velocity, blood oxygenation status, or tissue perfusion from different angles. Specifically, laser speckle imaging utilizes the speckle decorrelation phenomenon caused by moving erythrocytes to obtain relative blood flow information; OCT can obtain vascular structural information with a certain depth resolution; fluorescence imaging, two-photon imaging, and confocal imaging can obtain vascular images with high spatial resolution; and photoacoustic imaging, ultrasound imaging, MRI, and CT can be used for vascular or perfusion imaging at different depths and scales. However, some existing technologies require the use of exogenous fluorescent dyes, contrast agents, or microbubble contrast agents, thus being limited by the drug administration process, contrast agent metabolism, repeat testing conditions, and clinical applicability. Some imaging technologies also suffer from slow imaging speed, small field of view, short working distance, the need for contact detection, complex equipment structure, or insufficient spatial resolution. While near-infrared functional imaging can reflect tissue function based on changes in the concentrations of oxyhemoglobin and deoxyhemoglobin, it typically yields regional functional distributions with low spatial resolution, making it difficult to directly identify specific responding vessels. Existing label-free reflectance imaging or laser speckle imaging usually outputs surface structure maps, vascular structure maps, speckle contrast maps, or relative blood flow maps of the target area. These images typically contain information on static vessels, non-responsive vessels, tissue background, illumination inhomogeneities, physiological fluctuations caused by heartbeat or respiration, and device drift. When only some vessels in the target area respond to predetermined modulation conditions, the response signal may be superimposed on a strong static structural background and non-target frequency fluctuations, making it difficult to directly identify specific responding vessels from ordinary vascular maps or relative blood flow maps. Summary of the Invention

[0004] The purpose of this invention is to provide a method, system, and storage medium for responding to label-free vascular imaging, in order to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A method for responding to label-free vascular imaging includes:

[0007] S1. Acquire a label-free optical image sequence formed in the target area of ​​the target living body during the operation of a predetermined modulation condition, an acquisition time sequence corresponding to each frame of the label-free optical image sequence, and a modulation time reference characterizing the predetermined modulation condition. The label-free optical image sequence is formed based on the endogenous optical hemodynamic signal of the target area without the use of an exogenous optical contrast agent.

[0008] S2. For each pixel position in the labelless optical image sequence, perform feature calculation on the image data of the corresponding pixel position at different acquisition times to obtain a pixel-level optical feature sequence F(x,y,n). The pixel-level optical feature sequence includes at least one of the reflection or scattering intensity feature sequence and speckle dynamic blood flow feature sequence.

[0009] S3. Determine the target frequency corresponding to the predetermined modulation conditions based on the modulation time reference; when it is necessary to obtain the response phase, further determine the phase reference;

[0010] S4. For each pixel position, based on the pixel-level optical feature sequence, acquisition time sequence, target frequency and phase reference of the corresponding pixel position, perform complex exponential projection operation at the target frequency or orthogonal demodulation operation mathematically equivalent to complex exponential projection operation to obtain the response coefficient of the corresponding pixel position at the target frequency.

[0011] S5. Determine the response intensity of the corresponding pixel position based on the amplitude of the response coefficient, and map the response intensity according to the spatial coordinates of each pixel position to generate a response blood vessel image and / or response blood vessel mask for highlighting the blood vessels that respond at the target frequency.

[0012] Specifically, intrinsic optical hemodynamic signals include at least one of the following:

[0013] Changes in absorption intensity caused by variations in the concentrations of oxyhemoglobin and deoxyhemoglobin;

[0014] Changes in reflection or scattering intensity caused by changes in blood volume, blood vessel diameter, or tissue scattering state;

[0015] Dynamic changes in speckle patterns caused by changes in the motility of red blood cells or changes in blood flow velocity.

[0016] Specifically, when the pixel-level optical feature sequence includes a reflection or scattering intensity feature sequence, the original intensity I(x,y,n) at pixel position (x,y) in the nth frame image is subjected to dark field subtraction and reference illumination normalization to obtain the corrected reflection or scattering intensity: Where D(x,y) is the dark field intensity at pixel position (x,y) in the dark field image. The reference intensity is used for the pixel position (x,y) in the reference illumination image, and the corrected reflection or scattering intensity is used as the reflection or scattering intensity feature sequence.

[0017] Specifically, the illumination light used to form the reflection or scattering intensity characteristic sequence is at least one of incoherent light, low-coherence light, or partially coherent light, and the target area is simultaneously or time-divisionally illuminated using one or more different illumination bands.

[0018] Specifically, when the pixel-level optical feature sequence includes a speckle dynamic blood flow feature sequence, coherent illumination is used to illuminate the target area. For the nth frame image, a local window W is selected centered at pixel position (x,y), and the average gray value of the pixels within the local window is calculated. and pixel grayscale standard deviation The speckle contrast is calculated using the following formula: And based on the speckle contrast, a speckle dynamic blood flow characteristic sequence is formed.

[0019] Specifically, relative blood flow parameters are calculated based on speckle contrast: And by performing a time-averaged calculation on the relative blood flow parameters across multiple consecutive frames, including the current frame, we obtain: Where H is the number of frames included in the time-averaged window before or after the current frame, and the size of the local window W is 3×3, 5×5, 7×7 or other preset size.

[0020] Specifically, the first response vascular image and the second response vascular image are calculated based on the reflection or scattering intensity feature sequence and the speckle dynamic blood flow feature sequence, respectively. The first response vascular image and the second response vascular image are then spatially registered and fused to obtain the fused response vascular image.

[0021] Alternatively, the reflection or scattering intensity feature sequence and the speckle dynamic blood flow feature sequence can be fused before performing complex exponential projection operation or orthogonal demodulation operation at the target frequency.

[0022] Specifically, the same synchronous trigger signal is used to control the start of the predetermined modulation conditions and the start of image acquisition, or the timestamps of the predetermined modulation conditions and image acquisition are recorded separately and time-aligned, so that the acquisition time sequence has a definite time correspondence with the modulation time reference.

[0023] Specifically, the predetermined modulation conditions repeat periodically at the target frequency f1, the acquisition frame rate is f0, and the modulation period is:

[0024] ;

[0025] When the number of acquisition frames corresponding to each modulation cycle is an integer, the number of acquisition frames L in each modulation cycle and the total number of acquisition frames M after N modulation cycles are respectively:

[0026] ;

[0027] ;

[0028] The acquisition time of the nth frame image is:

[0029] ;

[0030] Where n = 0, 1, ..., M-1.

[0031] Specifically, at the end of the m-th modulation period, the M data collected up to the current time are... m =mL frames of image are subjected to one cumulative target frequency extraction, where the response coefficient at the corresponding pixel location is:

[0032] ;

[0033] The response intensity at the corresponding pixel location is:

[0034] ;

[0035] Here, φ0 is a phase reference determined according to the modulation time base, and the displayed vascular image is updated at the end of each modulation cycle.

[0036] Specifically, the orthogonal demodulation operation mathematically equivalent to the complex exponential projection operation includes:

[0037] Calculate the in-phase components:

[0038] ;

[0039] Calculate orthogonal components:

[0040] ;

[0041] The response intensity is calculated according to the following formula:

[0042] ;

[0043] Specifically, after completing a preset number of modulation cycles, the response vascular image obtained at the end of the last modulation cycle is taken as the final response vascular image:

[0044] ;

[0045] The response intensity of each pixel location is compared with a preset response threshold. Pixels whose response intensity meets the preset response threshold condition are identified as responsive blood vessel pixels to generate a responsive blood vessel mask.

[0046] Specifically, the predetermined modulation conditions include a stimulation phase and a non-stimulation phase. The target frequency is determined based on the repetition cycle formed by the stimulation phase and the non-stimulation phase, so that the static vascular component, the non-responsive vascular component, and the heartbeat component, respiratory component, and low-frequency drift component that deviate from the target frequency are suppressed during the target frequency extraction process.

[0047] Specifically, the predetermined modulation conditions include at least one of electrical stimulation, mechanical stimulation, tactile stimulation, visual stimulation, auditory stimulation, sound stimulation, thermal stimulation, physiological modulation, task induction, and drug action;

[0048] The target area includes at least one of the following: brain, spinal cord, retina, hand, foot, limb, tumor area, inflamed area, or other living tissue area.

[0049] Specifically, the illumination light used to form the reflection or scattering intensity characteristic sequence includes incoherent, low-coherence, or partially coherent illumination light with a wavelength of 660 nm and / or incoherent, low-coherence, or partially coherent illumination light with a wavelength of 850 nm;

[0050] The illumination light used to form the speckle dynamic blood flow characteristic sequence includes coherent illumination light with a wavelength of 850 nm;

[0051] The lighting power density is from 1mW / cm² to 200mW / cm², preferably from 10mW / cm² to 60mW / cm²;

[0052] The exposure time for image acquisition is 1ms to 1000ms, preferably 10ms to 100ms, and the acquisition frame rate is 1Hz to 1000Hz, preferably 10Hz to 100Hz, or other exposure times and acquisition frame rates can be set according to the target frequency and the required time resolution.

[0053] The present invention also provides a system applicable to the label-free imaging method for responding to blood vessels as described above, comprising:

[0054] An illumination module is used to illuminate the target area of ​​a living organism with a first illumination light that does not produce resolvable speckle and / or a second illumination light that produces speckle, without the use of an external optical contrast agent.

[0055] The image acquisition module is used to acquire a sequence of unmarked optical images of the target area during the operation of predetermined modulation conditions, and to record the acquisition time corresponding to each frame of the image.

[0056] The synchronization control module is used to receive a modulation time reference characterizing predetermined modulation conditions, or to output control signals to the modulation device used to generate predetermined modulation conditions, and to make the acquisition timing of the image acquisition module correspond to the modulation time reference.

[0057] The data processing module is used for:

[0058] The labelless optical image sequence is converted into a pixel-level optical feature sequence for each pixel location. The pixel-level optical feature sequence includes at least one of the reflection or scattering intensity feature sequence and the speckle dynamic blood flow feature sequence.

[0059] The target frequency and phase reference are determined based on the modulation time base;

[0060] For each pixel location, based on the pixel-level optical feature sequence, acquisition time, target frequency, and phase reference of the corresponding pixel location, perform a complex exponential projection operation at the target frequency or an orthogonal demodulation operation mathematically equivalent to the complex exponential projection operation to obtain the response coefficient of the corresponding pixel location.

[0061] The response intensity of the corresponding pixel position is determined based on the magnitude of the response coefficient, and a response blood vessel image and / or response blood vessel mask is generated based on the spatial coordinates of each pixel position.

[0062] Specifically, the lighting module includes:

[0063] The first illumination source is used to output incoherent light, low-coherence light, or partially coherent light to form a sequence of reflection or scattering intensity characteristics;

[0064] A second illumination source is used to output coherent light to form a speckle dynamic blood flow characteristic sequence;

[0065] A lighting switching component is used to selectively enable the first and second lighting sources to operate independently, synchronously, or in a time-sharing manner.

[0066] Specifically, the lighting module also includes a light homogenization component, a collimation component, and a light spot adjustment component;

[0067] The light homogenization component is used to spatially homogenize illumination light;

[0068] The collimation component is used to output a collimated illumination beam;

[0069] The spot adjustment component is used to adjust the size of the illumination spot by changing the illumination distance, the position of the optical element, or a combination of optical elements;

[0070] The lighting module can change the output lighting light to either a speckled or non-distinguishable speckled state by replacing or switching the light homogenization processing component.

[0071] Specifically, the lighting module includes a 660nm lighting source and an 850nm lighting source, and the lighting power density can be adjusted within the range of 1mW / cm² to 200mW / cm².

[0072] The lighting module also includes a water-cooling heat dissipation component for cooling the lighting source to maintain the stability of the light power of the lighting source under continuous output conditions.

[0073] Specifically, the image acquisition module includes:

[0074] sCMOS camera with a spectral response range covering 400nm to 1000nm;

[0075] Interchangeable imaging lens, wherein the magnification of the imaging lens includes at least one of 0.1x, 0.2x, 0.5x and 1x;

[0076] A filter switching assembly includes a filter turntable driven by a stepper motor and a plurality of long-pass filters and / or band-pass filters disposed on the filter turntable.

[0077] Specifically, it also includes a modulation device for generating predetermined modulation conditions, the modulation device including at least one of an electrical stimulation device, a mechanical stimulation device, a tactile stimulation device, a visual stimulation device, an auditory stimulation device, a sound stimulation device, or a thermal stimulation device.

[0078] The synchronization control module enables the modulation device, image acquisition module, and data processing module to operate synchronously through hardware trigger signals, a common clock, a timestamp, or a combination thereof.

[0079] Specifically, it also includes a display module and a storage module;

[0080] The display module is used to display the cumulative response vascular images up to the current cycle at the end of each predetermined modulation cycle;

[0081] The storage module is used to store at least one of the following: response vascular image, response vascular mask, original label-free optical image sequence, pixel-level optical feature sequence, image acquisition parameters, illumination parameters, predetermined modulation parameters, and synchronization time information.

[0082] The present invention also provides a computer-readable storage medium having a computer program or computer instructions stored thereon, wherein when the computer program or computer instructions are executed by a processor, the aforementioned responsive label-free vascular imaging method is implemented.

[0083] Compared with the prior art, the beneficial effects of the present invention are:

[0084] This invention acquires image sequences of reflection or scattering intensity and / or laser speckle image sequences of the target area without external optical contrast agents, maps the intrinsic optical hemodynamic feature sequences of each pixel position to the time reference of the predetermined modulation conditions, and extracts the response component at the target frequency. This enables the display of responsive vessels related to the predetermined modulation conditions against a background of ordinary vascular structures or relative blood flow, and suppresses the influence of static vessels, non-responsive vessels, low-frequency drift, and physiological fluctuations deviating from the target frequency on the imaging results.

[0085] Meanwhile, this application can perform imaging using intensity channels formed by changes in hemoglobin absorption, blood volume, blood vessel diameter, and tissue scattering, as well as imaging using blood flow channels formed by dynamic changes in speckle patterns caused by red blood cell movement. The two channels can be used individually or processed synchronously, in time-division, and fused, thus exhibiting good adaptability to different imaging modes. By changing imaging lenses of different magnifications, adjusting working distance, illumination spot, and illumination band, it is also possible to switch between different fields of view, spatial resolution, and detection depth to adapt to non-contact responsive vascular observation of brain, spinal cord, retina, limbs, and other living target areas. Attached Figure Description

[0086] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0087] Figure 1 This is a flowchart of the label-free imaging method for responding blood vessels according to the present invention;

[0088] Figure 2 This is a flowchart illustrating the generation process of the reflection or scattering intensity feature sequence of the present invention.

[0089] Figure 3 This is a flowchart of the generation process for the speckle dynamic blood flow feature sequence of the present invention;

[0090] Figure 4 This is a flowchart of the target frequency response extraction and blood vessel image generation process of the present invention;

[0091] Figure 5 This is a schematic diagram of the two types of optical feature fusion processing of the present invention;

[0092] Figure 6 This is a timing diagram of the modulation conditions, image acquisition, and cumulative update of the present invention. Detailed Implementation

[0093] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0094] Please see Figures 1-6 As shown, the present invention provides a technical solution: a method, system, and storage medium for responding to label-free vascular imaging, comprising the following steps:

[0095] S1. Acquire a label-free optical image sequence formed in the target area of ​​the target living body during the operation of a predetermined modulation condition, an acquisition time sequence corresponding to each frame of the label-free optical image sequence, and a modulation time reference characterizing the predetermined modulation condition. The label-free optical image sequence is formed based on the endogenous optical hemodynamic signal of the target area without the use of an exogenous optical contrast agent.

[0096] S2. For each pixel position in the labelless optical image sequence, perform feature calculation on the image data of the corresponding pixel position at different acquisition times to obtain a pixel-level optical feature sequence F(x,y,n). The pixel-level optical feature sequence includes at least one of the reflection or scattering intensity feature sequence and speckle dynamic blood flow feature sequence.

[0097] S3. Determine the target frequency corresponding to the predetermined modulation conditions based on the modulation time reference; when it is necessary to obtain the response phase, further determine the phase reference;

[0098] S4. For each pixel position, based on the pixel-level optical feature sequence, acquisition time sequence, target frequency and phase reference of the corresponding pixel position, perform complex exponential projection operation at the target frequency or orthogonal demodulation operation mathematically equivalent to complex exponential projection operation to obtain the response coefficient of the corresponding pixel position at the target frequency.

[0099] S5. Determine the response intensity of the corresponding pixel position based on the amplitude of the response coefficient, and map the response intensity according to the spatial coordinates of each pixel position to generate a response blood vessel image and / or response blood vessel mask for highlighting the blood vessels that respond at the target frequency.

[0100] This embodiment utilizes the periodic changes in blood volume, blood flow velocity, and red blood cell motility within the target living body under predetermined modulation conditions to create varying components in the reflected light intensity, scattered light intensity, or speckle dynamics of the vascular region that correspond to the modulation rhythm. By performing temporal analysis on the optical characteristics of each pixel location at continuous acquisition times, response components that are consistent with the modulation frequency and maintain a stable relationship with the modulation phase are extracted. These response components are then remapped to the image space according to their pixel locations, thereby forming an image or mask that highlights the responding blood vessels.

[0101] During the image acquisition phase, the target living organism is fixed in the supporting position of the imaging device, ensuring that the target area to be observed is within the effective field of view of the imaging lens. The imaging device includes an illumination source, an imaging lens, an image sensor, an image acquisition unit, and a modulation and control unit. The illumination source projects coherent light with a wavelength of 785 nanometers onto the target area, and the image sensor continuously acquires optical images of the target area at a acquisition rate of 50 frames per second, with an exposure time of 20 milliseconds for each frame. Throughout the entire acquisition process, the illumination power, illumination angle, lens focal length, aperture, exposure time, and acquisition speed remain constant to avoid temporal fluctuations in the image sequence that are unrelated to vascular response caused by changes in imaging parameters.

[0102] Predetermined modulation conditions are applied to the target living body by the modulation control unit. The modulation control unit repeatedly outputs active and inactive states according to a set cycle. Simultaneously changing the modulation state, the modulation control unit outputs a synchronization trigger signal to the image acquisition unit. The image acquisition unit records the actual time when the modulation state transitions from inactive to active, and also records the actual time when each frame of the image completes or begins exposure. This yields a sequence of acquisition times corresponding to each frame of the image, as well as a modulation time reference characterizing the changes in the modulation state.

[0103] The image acquisition time is determined by a hardware timestamp generated by the image sensor or image acquisition unit, rather than by converting the frame number to a preset frame rate. When frame interval changes, transmission delays, or individual frames are lost during image acquisition, subsequent processing still determines the position of each frame in the modulation period based on the actual acquisition time of each frame, thereby preventing phase errors caused by inconsistencies between the ideal and actual acquisition times.

[0104] Before acquiring an image sequence of the target area, the imaging lens is blocked, and 32 frames of dark-field images are acquired under the same exposure time and gain conditions as the actual acquisition. The arithmetic mean of the grayscale values ​​at the same pixel location in the 32 dark-field images is calculated to obtain the dark-field reference value for that pixel location. The dark-field reference value is used to characterize the dark current and fixed readout bias generated by the image sensor under no-incident light conditions.

[0105] After completing the dark-field acquisition, a uniform diffuse reflection standard plate is placed on the imaging plane of the target area, and 32 frames of flat-field images are acquired under the same illumination and exposure conditions. First, the dark-field reference value for the corresponding pixel position is subtracted from the grayscale value of each pixel in each flat-field image. Then, the arithmetic mean of the corrected grayscale values ​​for the same pixel position in the 32 flat-field images is calculated to obtain the flat-field reference value for that pixel position. Subsequently, the average value of all pixel flat-field reference values ​​is calculated, and this average value is divided by the flat-field reference value for each pixel position to obtain the brightness compensation ratio corresponding to each pixel position.

[0106] For each frame of the original image acquired, the dark field reference value of the corresponding pixel position is first subtracted from the original grayscale value of each pixel position, and then multiplied by the brightness compensation ratio corresponding to that pixel position to obtain the corrected image data. After dark field correction and flat field correction, the fixed dark current deviation in the image sequence and the spatial brightness deviation caused by uneven illumination, lens vignetting, and sensor response differences are suppressed.

[0107] This embodiment does not inject, ingest, coat, or infuse fluorescent dyes, chromogenic agents, nanoparticles, or other exogenous optical contrast agents into the target living organism during the entire image formation process. The temporal variations in the image originate from the target region's own light absorption, light reflection, light scattering, and motion scattering characteristics; therefore, the resulting image sequence is a label-free optical image sequence.

[0108] For each pixel location in a label-free optical image sequence, the corrected grayscale values ​​of that pixel location at all acquisition times are extracted according to the image acquisition order, thus forming a temporal grayscale sequence for that pixel location. When using reflection or scattering intensity features, the arithmetic mean of the corrected grayscale values ​​of that pixel location at all acquisition times is first calculated to obtain the average optical intensity of that pixel location. Then, the corrected grayscale value at each acquisition time is subtracted from the average optical intensity, and the difference is divided by the average optical intensity to obtain the normalized intensity change value at that acquisition time. All normalized intensity change values ​​are arranged in the order of acquisition times to obtain the reflection or scattering intensity feature sequence for that pixel location.

[0109] The above processing uses the average optical intensity of each pixel location as a normalization benchmark, which can reduce the influence of illumination intensity, tissue surface curvature, tissue basic reflectivity and imaging distance on the amplitude of different pixel features, so that subsequent analysis mainly reflects the relative changes of the same pixel location over time.

[0110] When using speckle dynamic blood flow features, a local calculation region is selected, centered on the pixel to be calculated, consisting of 5 pixels horizontally and 5 pixels vertically. For each frame of the image, the corrected gray values ​​of the 25 pixels within the local calculation region are first summed, then divided by 25 to obtain the average gray value of the local calculation region. Subsequently, the difference between the corrected gray values ​​of the 25 pixels and the average gray value is calculated, each difference is squared, all squared values ​​are summed, divided by 24, and the square root of the result is taken to obtain the standard deviation of the gray value of the local calculation region.

[0111] Dividing the standard deviation of the grayscale of the local computational region by the average grayscale value of the local computational region yields the speckle contrast at the pixel location to be calculated in that frame image. The speckle contrast decreases as the velocity of the local scattering particles increases. To ensure that the calculation result maintains the same direction of change as the dynamic changes in blood flow, the reciprocal of the square of the speckle contrast is taken to obtain the speckle dynamic blood flow characteristic value at that acquisition time.

[0112] The arithmetic mean of the speckle dynamic blood flow feature values ​​at the same pixel location across all acquisition times is calculated to obtain the average speckle dynamic blood flow feature value for that pixel location. Then, this average value is subtracted from the speckle dynamic blood flow feature value at each acquisition time, and the difference is divided by the average value to obtain the normalized speckle dynamic blood flow feature value for that acquisition time. All normalized speckle dynamic blood flow feature values ​​are arranged in order of acquisition time to obtain the speckle dynamic blood flow feature sequence for that pixel location.

[0113] When using reflection or scattering intensity features for analysis, subsequent calculations use the reflection or scattering intensity feature sequence; when using speckle dynamic blood flow features for analysis, subsequent calculations use the speckle dynamic blood flow feature sequence. Response extraction is performed separately for both types of features, and they are not directly added before response extraction to avoid affecting the response results due to differences in the numerical ranges and meanings of the variations of different features.

[0114] The target frequency is determined based on the modulation time reference. First, all moments when the modulation state transitions from an inactive to an active state are read sequentially. The time interval between two adjacent state transition moments is taken as a measurement of the modulation period. To reduce the impact of single-trigger jitter on the period calculation, the average modulation period is obtained by dividing the total time between the first and last effective state transition moments by the number of complete cycles contained in that interval. Then, 1 second is divided by the number of seconds contained in the average modulation period to obtain the target frequency.

[0115] The phase reference is the moment when the image transitions from an inactive state to an active state during the first complete modulation cycle. For each frame, the time difference between the actual acquisition time and the phase reference time is first calculated. Then, the number of complete modulation cycles contained in this time difference and the remaining time (less than one complete cycle) are calculated. The proportion of the remaining time to one modulation cycle is used to determine the phase position of the frame within the current modulation cycle. Thus, each frame is assigned a phase corresponding to the predetermined modulation conditions.

[0116] For the same pixel location, the optical feature value of that pixel location, the corresponding actual acquisition time, and the corresponding modulation phase are used together for quadrature demodulation. During demodulation, four time-varying components are established: a fixed component that does not change with time, a linear trend component that increases or decreases uniformly with time, a cosine component that changes in the same direction as the modulation reference phase, and a sine component that differs from the modulation reference phase by a quarter of a period.

[0117] First, calculate the average time across all image acquisition times. Then, calculate the time difference between each acquisition time and the average time. Using the time difference between each acquisition time and the average time as the time reference for linear trend calculation can reduce the numerical correlation between fixed components and linear trend components.

[0118] For each pixel location, the coefficients of the fixed component, the linear trend component, the cosine component, and the sine component are calculated separately. Specifically, based on the optical feature value corresponding to each acquisition time, the difference between this optical feature value and the superposition result of the four components is calculated. Then, the differences from all acquisition times are squared and summed. By adjusting the coefficients of the four components, the sum of the squares of all differences is minimized. The cosine and sine component coefficients corresponding to the minimum value are the two orthogonal response components of that pixel location at the target frequency.

[0119] A fixed component is used to absorb the average bias in the pixel feature sequence, while a linear trend component is used to absorb the unidirectional trend formed during acquisition by slow tissue movement, slow changes in illumination, temperature drift, or slow changes in physiological state. The cosine and sine components together characterize the periodic response consistent with the target frequency. By simultaneously calculating these four components, the incorrect inclusion of DC bias and slow trends in the vascular response can be avoided.

[0120] When calculating the orthogonal response components, an orthogonal decomposition method is used instead of directly inverting the intermediate calculation matrix. In a preferred embodiment, the acquisition duration covers at least three complete modulation cycles, with each modulation cycle containing at least 20 valid sampling moments. This ensures a sufficient number of phase sampling points within the modulation cycle and allows the fixed component, linear trend component, cosine component, and sine component to be calculated separately.

[0121] The cosine and sinusoidal response components together constitute the response coefficient at the target frequency. The cosine response component represents the portion of the response that is in phase with the modulation reference phase, while the sinusoidal response component represents the portion of the response that is offset from the modulation reference phase by a quarter of a period. The response phase is determined by comparing the magnitudes and signs of the sinusoidal and cosine response components.

[0122] When images are acquired at equal time intervals, with the acquisition duration covering an integer number of modulation cycles, and the optical feature sequence has been freed from average bias and linear trends, the same response result can be obtained using complex exponential projection. Specifically, the optical feature value at each acquisition moment is decomposed into in-phase and quadrature components corresponding to the modulation phase at that moment. The in-phase and quadrature components at all acquisition moments are then accumulated. After normalizing the accumulated result by the number of acquisition frames, two quadrature response components at the target frequency are obtained.

[0123] The vascular response, consistent with the target frequency and with a stable phase relationship, maintains the same direction during in-phase accumulation at each acquisition time, and the accumulation result increases with the number of acquisition frames. Random noise, non-periodic motion, and components with frequencies inconsistent with the target frequency have different projection directions at different acquisition times and cancel each other out during accumulation. Therefore, the hemodynamic response corresponding to predetermined modulation conditions can be separated from the optical feature sequence.

[0124] After obtaining the cosine and sine response components, each is squared, the two squared values ​​are added together, and the square root of the sum is taken as the response intensity at that pixel location. The response intensity is unaffected by changes in the phase of the response and is used to characterize the periodic variation amplitude of the pixel location at the target frequency.

[0125] Based on the horizontal and vertical positions of each pixel in the original image, the corresponding response intensities are written to the same positions, forming a response intensity matrix consistent with the width and height of the original image. Vascular regions, influenced by predetermined modulation conditions, exhibit stronger periodic hemodynamic changes, resulting in higher response intensities than surrounding tissue regions lacking a corresponding frequency response.

[0126] When generating a response blood vessel image, the response intensity of all pixels in the target area is read, and the minimum and maximum response intensities are determined. For any pixel, the difference between the response intensity of the pixel and the minimum response intensity is calculated first, and then the difference between the maximum and minimum response intensities is calculated. The first difference is divided by the second difference to obtain the proportional position of the pixel's response intensity within the current response range.

[0127] Multiply the obtained ratio by 65535 and round the result to the nearest integer to obtain the grayscale value of the pixel in the 16-bit response blood vessel image. Pixels with higher response intensity have higher display grayscale, and pixels with lower response intensity have lower display grayscale. When the maximum response intensity and the minimum response intensity in the target area are equal, it means that there is no difference in the response intensity of all pixels, and the display grayscale of all pixels is set to 0.

[0128] When generating the responsive blood vessel mask, the segmentation threshold is automatically determined based on the distribution of the current response intensity matrix. First, the numerical range between the minimum and maximum response intensities is divided into 256 equal intervals. The number of pixels contained in each interval is counted, and then the pixel ratio of each interval is obtained by dividing the number of pixels in each interval by the total number of pixels in the target region.

[0129] The boundary positions of each adjacent interval are sequentially used as candidate segmentation positions. For each candidate segmentation position, all intervals below that position are classified as low-response categories, and all intervals above that position are classified as high-response categories. The pixel proportions of each interval in the low-response and high-response categories are summed to obtain the pixel proportions contained in each category.

[0130] In the low-response category, the index of each interval is multiplied by the pixel ratio of that interval. All products are summed, and then divided by the pixel ratio of the low-response category to obtain the average interval position for the low-response category. The average interval position for the high-response category is calculated using the same method.

[0131] Subsequently, the product of the proportion of pixels in the low-response category, the proportion of pixels in the high-response category, and the square of the difference between the average interval positions of the two categories is calculated, and the result is used as the inter-class difference value for the candidate segmentation position. The inter-class difference value is calculated for each candidate segmentation position, and the candidate segmentation position with the largest inter-class difference value is selected as the final segmentation position. When multiple candidate segmentation positions obtain the same maximum inter-class difference value, the segmentation position with the lower corresponding response intensity is selected to avoid missing vascular pixels with relatively low response intensity but a stable correspondence with the modulation frequency.

[0132] The final threshold is the median response intensity of the interval corresponding to the selected segmentation position. For pixels with a response intensity greater than or equal to the threshold, the value at the corresponding position in the response vessel mask is set to 1; for pixels with a response intensity less than the threshold, the value at the corresponding position in the response vessel mask is set to 0.

[0133] The aforementioned thresholds are calculated based on the pixel distribution of the current response intensity matrix, without relying on subjective judgment by the operator. For different live targets, different target regions, different blood vessel densities, and different baseline optical intensities, the thresholds are determined using the same interval statistics, category classification, and inter-class difference calculation process.

[0134] Through the above processing, the vascular response within the target area that maintains the same frequency and has a stable phase relationship with the predetermined modulation conditions is preserved, while random noise, slow drift, static tissue reflection, and physiological changes that are out of sync with the modulation conditions are suppressed. Ultimately, responsive vascular images and responsive vascular masks are obtained without the use of exogenous optical contrast agents.

[0135] Specifically, intrinsic optical hemodynamic signals include at least one of the following:

[0136] Changes in absorption intensity caused by variations in the concentrations of oxyhemoglobin and deoxyhemoglobin;

[0137] Changes in reflection or scattering intensity caused by changes in blood volume, blood vessel diameter, or tissue scattering state;

[0138] Dynamic changes in speckle patterns caused by changes in the motility of red blood cells or changes in blood flow velocity.

[0139] Specifically, when the pixel-level optical feature sequence includes a reflection or scattering intensity feature sequence, the original intensity I(x,y,n) at pixel position (x,y) in the nth frame image is subjected to dark field subtraction and reference illumination normalization to obtain the corrected reflection or scattering intensity:

[0140] Specifically, when the pixel-level optical feature sequence includes a reflection or scattering intensity feature sequence, the original intensity I(x,y,n) at pixel position (x,y) in the nth frame image is subjected to dark field subtraction and reference illumination normalization to obtain the corrected reflection or scattering intensity: Where D(x,y) is the dark field intensity at pixel position (x,y) in the dark field image. The reference intensity is used for the pixel position (x,y) in the reference illumination image, and the corrected reflection or scattering intensity is used as the reflection or scattering intensity feature sequence.

[0141] Specifically, the illumination light used to form the reflection or scattering intensity characteristic sequence is at least one of incoherent light, low-coherence light, or partially coherent light, and the target area is simultaneously or time-divisionally illuminated using one or more different illumination bands.

[0142] Specifically, when the pixel-level optical feature sequence includes a speckle dynamic blood flow feature sequence, coherent illumination is used to illuminate the target area. For the nth frame image, a local window W is selected centered at pixel position (x,y), and the average gray value of the pixels within the local window is calculated. and pixel grayscale standard deviation The speckle contrast is calculated using the following formula: And based on the speckle contrast, a speckle dynamic blood flow characteristic sequence is formed.

[0143] Specifically, relative blood flow parameters are calculated based on speckle contrast: And by performing a time-averaged calculation on the relative blood flow parameters across multiple consecutive frames, including the current frame, we obtain: Where H is the number of frames included in the time-averaged window before or after the current frame, and the size of the local window W is 3×3, 5×5, 7×7 or other preset size.

[0144] Specifically, the first response vascular image and the second response vascular image are calculated based on the reflection or scattering intensity feature sequence and the speckle dynamic blood flow feature sequence, respectively. The first response vascular image and the second response vascular image are then spatially registered and fused to obtain the fused response vascular image.

[0145] Alternatively, the reflection or scattering intensity feature sequence and the speckle dynamic blood flow feature sequence can be fused before performing complex exponential projection operation or orthogonal demodulation operation at the target frequency.

[0146] Specifically, the same synchronous trigger signal is used to control the start of the predetermined modulation conditions and the start of image acquisition, or the timestamps of the predetermined modulation conditions and image acquisition are recorded separately and time-aligned, so that the acquisition time sequence has a definite time correspondence with the modulation time reference.

[0147] Specifically, the predetermined modulation conditions repeat periodically at the target frequency f1, the acquisition frame rate is f0, and the modulation period is:

[0148] ;

[0149] When the number of acquisition frames corresponding to each modulation cycle is an integer, the number of acquisition frames L in each modulation cycle and the total number of acquisition frames M after N modulation cycles are respectively:

[0150] ;

[0151] ;

[0152] The acquisition time of the nth frame image is:

[0153] ;

[0154] Where n = 0, 1, ..., M-1.

[0155] Specifically, at the end of the m-th modulation period, the M data collected up to the current time are... m =mL frames of image are subjected to one cumulative target frequency extraction, where the response coefficient at the corresponding pixel location is:

[0156] ;

[0157] The response intensity at the corresponding pixel location is:

[0158] ;

[0159] Here, φ0 is a phase reference determined according to the modulation time base, and the displayed vascular image is updated at the end of each modulation cycle.

[0160] Specifically, the orthogonal demodulation operation mathematically equivalent to the complex exponential projection operation includes:

[0161] Calculate the in-phase components:

[0162] ;

[0163] Calculate orthogonal components:

[0164] ;

[0165] The response intensity is calculated according to the following formula:

[0166] ;

[0167] Specifically, after completing a preset number of modulation cycles, the response vascular image obtained at the end of the last modulation cycle is taken as the final response vascular image:

[0168] ;

[0169] The response intensity of each pixel location is compared with a preset response threshold. Pixels whose response intensity meets the preset response threshold condition are identified as responsive blood vessel pixels to generate a responsive blood vessel mask.

[0170] In one possible implementation, after the target area is subjected to predetermined modulation conditions, the metabolic activity, vasomotor state, or blood perfusion state of the local tissue changes with the modulation rhythm. Oxyhemoglobin and deoxyhemoglobin have different light absorption characteristics in different illumination bands. When local oxygen supply, oxygen consumption, or venous return changes, the intensity of reflected light entering the image sensor changes with the concentration of the two types of hemoglobin. The image acquisition module continuously records the optical intensity at each pixel location during modulation, and the temporal intensity changes at each pixel location form a hemoglobin-related optical feature sequence.

[0171] When blood volume increases, the amount of blood components involved in absorption and scattering within the target area increases; when the vessel diameter increases, the proportion of the vessel within the corresponding pixel location increases; when the tissue hydration state, cell arrangement, or morphology of the tissue surrounding the vessel changes, the propagation path and scattering direction of light within the tissue change. All of these changes are reflected as variations in reflection or scattering intensity over time. By extracting the intensity change components that maintain the same repetition frequency as the modulation time reference, the vascular response at the corresponding pixel location is obtained.

[0172] When coherent illumination is used, the movement of red blood cells causes a time-varying phase difference in the scattered light returning to the image sensor. The interference of multiple scattered light components forms a speckle pattern. As blood flow velocity increases, the rate of speckle change during exposure increases, resulting in decreased speckle contrast in the image; conversely, as blood flow velocity decreases, the rate of speckle change decreases, resulting in increased speckle contrast. Calculating the speckle contrast in continuous images yields the dynamic blood flow characteristics of the speckle pattern, reflecting the red blood cell movement and changes in blood flow velocity.

[0173] When simultaneously acquiring changes in absorption intensity, reflection or scattering intensity, and speckle dynamics within the same target area, each signal undergoes correction, feature extraction, and target frequency response calculation. Absorption and reflection signals characterize changes in blood composition and blood volume, while speckle dynamics characterize changes in blood flow. All results are correlated using the same spatial coordinates, thus distinguishing changes in blood composition, vascular morphology, and blood flow velocity within the same vascular region.

[0174] The above processing does not rely on exogenous optical contrast agents, thus avoiding interference with the physiological state of the target organism during contrast agent injection, and allows for continuous and repeated acquisition in the same target area. In another embodiment, only one type of endogenous optical hemodynamic signal is used to complete the responsive vascular imaging; when it is necessary to distinguish between changes in blood oxygenation and changes in blood flow, intensity signals of different wavelengths and speckle signals under coherent illumination conditions are acquired separately.

[0175] In one possible implementation, dark-field images are acquired with the illumination source turned off and the imaging lens blocked. The exposure time, camera gain, pixel readout method, and sensor operating temperature used in dark-field acquisition remain consistent with those used in formal image acquisition. After acquiring 32 consecutive frames of dark-field images, the arithmetic mean of the 32 grayscale values ​​at the same pixel location is calculated, and this average is taken as the dark-field intensity at that pixel location. Using multi-frame averaging reduces the impact of sensor random readout noise on the dark-field intensity.

[0176] Reference illumination images are acquired before the target live object enters the imaging position. A diffuse reflection reference plate with uniform surface reflection is placed on the plane of the target area, and 32 frames of images are continuously acquired under the illumination power, illumination angle, exposure time, lens aperture, and camera gain conditions used in the actual acquisition. For each frame of reference illumination image, dark field subtraction is first performed, and then the processing results for the same pixel location are calculated by arithmetic mean to obtain the effective reference illumination intensity at that pixel location.

[0177] During formal acquisition, the original intensity of any pixel location in the current frame is first subtracted from the dark field intensity of that pixel location to obtain the effective imaging intensity after eliminating dark current and fixing readout bias. Then, this effective imaging intensity is divided by the effective reference illumination intensity at the same pixel location, and the resulting ratio is used as the corrected reflection or scattering intensity.

[0178] When the effective reference illumination intensity is less than one percent of the average intensity of all pixels in the reference image, the corresponding pixel is marked as an invalid pixel and is not included in subsequent response calculations. This process is used to eliminate abnormal locations caused by lens occlusion, insufficient illumination coverage, sensor defects, and local saturation of the reference image. Invalid pixels are assigned a value of 0 in the response vessel image and a non-vessel state in the response vessel mask.

[0179] After dark field subtraction, the sensor's dark current and fixed readout bias are no longer superimposed on the optical intensity of the target area; after reference illumination normalization, uneven illumination spot, lens vignetting, and differences in the photosensitivity response of each pixel are compensated. The resulting reflection or scattering intensity feature sequence mainly characterizes the optical changes of the target area itself over time.

[0180] In the implementation method employing a fixed optical path and with illumination uniformity factory-calibrated, the reference illumination image is stored in the data processing module and calibrated using a standard reference board before each acquisition. When the average intensity of the reference image deviates from the stored calibration value by more than 5%, the reference illumination image is reacquired.

[0181] In one possible implementation, for any pixel location, the dark field intensity of the same pixel location is first subtracted from the original intensity of the current frame. Then, the same dark field intensity is subtracted from the reference intensity of that pixel location in the reference illumination image to obtain the effective reference intensity under the reference illumination condition. The corrected intensity for the corresponding pixel location in the current frame is obtained by dividing the effective imaging intensity of the current frame by the effective reference intensity.

[0182] When using multi-band illumination, a dark-field image and a reference illumination image are acquired separately for each illumination band. When the current frame belongs to the 660 nm illumination channel, normalization is performed using the reference intensity corresponding to the 660 nm channel; when the current frame belongs to the 850 nm illumination channel, normalization is performed using the reference intensity corresponding to the 850 nm channel. Reference images from different illumination channels are not interchangeable.

[0183] After all frames have been corrected, the correction intensity of the same pixel position is arranged from earliest to latest according to the hardware timestamp. For positions where frames are lost, manually generated intensity values ​​are not added; instead, the actual time interval between adjacent valid frames is retained, and the subsequent demodulation process calculates the intensity based on the actual acquisition time.

[0184] To suppress pixel misalignment caused by the overall movement of the target living organism, spatial registration is performed on each frame before the intensity sequence is formed. Using the first frame of the acquisition sequence as the reference image, stable tissue boundaries or fixed marker points within the target region are extracted. The translation and rotation of the current frame relative to the reference image are calculated, and then the coordinates of the current frame are transformed according to the calculation results. The registered pixel positions correspond to the same tissue positions in the target region at different acquisition times.

[0185] When regions with no valid data appear at the edges of the registered image, these regions are removed from the effective imaging range. Subsequent response vascular images are generated only within common regions where valid image data exists in all frames, avoiding the involvement of edge padding values ​​in target frequency extraction.

[0186] The aforementioned correction and alignment process ensures that the reflection or scattering intensity characteristic sequence possesses both spatial and temporal consistency. This reduces the impact of fixed optical biases, inter-frame shifts, and acquisition time errors on the response coefficients.

[0187] In another embodiment, the reference illumination image is formed from multiple baseline images of the target region before the predetermined modulation begins. In this approach, dark field subtraction is first performed on each frame of the baseline phase, and then the average value is calculated for the same pixel location, with the resulting average value used as the reference intensity for that pixel location. The formal modulation phase follows the same normalization process to form an intensity feature sequence.

[0188] In one possible implementation, when incoherent illumination is used, the illumination module uses light-emitting diodes as the light source, and the light output is expanded by a beam-diffusing component to cover the entire target area. The interference effect produced by incoherent illumination is lower than the spatial resolution capability of the image sensor, and the acquired image mainly reflects the intensity distribution of absorption, reflection, and scattering in the target area.

[0189] When using low-coherence illumination, the illumination module employs a broadband semiconductor light source, and the output wavelength is limited by a bandpass filter. Low-coherence illumination reduces the impact of stable speckle on intensity measurements while maintaining high optical power. When using partially coherent illumination, the light source output is processed by a rotating or vibrating diffuser, causing multiple speckle states formed during exposure to be superimposed on the image sensor, thereby reducing the contrast of resolvable speckle in the image.

[0190] In the single-band illumination implementation, the target area is continuously illuminated by the same band throughout the acquisition process, and the image acquisition module continuously acquires images according to a set frame rate. This method reduces illumination switching time and is suitable for scenarios where the main focus is on extracting changes in blood volume, blood vessel diameter, or tissue scattering state.

[0191] In the multi-band synchronous illumination implementation, illumination light of different bands simultaneously illuminates the target area. The returned light is split into different optical paths by a beam splitter and then passes through corresponding bandpass filters to enter different image sensors. Each image sensor is triggered by the same trigger signal to start exposure, thereby obtaining multi-band images with consistent acquisition times.

[0192] In the multi-band time-division illumination implementation, the illumination sources are alternately lit in a fixed sequence. Each illumination source completes stable output before the start of the corresponding image exposure and turns off after the exposure ends. The image acquisition module records the current illumination band and the actual acquisition time in the metadata of each frame. Images belonging to the same band independently form an intensity feature sequence according to time order.

[0193] When using time-division illumination, the illumination switching frequency is more than 10 times the target modulation frequency, and each illumination band obtains no less than 20 valid sampling moments within one modulation cycle. Subsequent target frequency extraction uses the actual timestamps of the images for each band, and images of different bands are not considered to have been acquired simultaneously.

[0194] In multi-band processing, the response intensity of each band is calculated separately. When a composite response image is needed, the response intensity of each band is first normalized to the same numerical range, and then the weight of each band is determined based on the noise level in the baseline phase. Bands with lower baseline noise are assigned higher weights, and bands with higher baseline noise are assigned lower weights.

[0195] The aforementioned illumination method reduces the interference of stable speckle on the measurement of reflection or scattering intensity, and preserves the hemoglobin absorption and tissue scattering information reflected by different wavelengths. In another embodiment, different wavelengths are selected in the same light source output by a filter switching assembly, while the image acquisition and wavelength marking processes remain unchanged.

[0196] In one possible implementation, the coherent illumination light is generated by a narrow-linewidth semiconductor laser source, and the illumination light is collimated and expanded to cover the target area. The relative positions of the light source, imaging lens, and target live object remain fixed during acquisition. The exposure time, gain, and aperture of the image sensor remain constant throughout the modulation cycle, so that the speckle contrast variation mainly originates from the motion changes of scattering particles within the target area.

[0197] For any pixel to be calculated in the current frame, a 5x5 pixel local window is selected centered on that pixel. Pixels located at the image edge that cannot form a complete local window are not included in the calculation, or the window data is supplemented by mirror extension. The same edge processing method is used for the entire image sequence.

[0198] Add the grayscale values ​​of the 25 pixels within the local window and divide by 25 to obtain the local grayscale mean. Then calculate the difference between the grayscale value of each pixel and the local grayscale mean, square all the differences, add the squared results, divide by 24, and take the square root of the result to obtain the local grayscale standard deviation.

[0199] Dividing the local grayscale standard deviation by the local grayscale mean yields the speckle contrast of the pixel to be calculated in the current frame. If the local grayscale mean is less than one percent of the camera's full scale, the pixel is marked as a low-signal pixel and its speckle contrast is not calculated. If there are saturated pixels within a local window, the window is marked as a saturated window and is not included in subsequent response extraction.

[0200] The above processing is performed on all valid pixels one by one to form the speckle contrast image of the current frame. Then, the speckle contrast of the same pixel position in each frame is arranged according to the acquisition time order to obtain the speckle dynamic feature sequence of the corresponding pixel position.

[0201] During the phase of increased blood flow velocity, red blood cells exhibit faster speckle changes during exposure, with multiple speckle states averaging in a single exposure, resulting in decreased speckle contrast. Conversely, during the phase of decreased blood flow velocity, speckle changes during exposure decrease, leading to increased speckle contrast. Therefore, the temporal variation in speckle contrast can reflect changes in red blood cell motility under predetermined modulation conditions.

[0202] To reduce the impact of local brightness gradients on standard deviation calculation, another implementation first performs planar background fitting on the local window, then subtracts the fitted background component from the gray values ​​of each pixel, and finally calculates the standard deviation of the remaining gray values. This process does not change the basic process of determining speckle contrast based on local mean and local standard deviation.

[0203] In one possible implementation, for each valid pixel location, the speckle contrast of that pixel in the current frame is first squared, and then 1 is divided by the squared value to obtain the relative blood flow index of the current frame. When the speckle contrast decreases, the relative blood flow index increases; when the speckle contrast increases, the relative blood flow index decreases, thereby ensuring that the direction of change of the feature value is consistent with the direction of increased blood flow.

[0204] To suppress single-frame speckle noise, a time-averaged relative blood flow index is applied. During offline processing, with the current frame as the center, the 5 frames before the current frame, the current frame, and the 5 frames after the current frame are selected, for a total of 11 frames of relative blood flow index. The arithmetic mean of the 11 values ​​at the same pixel position is calculated to obtain the time-averaged relative blood flow index corresponding to the current frame.

[0205] During online processing, back-side images that have not yet been acquired are not used. The data processing module selects the current frame and the 10 frames preceding it, for a total of 11 frames, for averaging relative to blood flow parameters. If there are fewer than 11 frames at the beginning of the sequence, all frames acquired up to the current time are used for averaging, and the actual number of frames involved in the averaging is recorded.

[0206] The duration of the time-averaged window is less than one-quarter of the modulation period. When sampling at a frame rate of 100 Hz and a modulation period of 1 second, an 11-frame time window corresponds to 0.11 seconds, preventing the mixing of most of the time variations between the stimulus and non-stimulation phases. As the modulation period changes, the time-averaged frame count is adjusted accordingly to keep the average window duration below one-quarter of the modulation period.

[0207] The local window size is determined based on the horizontal pixel width of a single target blood vessel in the image. A 3x3 window is used when the vessel width is 3 to 5 pixels; a 5x5 window is used when the vessel width is 5 to 9 pixels; and a 7x7 window is used when the vessel width exceeds 9 pixels. The window size remains constant throughout a single image sequence processing operation.

[0208] When the local window is too small, the statistical fluctuation of the local standard deviation increases; when the local window is too large, blood vessels and surrounding tissues enter the computational region simultaneously, resulting in a decrease in spatial resolution. The above window selection process establishes a definite correspondence between statistical stability and the ability to preserve blood vessel edges.

[0209] The time-averaged relative blood flow indices form a speckle dynamic blood flow feature sequence according to the acquisition time sequence, and then enter the target frequency extraction process. In another embodiment, the time averaging adopts a weighted average that assigns higher weight to the current frame and lower weight to frames that are farther away from the current frame, and the sum of all weights is normalized to 1.

[0210] In one possible implementation, when fusing image results, the reflection or scattering intensity image and the speckle dynamic image are used to perform dark field correction, reference illumination normalization, feature sequence generation, and target frequency response extraction, respectively. A first-response vascular image is obtained from the intensity features, and a second-response vascular image is obtained from the speckle dynamic features.

[0211] When two types of images are acquired by the same image sensor in a time-division manner, the images from different illumination channels are first grouped according to the acquisition time, and then spatial registration is completed using stable tissue boundaries and vascular bifurcation locations within the target area. When two types of images are acquired synchronously by different image sensors, a planar calibration plate with regularly marked points is used to establish the mapping relationship between the coordinates of the two images before formal acquisition, and secondary translation correction is performed using the vascular bifurcation locations in the live images.

[0212] After spatial registration, the first and second response vascular images are normalized. During normalization, the response intensity at the 1st percentile within each target region is used as the lower limit, and the response intensity at the 99th percentile is used as the upper limit. Pixels below the lower limit are assigned a value of 0, pixels above the upper limit are assigned a value of 1, and pixels in between are assigned a value according to their proportion between the upper and lower limits.

[0213] The fusion weights are determined based on the degree of fluctuation in the non-response background region. The standard deviation of the background region response intensity is calculated for both the first and second response vascular images. The reciprocal of this standard deviation is used as the initial weight, and the two initial weights are normalized until their sum is 1. The two normalized response images are then multiplied by their respective weights and summed to obtain the fused response vascular image.

[0214] When one type of image is marked as saturated, low signal, or invalid data at a certain pixel location, that pixel location only uses the valid response value of the other type of image, and invalid data is not used in the fusion. When both types of images are invalid at the same pixel location, that location is assigned a value of 0 in the fused image.

[0215] When using feature-level fusion, the intensity feature sequence and the speckle dynamic blood flow feature sequence are first normalized by removing the mean and standard deviation, respectively. For each pixel location, the two normalized features at the same acquisition time are added together according to predetermined weights to form a fused feature sequence. The weights are still determined by the background noise during the baseline phase. The fused feature sequence is then used to extract the target frequency according to the acquisition time, target frequency, and phase reference.

[0216] Image result fusion preserves the independent responses of the two types of features, facilitating the separate observation of changes in blood volume and blood flow. Feature-level fusion reduces the number of subsequent response calculations and is suitable for situations where the two types of images have consistent spatial coordinates and are close to the acquisition time. Either of the two processing paths is executed in the same imaging task, and the fusion method used is recorded in the result data.

[0217] In one possible implementation, when using the same synchronization trigger signal, the synchronization control module outputs a start pulse generated by the same hardware clock to both the modulation device and the image acquisition module. Upon receiving the start pulse, the image acquisition module begins the first frame exposure, and upon receiving the same start pulse, the modulation device enters the first modulation cycle. The synchronization control module simultaneously records the output time of the start pulse and uses this time as the common zero point for both the image acquisition and modulation processes.

[0218] To eliminate the internal response delay of the modulation device, the start pulse output by the synchronization control module and the actual output signal of the modulation device were simultaneously recorded using the oscilloscope recording channel before formal acquisition. The time difference between the two was written as a fixed delay into the modulation parameters. When determining the phase reference later, this fixed delay was added to the modulation start time, thus ensuring that the phase reference corresponds to the moment when the modulation conditions actually act on the target living organism.

[0219] When using independent timestamps, the modulation device records its own clock time at each state switch, and the image acquisition module records its own clock time for each frame of image. The synchronization control module sends time synchronization pulses to both devices before, during, and after acquisition. Both devices record the local time of each time synchronization pulse.

[0220] The data processing module establishes an initial time offset based on the first time synchronization pulse, determines the cumulative drift of the two devices' clocks during the acquisition period based on the last time synchronization pulse, and then converts all timestamps from the image acquisition module to the time reference used by the modulation device according to the linear time relationship. When the acquisition time is long, a time synchronization pulse is sent every 10 modulation cycles to correct nonlinear clock drift.

[0221] After time conversion, the time interval relative to the most recent modulation state switching moment is determined for each frame, and the phase position of that frame in the modulation period is determined accordingly. If the modulation time record is missing or the interval between two consecutive modulation switching deviates from the average period by more than 5%, the corresponding modulation period is marked as an abnormal period and is not included in the final response calculation.

[0222] The above processing establishes a definite time correspondence between image feature values ​​and modulation states, avoiding response phase errors caused by trigger delay, clock skew, and clock drift. Hardware triggering is suitable for configurations with trigger interfaces between devices, while timestamp alignment is suitable for configurations where each device operates independently and can record high-precision time information.

[0223] In one possible implementation, the modulation period is obtained by dividing 1 second by the target frequency. The number of frames acquired per modulation period is obtained by dividing the frame rate by the target frequency. When the resulting number of frames is an integer, this integer is used as the number of image frames contained in each modulation period. After a preset number of modulation periods, the total number of acquired frames is obtained by multiplying the number of image frames in each period by the number of modulation periods.

[0224] For any image frame numbered starting from 0, its acquisition time is obtained by dividing the frame number by the acquisition frame rate. The acquisition time of the first image frame is 0, and the acquisition time of the subsequent image frame is increased by one frame interval relative to the previous frame. The frame interval is obtained by dividing 1 second by the acquisition frame rate.

[0225] In one specific embodiment, the target frequency is set to 1 Hz, the modulation period is 1 second, and the frame rate is set to 100 Hz. Each modulation period contains 100 frames. When 10 modulation periods are executed consecutively, the total number of frames acquired is 1000. The first frame is acquired at 0 seconds, the second frame at 0.01 seconds, the 100th frame at 0.99 seconds, the 101st frame at 1 second, and the last frame at 9.99 seconds.

[0226] When the frame rate divided by the target frequency does not yield an integer, the number of frames per cycle is not forcibly determined through rounding; instead, demodulation at the target frequency is performed based on the hardware timestamp of each frame. Another approach is to adjust either the frame rate or the target frequency so that their ratio is an integer. Even after adjustment, the frame rate is guaranteed to be no less than 20 times the target frequency.

[0227] When an integer number of frames is used per cycle, the same modulation phase corresponds to the same frame sequence position in different cycles, which facilitates the cumulative response extraction at the end of each modulation cycle and reduces phase position drift caused by non-integer sampling.

[0228] In one possible implementation, after the first modulation cycle ends, the data processing module reads all image feature values ​​within the first cycle; after the second modulation cycle ends, it reads all image feature values ​​within the previous two cycles; thereafter, after each modulation cycle is completed, the newly obtained image feature values ​​are added to the previous data to form the cumulative data up to the current cycle.

[0229] For any pixel location, the optical feature values ​​at each acquisition time in the accumulated data are read sequentially. Based on the time difference between the acquisition time and the phase reference, the phase position at the target frequency at that acquisition time is determined. The current optical feature value is multiplied by the cosine and sine reference values ​​corresponding to the phase position, and then summed over all accumulated frames.

[0230] Divide the in-phase accumulation result and the quadrature accumulation result by the cumulative frame number, then multiply by 2 to obtain the in-phase response component and quadrature response component of the pixel position up to the current period. Square the two response components respectively, sum the squared results, and take the square root to obtain the response intensity up to the current period.

[0231] The cumulative frame count increases with the number of completed modulation cycles. At the end of the first cycle, the cumulative frame count equals the number of frames contained in one cycle; at the end of the fifth cycle, the cumulative frame count equals five times the number of frames per cycle; at the end of the last cycle, the cumulative frame count equals the total number of frames acquired.

[0232] After each response intensity calculation, a cumulative response vascular image is generated according to pixel spatial coordinates and sent to the display module. The display module marks the number of modulation cycles that have been completed in the image title area, allowing the operator to observe the changes in the response vascular image as the cumulative cycle increases.

[0233] To avoid repeatedly reading all historical images, the data processing module saves the in-phase cumulative value, orthogonal cumulative value, and effective frame number for each pixel position up to the previous cycle. At the end of a new cycle, only the in-phase projection value and orthogonal projection value for each frame within the new cycle are calculated, and the previously accumulated results are added. The final response coefficient is consistent with the result of recalculating all historical frames.

[0234] When a modulation period is marked as an abnormal period by the synchronization control module, the image of that period is not added to the cumulative result, and the number of effective periods does not increase. The above accumulation method can reduce the impact of random noise on the response image as the effective modulation period increases, and supports real-time observation of the gradual emergence of response vessels during the acquisition process.

[0235] In one possible implementation, for any pixel location, the modulation phase corresponding to each acquisition time is determined based on the acquisition time of each frame image, the target frequency, and the phase reference. For each modulation phase, a cosine reference value and a sine reference value are obtained respectively.

[0236] When calculating the in-phase component, the pixel-level optical feature value at each acquisition moment is multiplied by the corresponding cosine reference value, and then all products are summed. The resulting accumulated value is divided by the number of valid frames involved in the calculation and multiplied by 2 to obtain the in-phase response component of that pixel position at the target frequency.

[0237] When calculating the orthogonal components, the pixel-level optical feature value at each acquisition moment is multiplied by the corresponding sine reference value, and then all products are summed. The resulting accumulated value is divided by the number of valid frames and multiplied by 2 to obtain the orthogonal response component of that pixel position at the target frequency.

[0238] When the acquisition time does not cover an integer number of modulation periods or there are unequal intervals between image acquisition times, the time average and linear trend are first removed from the optical feature sequence, and then the cosine reference value and sine reference value are calculated based on the actual acquisition time. At this time, the in-phase response component and the quadrature response component are obtained by minimizing the sum of squared residuals of all acquisition times.

[0239] The response intensity is determined by the following process: first, square the in-phase response component, then square the quadrature response component, add the two squared results, and finally take the square root of the sum. This value characterizes the amplitude of the target frequency response and is unaffected by the position of the response phase within the modulation period.

[0240] The response phase is determined by the sign relationship between the in-phase and quadrature response components and the quadrant in which their ratio lies. The response phase is used to distinguish between blood flow changes that occur earlier and those that occur later after the stimulus begins, but it is not a necessary condition for generating a response vascular image.

[0241] Quadrature demodulation retains only the varying components that maintain a stable phase relationship with the target frequency reference. Static intensity cancels each other out after entering the positive and negative ranges of the cosine and sine references, while random variations weaken during accumulation due to their non-fixed phase.

[0242] In one possible implementation, the preset number of modulation cycles is written into the modulation parameters before the acquisition begins. This embodiment uses 10 modulation cycles. After the 10th cycle, the data processing module uses all valid images from the 10 cycles to calculate the final response intensity at each pixel location and forms the final response vascular image according to spatial coordinates. Intermediate response images during the acquisition process are only used for display and do not replace the final response image.

[0243] The preset response threshold was determined through baseline calibration before the formal acquisition. Without applying periodic modulation, the illumination power, exposure time, acquisition frame rate, target area location, and total acquisition duration were kept consistent with the formal acquisition, and three baseline acquisitions were completed. For each baseline data set, the same response extraction was performed using the target frequency and phase reference used in the formal acquisition, resulting in three baseline response intensity images.

[0244] Saturated pixels, low-signal pixels, and invalid edge pixels are excluded from each baseline response intensity image. The response intensity of the remaining pixels is then arranged from smallest to largest. The response intensity corresponding to the position where the arrangement reaches 99.7% of all valid pixels is taken as the candidate threshold for that baseline acquisition. Three baseline acquisitions generate three candidate thresholds, and the maximum value among them is taken as the preset response threshold used in the actual acquisition.

[0245] The threshold determined using the above method corresponds to the high level of background fluctuations under aperiodic modulation conditions. The threshold is fixed before the formal acquisition begins and is stored along with camera gain, exposure time, illumination power, target frequency, acquisition frame rate, and feature type. If any of these parameters changes, baseline calibration is re-performed, and the previous threshold is not used.

[0246] After the final response vessel image is formed, the response intensity is read pixel by pixel. When the response intensity is greater than or equal to a preset response threshold, the corresponding pixel is identified as a response vessel pixel and assigned a value of 1 in the response vessel mask; when the response intensity is lower than the preset response threshold, a value of 0 is assigned in the response vessel mask.

[0247] To remove isolated regions formed by a single anomalous pixel, a connected component screening is performed on the binary mask. Regions with fewer than 9 connected pixels are deleted. This area threshold is determined by the number of pixels the smallest target blood vessel covers at the current imaging magnification and is recalibrated when the imaging magnification changes.

[0248] After threshold comparison and connected region filtering, the response vascular mask retains only the response regions that exceed the baseline noise level at the target frequency and have a defined spatial range, reducing the result differences caused by manually selected thresholds.

[0249] In one possible implementation, the predetermined modulation conditions include a stimulation phase and a non-stimulation phase, and the target frequency is determined based on the repetition period formed by the stimulation phase and the non-stimulation phase.

[0250] Each modulation cycle consists of consecutive stimulation and non-stimulation phases. At the start of the stimulation phase, the synchronization control module outputs a start signal to the modulation device; at the end of the stimulation phase, the synchronization control module stops the modulation output and enters the non-stimulation phase. The stimulation and non-stimulation phases are repeated for a fixed duration.

[0251] In one embodiment, the stimulation phase lasts 0.5 seconds, the non-stimulation phase lasts 0.5 seconds, the complete modulation cycle is 1 second, and the target frequency is 1 Hz. Ten modulation cycles are executed continuously, for a total acquisition time of 10 seconds. The image acquisition module acquires images continuously throughout the entire process, without stopping between the stimulation and non-stimulation phases.

[0252] The static vascular component remains approximately constant throughout the entire acquisition period. Its positive and negative components cancel each other out when multiplied by the cosine and sine references within the complete modulation cycle. Unmodulated vessels lack the variation to maintain a stable phase relationship with the target frequency, and their cumulative response intensity is lower than that of the responding vessels.

[0253] The heart rate component, respiratory component, and low-frequency drift component each have their own frequency. The target frequency extraction only accumulates the changing components that are consistent with the modulation frequency and phase reference. The periodic components whose frequency deviates from the target frequency gradually become phase-shifted over multiple modulation cycles and are suppressed during the accumulation process.

[0254] To reduce interference caused by the target frequency coinciding with heart or respiratory rates, a 10-second baseline image is recorded before acquisition begins, and the spectrum of the average intensity or average speckle flow index in the target region is calculated. When the interval between the modulation frequency and the main physiological frequency in the baseline signal is less than twice the frequency resolution, the duration of the stimulation and non-stimulation phases is reset so that the target frequency avoids the recorded main physiological frequency.

[0255] In another implementation, the durations of the stimulation and non-stimulation phases differ, but their sum remains constant. The target frequency is still determined by the complete repetition cycle formed by both phases, and the demodulation reference is generated according to the actual stimulation waveform.

[0256] In one possible implementation, the predetermined modulation conditions employ electrical stimulation, mechanical stimulation, tactile stimulation, visual stimulation, auditory stimulation, sound stimulation, thermal stimulation, physiological modulation, task induction, or pharmacological effects, with the target area set as the brain, spinal cord, retina, hand, foot, limb, tumor area, inflammatory area, or living tissue area.

[0257] In this embodiment of visual response imaging of the cerebral cortex, the target living head is fixed to a support platform, and the target area of ​​the cerebral cortex is located within the imaging field of view. The display device shows a periodically flipping black-and-white pattern during the stimulation phase and a uniform grayscale image during the non-stimulation phase. The image switching signal of the display device is simultaneously input to the synchronization control module as a modulation time reference. The image acquisition module records label-free optical images of the cerebral cortex surface and extracts the vascular response consistent with the visual stimulation cycle.

[0258] In this embodiment of limb tactile response imaging, the contact end of the tactile stimulation device contacts the skin of the fingers or feet. During the stimulation phase, a driver applies reciprocating pressure at a fixed frequency and amplitude; during the non-stimulation phase, the driver stops and no additional pressure is applied to the contact end. A synchronization control module records the start and stop times of the driver, and an image acquisition module continuously acquires images of superficial blood vessels around the stimulation site.

[0259] In the spinal cord electrical stimulation response imaging embodiment, the electrical stimulation device outputs electrical stimulation according to a set pulse width, pulse frequency, and stimulation duration. Each stimulation phase contains the same number of electrical pulses, and pulse output ceases during non-stimulation phases. The pulse envelope signal of the electrical stimulation device serves as the modulation time reference, and the target frequency used for response extraction is determined by the repetition period of the stimulation and non-stimulation phases, not by the frequency of a single electrical pulse.

[0260] In the tumor region drug action imaging embodiment, the drug delivery device performs a drug delivery phase and a non-drug delivery observation phase at fixed time intervals, and image acquisition continuously covers the pre-drug baseline, the drug delivery phase, and the recovery phase. For single drug actions that do not have strict periodicity, a reference sequence corresponding to the drug action time process is first constructed based on the drug delivery time, and then the response region is extracted using a correlation projection process equivalent to orthogonal demodulation.

[0261] When using thermal stimulation, the thermal stimulation device uses closed-loop temperature control to ensure the contact surface reaches the set temperature during the stimulation phase and returns to the baseline temperature during the non-stimulation phase. The actual temperature change recorded by the temperature sensor serves as the modulation time reference, avoiding the need to determine the action time solely based on the heating control signal.

[0262] Each modulation device underwent output calibration before formal imaging. Stimulation amplitude, duration, and repetition count were maintained within the limits specified in the established experimental protocol, and the actual output parameters were recorded in the stored data. This process establishes a time-correlation relationship between vascular response and specific stimulus events, making it applicable to different tissue regions and physiological tasks.

[0263] In one possible implementation, the illumination light used to form a sequence of reflection or scattering intensity characteristics includes incoherent, low-coherent, or partially coherent illumination light with a wavelength of 660 nm and incoherent, low-coherent, or partially coherent illumination light with a wavelength of 850 nm; the illumination light used to form a speckle dynamic blood flow characteristic sequence includes coherent illumination light with a wavelength of 850 nm.

[0264] The 660 nm intensity illumination channel uses a light-emitting diode (LED) source with a center wavelength of 660 nm. The returned light passes through a bandpass filter matched to this wavelength before entering the image sensor. The 850 nm intensity illumination channel uses a broadband semiconductor source with a center wavelength of 850 nm. The returned light passes through an 850 nm bandpass filter for acquisition.

[0265] The 850 nm speckle illumination channel uses a narrow-linewidth semiconductor laser source with a center wavelength of 850 nm. Coherent light is expanded to illuminate the target area, and the image sensor acquires speckle images using a bandpass filter in the same wavelength band. The 850 nm intensity illumination and 850 nm speckle illumination operate in a time-division multiplexing manner to avoid altering the speckle contrast by simultaneously introducing broadband and coherent light sources into the image sensor.

[0266] The illumination power density was measured on the plane of the target living body surface. During measurement, the target living body was removed, and a calibrated optical power meter probe was placed at five locations around the center of the target area. The incident power at each location was measured, and then divided by the effective receiving area of ​​the probe to obtain the power density. The average power density at the five locations was controlled within the range of 1 milliwatt per square centimeter to 200 milliwatts per square centimeter.

[0267] In this embodiment, the illumination power density is set to 30 milliwatts per square centimeter, falling within the operating range of 10 to 60 milliwatts per square centimeter. The difference between the maximum and minimum power densities at the five measurement locations does not exceed 10 percent of the average. If this range is exceeded, the illumination distribution is recalibrated by adjusting the homogenizing components and the spot size.

[0268] The image exposure time is set to 10 milliseconds, and the acquisition frame rate is set to 100 Hz. Each exposure is triggered by the synchronization control module, and the image sensor completes the preparation for the next frame exposure during the readout of the previous frame. The actual hardware timestamp is used to record the acquisition time of each frame.

[0269] When the target frequency is increased, the acquisition frame rate is adjusted to at least 20 times the target frequency; when the reflection intensity of the target area is low, the exposure time or illumination power is increased without causing pixel saturation. After adjustment, dark field images, reference illumination images, and baseline threshold data are reacquired.

[0270] The above parameter configuration takes into account intensity signal, speckle dynamic signal and time resolution, so that the same target area can be obtained separately for hemoglobin-related intensity changes and blood flow velocity-related speckle changes.

[0271] In one possible implementation, the responsive label-free vascular imaging system includes an illumination module, an image acquisition module, a synchronization control module, and a data processing module.

[0272] The illumination module is positioned above the target live object, and its illumination direction and distance are maintained by a fixed bracket. The illumination module outputs a first illumination light that does not produce resolvable speckle and a second illumination light that produces speckle. The first illumination light is used to form a reflection or scattering intensity image, and the second illumination light is used to form a dynamic speckle image. The two illumination lights can be output individually, simultaneously, or in a time-division manner according to the control signal from the synchronization control module.

[0273] The image acquisition module includes an imaging lens, a filter assembly, an image sensor, and an acquisition controller. The imaging lens images the reflected light from the target area onto the image sensor. The acquisition controller records a hardware timestamp, illumination channel, exposure time, gain, and frame number for each frame exposure, and transmits the image data along with the corresponding parameters to the data processing module.

[0274] The synchronization control module receives the modulation time reference output by the modulation device, or outputs control signals to the modulation device. The synchronization control module uses the same clock to generate the image exposure trigger signal and the modulation control signal, and records the actual output time of each signal. When using independent clock devices, the synchronization control module establishes the time correspondence between devices through time synchronization pulses.

[0275] After receiving the image sequence, the data processing module first performs dark field subtraction, reference illumination normalization, invalid pixel marking, and inter-frame registration. Then, it generates a reflection or scattering intensity feature sequence, a speckle dynamic blood flow feature sequence, or both for each pixel location.

[0276] The data processing module calculates the modulation period, target frequency, and phase reference based on the modulation time base, and determines the phase position according to the actual acquisition time of each frame. For each pixel position, complex exponential projection or quadrature demodulation is performed to obtain the in-phase response component and the quadrature response component, and then the response intensity is determined based on the square root of the sum of the squares of the two components.

[0277] The response intensity of all pixels is written into a two-dimensional response matrix according to the coordinates of the original image. After grayscale mapping, the response matrix forms a response vessel image, which is then compared with a threshold to form a response vessel mask. The processing result, along with the original image, acquisition parameters, and synchronization information, is output together.

[0278] The lighting module, image acquisition module, and modulation device employ electrically isolated control interfaces to prevent electromagnetic interference from the lighting drive current or modulation device output to the image sensor. Each module is mounted on a rigid support, and the target live subject platform is equipped with a position locking mechanism to maintain spatial stability during continuous acquisition.

[0279] In one possible implementation, the lighting module includes a first lighting source, a second lighting source, and a lighting switching component.

[0280] The first illumination source uses a light-emitting diode (LED) light source or a broadband semiconductor light source to output incoherent light, low-coherence light, or partially coherent light. The output light from the first illumination source illuminates the target area after passing through a homogenizing component and a beam expander component, which is used to form a sequence of reflection or scattering intensity characteristics.

[0281] The second illumination source uses a narrow-linewidth semiconductor laser source to output coherent light. The output light from the second illumination source is collimated, expanded, and shaped to illuminate the target area, causing the scattered light returning from the target area to form resolvable speckle patterns on the image sensor, which is used to form a speckle dynamic blood flow feature sequence.

[0282] The lighting switching assembly includes electronic switches respectively located in the driving circuits of the two light sources. The synchronization control module outputs control signals to the electronic switches according to a preset lighting sequence, enabling the first and second lighting sources to operate independently, synchronously, or in a time-sharing manner.

[0283] In time-sharing operation, the first illumination source is lit during the first frame of exposure, the second illumination source is lit during the second frame of exposure, and so on, alternating in the same order thereafter. Each light source completes stable output before the start of exposure and turns off after the exposure ends. The image acquisition module writes the corresponding illumination channel identifier into each frame of data.

[0284] When operating synchronously, the two illumination lights are output at different wavelengths, and the returned light enters different image sensors or different spectral channels through a beam splitter. The coherent speckle channel does not receive the spectral components of the first illumination source, and the intensity channel does not receive the coherent light components of the second illumination source.

[0285] The lighting switching assembly also includes a mechanical light-blocking component. When the electronic switch is turned off, the mechanical light-blocking component closes the corresponding optical path to prevent the weak output generated by the residual current driving the light source from entering the non-corresponding image channel.

[0286] The above configuration enables intensity imaging and speckle imaging to use their respective matched illumination states, and maintains a definite correspondence between image data and illumination mode through channel marking.

[0287] In one possible implementation, the lighting module further includes a light homogenization component, a collimation component, and a light spot adjustment component.

[0288] The homogenization processing component is positioned between the light source output and the collimation component. The homogenization processing component corresponding to the first illumination source employs an integrating rod and a rotating diffuser. The integrating rod performs multiple reflections and mixing of the incident light, while the rotating diffuser alters the local phase distribution during exposure, enabling the target area to obtain spatially uniform illumination light without stable, resolvable speckle.

[0289] The homogenization processing component corresponding to the second illumination source does not use a rotating diffuser; the coherent light is output through a static beam expander, thereby preserving the spatial coherence required to form speckle. The illumination module switches between the rotating diffuser optical path and the static beam expander optical path via an electric switching frame.

[0290] The collimation assembly includes a beam expander and a collimating lens arranged along the optical axis. The beam expander increases the beam diameter, while the collimating lens converts the diverging beam into an approximately parallel beam. The collimation state is verified by measuring the spot diameter at two different propagation distances, with the difference in spot diameter between the two locations controlled within five percent of the average diameter.

[0291] The beam adjustment assembly includes a lens group that moves along the optical axis and an adjustable aperture. A drive mechanism changes the position of the lens group to adjust the convergence or divergence of the beam; the adjustable aperture limits the outer edge of the beam. After adjustment, the illumination spot covers the target area, with an illumination margin of no less than 10% of the width of the target area remaining outside the edge of the target area.

[0292] When outputting illumination light without resolvable speckle, the rotating diffuser completes multiple scattering state changes during a single frame exposure, averaging the multiple speckle states in the image sensor. When outputting illumination light with speckle, the rotating diffuser moves out of the optical path, and the coherent light is directly collimated and expanded to illuminate the target area.

[0293] In another configuration, the homogenization component uses an optical fiber bundle or an integrating sphere, and the illumination state switching is still achieved by adding or removing dynamic diffusion elements in the coherent optical path. After replacing the homogenization component, the illumination uniformity and speckle contrast of the target plane are remeasured, and the calibration data is saved to the illumination parameter file.

[0294] In one possible implementation, the lighting module includes a 660 nm lighting source and an 850 nm lighting source, with the lighting power density adjustable in the range of 1 mW / cm² to 200 mW / cm², and the light power stability under continuous output conditions is maintained by a water-cooled heat dissipation component.

[0295] The 660 nm and 850 nm lighting sources are respectively mounted on metal heat-conducting bases. Each heat-conducting base has a coolant channel inside, and the coolant inlet and outlet are connected to the circulation pump, the storage container, and the radiator, respectively, forming a closed circulation loop.

[0296] The temperature sensor in the water-cooled heat dissipation assembly is located near the light source package. The controller reads the temperature sensor data and adjusts the circulation pump speed and radiator fan speed to maintain the light source base temperature within a range of ±1 degree Celsius of 25 degrees Celsius. Before formal data acquisition, the illumination source runs continuously for 5 minutes to allow the base temperature and output light power to stabilize before image acquisition begins.

[0297] The lighting power is regulated by a constant current driver. The constant current driver receives the set value output from the synchronization control module and changes the light source drive current. A light power detection position is set on the plane where the target area is located, and the incident power density is measured using a calibrated power meter. The correspondence between the drive current and the power density is stored in the lighting calibration table.

[0298] During continuous output for 30 minutes, the optical power is recorded every minute. The optical power is considered stable when the difference between the maximum and minimum recorded values ​​does not exceed two percent of the average optical power. If it exceeds this range, stop the acquisition process and check the coolant flow rate, light source base temperature, and constant current drive status.

[0299] Water-cooled heat dissipation components reduce center wavelength drift and output power drift caused by changes in light source temperature, ensuring consistent lighting conditions across different modulation cycles. In another embodiment, the 660nm light source uses air cooling, while the 850nm high-power light source uses water cooling; the actual operating temperature and output power of each light source are still recorded separately.

[0300] In one possible implementation, the image acquisition module includes a scientific-grade complementary metal-oxide-semiconductor camera with a spectral response range covering 400 nanometers to 1000 nanometers, an interchangeable imaging lens, and a filter switching assembly.

[0301] The camera is fixed to a vertically adjustable stand, with the sensor imaging plane parallel to the target area. The camera operates in global exposure mode, with each frame triggered by a synchronization control module. The camera outputs raw grayscale images and hardware timestamps, without performing automatic exposure, automatic gain control, or image sharpening internally.

[0302] The imaging lens connects to the camera via a standard interface. Lenses with magnifications of 0.1x, 0.2x, 0.5x, or 1x are installed depending on the size of the target area. After changing the lens, a calibration plate with known spacing markings is placed on the plane of the target area, the actual length corresponding to each image pixel is calculated, and the calibration results are written into the acquisition parameters.

[0303] The filter switching assembly includes a filter turntable driven by a stepper motor. The filter turntable has multiple filter mounting positions along its circumference, and each mounting position is respectively equipped with a long-wavelength pass filter, a 660 nm bandpass filter, an 850 nm bandpass filter, or a blank pass filter.

[0304] The stepper motor rotates the designated filter to the imaging optical path according to the instructions of the synchronization control module. After the ring is in position, the position sensor outputs a position signal; the image acquisition module only starts exposure after receiving the position signal. If the filter is not in position or the position feedback is inconsistent with the target position, the current frame will not be included in the valid image sequence.

[0305] When using time-division multi-band acquisition, the filter switching speed is consistent with the illumination switching sequence. When using high-speed acquisition mode, each band is configured with an independent image channel to avoid the mechanical ring speed limiting the acquisition frame rate.

[0306] Before formal data acquisition, the camera undergoes dark field calibration, pixel defect calibration, and linear response calibration. Pixels with grayscale values ​​above 95% of the camera's full scale are marked as saturated pixels, and pixels below 1% of full scale are marked as low-signal pixels. These pixels are not included in speckle contrast and target frequency response calculations.

[0307] In one possible implementation, the system further includes a modulation device for generating predetermined modulation conditions, and a synchronization control module enables the modulation device, image acquisition module, and data processing module to operate in a unified time relationship via a hardware trigger signal, a common clock, a timestamp, or a combination thereof.

[0308] The electrical stimulation device includes a pulse generator, an isolated output unit, and stimulation electrodes. A synchronization control module outputs a stimulation phase start signal to the pulse generator, which outputs pulses according to set parameters during the stimulation phase and stops outputting during the non-stimulation phase. The isolated output unit electrically isolates the stimulation circuit from the image acquisition equipment.

[0309] The mechanical stimulation device includes a drive motor, a displacement actuator, and a position sensor. A synchronization control module controls the drive motor to execute a set displacement, and the position sensor records the start and end times of the actual displacement. The actual displacement signal serves as the modulation time reference.

[0310] The tactile stimulation device uses a vibratory actuator, the visual stimulation device uses a display, the auditory or sound stimulation device uses a speaker, and the thermal stimulation device uses a heating element with temperature feedback. Each device returns its actual output status to the synchronization control module, recording not only the output time of the control commands.

[0311] When a common clock is used, the timing circuits of the image acquisition module, synchronization control module, and modulation device receive the same reference clock. When hardware triggering is used, the synchronization control module outputs pulses to the image acquisition module and modulation device respectively through independent trigger lines. When timestamps are used, each device records its local time and establishes a time conversion relationship through periodic time synchronization pulses.

[0312] After receiving each image frame, the data processing module matches the actual state of the modulation device according to the corresponding hardware timestamp. If the image acquisition time falls within the uncertain range of modulation state switching, the frame is marked as a switching frame. Whether a switching frame is included in the calculation is predetermined in the acquisition parameters; in this embodiment, switching frames are excluded.

[0313] By recording the actual output state of the modulation device, the modulation time reference offset caused by mechanical execution delay, temperature rise delay, and display refresh delay can be avoided.

[0314] In one possible implementation, the system further includes a display module and a storage module. The display module displays the cumulative response vascular images obtained up to the current cycle at the end of each predetermined modulation cycle, and the storage module saves the image data, processing results, and acquisition parameters.

[0315] After each modulation cycle, the data processing module updates the cumulative in-phase response component, cumulative quadrature response component, and response intensity at each pixel location, and generates the current cumulative response vascular image. Upon receiving the image, the display module displays the original image, the current response vascular image, the overlay of the response vascular image, and the number of completed cycles on the same display interface.

[0316] The response vessel overlay uses the original grayscale image as a background, and then overlays the response intensities onto the same spatial location after color mapping. The response vessel mask is used to limit the color coverage, ensuring that areas below the response threshold retain their original grayscale display. The display process does not alter the stored original response intensities.

[0317] The display module simultaneously shows the acquisition status, including the current lighting channel, current modulation stage, exposure time, acquisition frame rate, number of valid periods, number of abnormal periods, and number of remaining periods. If a module experiences trigger loss, image frame loss, or abnormal lighting power, the display interface will indicate the corresponding status.

[0318] The storage module establishes an independent data directory for each imaging task. The directory stores the original label-free optical image sequence, the corrected image, the pixel-level optical feature sequence, the cumulative response vascular images for each cycle, the final response vascular image, and the response vascular mask, respectively.

[0319] The acquisition parameter file records the image width and height, pixel depth, exposure time, acquisition frame rate, lens magnification, filter position, illumination wavelength, illumination power density, modulation type, duration of the stimulation phase, duration of the non-stimulation phase, number of modulation cycles, and threshold determination method.

[0320] The synchronization information file records the hardware timestamp for each frame, the time of each modulation state switch, the time synchronization pulse time, the abnormal period marker, and the illumination channel identifier. The processing result file records the feature type used, the local window size, the time-averaged frame count, the registration parameters, the fusion weights, and the response threshold.

[0321] Data is written to a temporary file first, then verified, and finally renamed to the official filename. If the data acquisition process is interrupted, the verified data is retained, and the task status is marked as incomplete.

[0322] In one possible implementation, a computer-readable storage medium stores a computer program or computer instructions, which, when executed by a processor, implement a label-free vascular imaging method.

[0323] Computer programs include image input programs, correction processing programs, feature calculation programs, synchronization processing programs, target frequency extraction programs, image generation programs, and data storage programs.

[0324] The image input program reads each frame of unmarked optical image, hardware timestamp, illumination channel, and acquisition parameters, and sorts the images according to the timestamp. If duplicate timestamps, missing frame numbers, or inconsistent image sizes are found, the corresponding data is marked as abnormal.

[0325] The correction process reads the dark-field image and the reference illumination image, and performs dark-field subtraction, reference illumination normalization, invalid pixel identification, and inter-frame spatial registration on the original image. The feature calculation program generates a reflection or scattering intensity feature sequence or a speckle dynamic blood flow feature sequence for each pixel location.

[0326] The synchronization processing program reads the modulation time base, determines the modulation state switching time, target frequency, and phase reference, and converts the acquisition time of each frame to a unified time base. The target frequency extraction program calculates the in-phase reference and quadrature reference based on the actual acquisition time, and performs cumulative projection or quadrature demodulation on each pixel position.

[0327] The image generation program determines the response intensity based on the in-phase and quadrature response components, generates a response vessel image according to pixel spatial coordinates, and generates a response vessel mask based on a preset response threshold. The data storage program saves the original data, processing parameters, intermediate results, and final results.

[0328] The processor reads image data in blocks during program execution. A preset number of consecutive images are loaded into memory each time, feature statistics and target frequency accumulations are updated for all pixel positions, and then the processed raw data cache is released. This processing method reduces the memory usage of long image sequences.

[0329] The computer-readable storage medium may be a solid-state memory, flash memory, magnetic storage medium, or optical storage medium. The program runs on a general-purpose processor, a graphics processor, or a processing device with parallel computing capabilities. All processing devices perform the same time alignment, feature calculation, orthogonal demodulation, and spatial mapping steps.

[0330] To ensure the processing results are verifiable, the program writes the version number, parameter file checksum, and input data checksum into the results file. When executed again using the same input data, the same program version, and the same processing parameters, the same response vascular image and response vascular mask will be output.

[0331] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for responding to label-free vascular imaging, characterized in that, include: S1. Acquire a label-free optical image sequence formed in the target area of ​​the target living body during the operation of a predetermined modulation condition, an acquisition time sequence corresponding to each frame of the label-free optical image sequence, and a modulation time reference characterizing the predetermined modulation condition. The label-free optical image sequence is formed based on the endogenous optical hemodynamic signal of the target area without the use of an exogenous optical contrast agent. S2. For each pixel position in the labelless optical image sequence, perform feature calculation on the image data of the corresponding pixel position at different acquisition times to obtain a pixel-level optical feature sequence F(x,y,n). The pixel-level optical feature sequence includes at least one of the reflection or scattering intensity feature sequence and speckle dynamic blood flow feature sequence. S3. Determine the target frequency corresponding to the predetermined modulation conditions based on the modulation time reference; when it is necessary to obtain the response phase, further determine the phase reference; S4. For each pixel position, based on the pixel-level optical feature sequence, acquisition time sequence, target frequency and phase reference of the corresponding pixel position, perform complex exponential projection operation at the target frequency or orthogonal demodulation operation mathematically equivalent to complex exponential projection operation to obtain the response coefficient of the corresponding pixel position at the target frequency. S5. Determine the response intensity of the corresponding pixel position based on the amplitude of the response coefficient, and map the response intensity according to the spatial coordinates of each pixel position to generate a response blood vessel image and / or response blood vessel mask for highlighting the blood vessels that respond at the target frequency.

2. The responsive label-free vascular imaging method according to claim 1, characterized in that, Intrinsic optical hemodynamic signals include at least one of the following: Changes in absorption intensity caused by variations in the concentrations of oxyhemoglobin and deoxyhemoglobin; Changes in reflection or scattering intensity caused by changes in blood volume, blood vessel diameter, or tissue scattering state; Dynamic changes in speckle patterns caused by changes in the motility of red blood cells or changes in blood flow velocity.

3. The responsive label-free vascular imaging method according to claim 1, characterized in that, When the pixel-level optical feature sequence includes a reflection or scattering intensity feature sequence, the original intensity I(x,y,n) at pixel position (x,y) in the nth frame image is subjected to dark field subtraction and reference illumination normalization to obtain the corrected reflection or scattering intensity: Where D(x,y) is the dark field intensity at pixel position (x,y) in the dark field image. The reference intensity is used for the pixel position (x,y) in the reference illumination image, and the corrected reflection or scattering intensity is used as the reflection or scattering intensity feature sequence.

4. The responsive label-free vascular imaging method according to claim 1 or 3, characterized in that, The illumination light used to form the reflection or scattering intensity characteristic sequence is at least one of incoherent light, low-coherence light, or partially coherent light, and the target area is simultaneously or time-divisionally illuminated using one or more different illumination bands.

5. The responsive label-free vascular imaging method according to claim 1, characterized in that, In the case where the pixel-level optical feature sequence includes speckle dynamic blood flow feature sequence, coherent illumination is used to illuminate the target area, and for the nth frame image, a local window W is selected centered at pixel position (x,y), and the average pixel gray value within the local window is calculated. and pixel grayscale standard deviation The speckle contrast is calculated using the following formula: And based on the speckle contrast, a speckle dynamic blood flow characteristic sequence is formed.

6. The responsive label-free vascular imaging method according to claim 5, characterized in that, Calculation of relative blood flow parameters based on speckle contrast: And by performing a time-averaged calculation on the relative blood flow parameters across multiple consecutive frames, including the current frame, we obtain: Where H is the number of frames included in the time-averaged window before or after the current frame, and the size of the local window W is 3×3, 5×5, 7×7 or other preset size.

7. The responsive label-free vascular imaging method according to claim 1, characterized in that, The first response vascular image and the second response vascular image are calculated based on the reflection or scattering intensity feature sequence and the speckle dynamic blood flow feature sequence, respectively. The first response vascular image and the second response vascular image are then spatially registered and fused to obtain the fused response vascular image. Alternatively, the reflection or scattering intensity feature sequence and the speckle dynamic blood flow feature sequence can be fused before performing complex exponential projection operation or orthogonal demodulation operation at the target frequency.

8. The responsive label-free vascular imaging method according to claim 1, characterized in that, The start of the predetermined modulation conditions and the start of image acquisition are controlled by the same synchronous trigger signal, or the timestamps of the predetermined modulation conditions and image acquisition are recorded separately and time-aligned so that the acquisition time sequence has a definite time correspondence with the modulation time reference.

9. The responsive label-free vascular imaging method according to claim 1 or 7, characterized in that, The predetermined modulation conditions repeat periodically at the target frequency f1, with a frame rate of f0 and a modulation period of: ； When the number of acquisition frames corresponding to each modulation cycle is an integer, the number of acquisition frames L in each modulation cycle and the total number of acquisition frames M after N modulation cycles are respectively: ； ； The acquisition time of the nth frame image is: ； Where n = 0, 1, ..., M-1.

10. The responsive label-free vascular imaging method according to claim 9, characterized in that, At the end of the m-th modulation period, the M data collected up to the current time are... m =mL frames of image are subjected to one cumulative target frequency extraction, where the response coefficient at the corresponding pixel location is: ； The response intensity at the corresponding pixel location is: ； Here, φ0 is a phase reference determined according to the modulation time base, and the displayed vascular image is updated at the end of each modulation cycle.

11. The responsive label-free vascular imaging method according to claim 10, characterized in that, The orthogonal demodulation operations mathematically equivalent to complex exponential projection operations include: Calculate the in-phase components: ； Calculate orthogonal components: ； The response intensity is calculated according to the following formula: 。 12. The method for responding to label-free vascular imaging according to claim 10, characterized in that, After completing a preset number of modulation cycles, the response vascular image obtained at the end of the last modulation cycle is taken as the final response vascular image: ； The response intensity of each pixel location is compared with a preset response threshold. Pixels whose response intensity meets the preset response threshold condition are identified as responsive blood vessel pixels to generate a responsive blood vessel mask.

13. The responsive label-free vascular imaging method according to claim 1, characterized in that, The predetermined modulation conditions include a stimulation phase and a non-stimulation phase. The target frequency is determined based on the repetition cycle formed by the stimulation phase and the non-stimulation phase, so that the static vascular component, the non-responsive vascular component, and the heartbeat component, respiratory component, and low-frequency drift component that deviate from the target frequency are suppressed during the target frequency extraction process.

14. The responsive label-free vascular imaging method according to claim 1, characterized in that, The predetermined modulation conditions include at least one of electrical stimulation, mechanical stimulation, tactile stimulation, visual stimulation, auditory stimulation, sound stimulation, thermal stimulation, physiological modulation, task induction, and drug action; The target area includes at least one of the following: brain, spinal cord, retina, hand, foot, limb, tumor area, inflamed area, or other living tissue area.

15. The responsive label-free vascular imaging method according to claim 1, characterized in that: Illumination light used to form a sequence of reflection or scattering intensity characteristics includes incoherent, low-coherence, or partially coherent illumination light with a wavelength of 660 nm and / or incoherent, low-coherence, or partially coherent illumination light with a wavelength of 850 nm. The illumination light used to form the speckle dynamic blood flow characteristic sequence includes coherent illumination light with a wavelength of 850 nm; The lighting power density ranges from 1 mW / cm² to 200 mW / cm². The exposure time for image acquisition is 1ms to 1000ms, and the acquisition frame rate is 1Hz to 1000Hz, or other exposure times and acquisition frame rates can be set according to the target frequency and the required time resolution.

16. A responsive label-free vascular imaging system, applicable to the responsive label-free vascular imaging method according to any one of claims 1-15, characterized in that, include: An illumination module is used to illuminate the target area of ​​a living organism with a first illumination light that does not produce resolvable speckle and / or a second illumination light that produces speckle, without the use of an external optical contrast agent. The image acquisition module is used to acquire a sequence of unmarked optical images of the target area during the operation of predetermined modulation conditions, and to record the acquisition time corresponding to each frame of the image. The synchronization control module is used to receive a modulation time reference characterizing predetermined modulation conditions, or to output control signals to the modulation device used to generate predetermined modulation conditions, and to make the acquisition timing of the image acquisition module correspond to the modulation time reference. The data processing module is used for: The labelless optical image sequence is converted into a pixel-level optical feature sequence for each pixel location. The pixel-level optical feature sequence includes at least one of the reflection or scattering intensity feature sequence and the speckle dynamic blood flow feature sequence. The target frequency and phase reference are determined based on the modulation time base; For each pixel location, based on the pixel-level optical feature sequence, acquisition time, target frequency, and phase reference of the corresponding pixel location, perform a complex exponential projection operation at the target frequency or an orthogonal demodulation operation mathematically equivalent to the complex exponential projection operation to obtain the response coefficient of the corresponding pixel location. The response intensity of the corresponding pixel position is determined based on the magnitude of the response coefficient, and a response blood vessel image and / or response blood vessel mask is generated based on the spatial coordinates of each pixel position.

17. The responsive label-free vascular imaging system according to claim 16, characterized in that, The lighting module includes: The first illumination source is used to output incoherent light, low-coherence light, or partially coherent light to form a sequence of reflection or scattering intensity characteristics; A second illumination source is used to output coherent light to form a speckle dynamic blood flow characteristic sequence; A lighting switching component is used to selectively enable the first and second lighting sources to operate independently, synchronously, or in a time-sharing manner.

18. The responsive label-free vascular imaging system according to claim 17, characterized in that, The lighting module also includes a light homogenization component, a collimation component, and a light spot adjustment component; The light homogenization component is used to spatially homogenize illumination light; The collimation component is used to output a collimated illumination beam; The spot adjustment component is used to adjust the size of the illumination spot by changing the illumination distance, the position of the optical element, or a combination of optical elements; The lighting module can change the output lighting light to either a speckled or non-distinguishable speckled state by replacing or switching the light homogenization processing component.

19. The responsive label-free vascular imaging system according to claim 16, characterized in that, The lighting module includes a 660nm lighting source and an 850nm lighting source, and the lighting power density can be adjusted from 1mW / cm² to 200mW / cm². The lighting module also includes a water-cooling heat dissipation component for cooling the lighting source to maintain the stability of the light power of the lighting source under continuous output conditions.

20. The responsive label-free vascular imaging system according to claim 16, characterized in that, The image acquisition module includes: sCMOS camera with a spectral response range covering 400nm to 1000nm; The interchangeable imaging lens has a magnification of at least one of 0.1x, 0.2x, 0.5x, and 1x. A filter switching assembly includes a filter turntable driven by a stepper motor and a plurality of long-pass filters and / or band-pass filters disposed on the filter turntable.

21. The responsive label-free vascular imaging system according to claim 16, characterized in that, It also includes a modulation device for generating predetermined modulation conditions, the modulation device including at least one of an electrical stimulation device, a mechanical stimulation device, a tactile stimulation device, a visual stimulation device, an auditory stimulation device, a sound stimulation device, or a thermal stimulation device. The synchronization control module enables the modulation device, image acquisition module, and data processing module to operate synchronously through hardware trigger signals, a common clock, a timestamp, or a combination thereof.

22. The responsive label-free vascular imaging system according to claim 16, characterized in that, It also includes a display module and a storage module; The display module is used to display the cumulative response vascular images up to the current cycle at the end of each predetermined modulation cycle; The storage module is used to store at least one of the following: response vascular image, response vascular mask, original label-free optical image sequence, pixel-level optical feature sequence, image acquisition parameters, illumination parameters, predetermined modulation parameters, and synchronization time information.

23. A computer-readable storage medium having a computer program or computer instructions stored thereon, characterized in that, When the computer program or computer instructions are executed by the processor, the responsive label-free imaging method of any one of claims 1 to 15 is implemented.

Images