Method, system and storage medium for real-time monitoring of blood labyrinth barrier permeability

Abstract

This application relates to the field of biomedical monitoring technology, and discloses a method, system, and storage medium for real-time monitoring of blood labyrinth barrier permeability. The method includes: acquiring raw signal data and flow velocity measurement data of a fluorescent tracer within the blood labyrinth barrier region; preprocessing these data to obtain a fluorescence signal sequence and a flow velocity data sequence; classifying the sampling segments corresponding to the fluorescence signal sequence according to the flow velocity data sequence to determine high-flow velocity and low-flow velocity segments; differentially processing the fluorescence signals corresponding to the high-flow velocity and low-flow velocity segments to obtain a high-flow velocity compensation sequence and a low-flow velocity integral sequence, integrating them to obtain an overall compensation sequence; separating the flow contribution based on the temporal correspondence between the overall compensation sequence and the flow velocity data sequence to obtain a pure concentration change sequence; calculating the blood labyrinth barrier permeability value based on the pure concentration change sequence, and outputting the permeability monitoring result. This invention can improve the accuracy and real-time performance of blood labyrinth barrier permeability monitoring.

Description

Technical Field

[0001] This application relates to the field of biomedical monitoring technology, and in particular to a method, system and storage medium for real-time monitoring of blood labyrinth barrier permeability. Background Technology

[0002] The blood-labyrinthine barrier is a crucial physiological barrier structure for maintaining the stability of the inner ear microenvironment. Changes in its integrity and permeability are closely related to various pathological processes, including sudden deafness, Meniere's disease, drug-induced ototoxicity, and inflammatory inner ear diseases. Accurate monitoring of blood-labyrinthine barrier permeability helps to elucidate inner ear pathological processes, assess drug delivery efficacy, and provide technical support for disease diagnosis, efficacy evaluation, and mechanistic research. Therefore, the detection and analysis of blood-labyrinthine barrier permeability has always been an important research direction in the fields of otology, biomedical testing, and medical signal processing.

[0003] Existing techniques for assessing blood labyrinth barrier permeability primarily rely on methods such as tissue staining, in vitro sampling, biochemical testing, and pathological observation. These methods indirectly assess the state of the blood labyrinth barrier by detecting the degree of tracer exudation or changes in tissue morphology. While these methods can reflect barrier damage to some extent, they are typically endpoint detection or offline analysis methods, making it difficult to continuously monitor the dynamic changes in the blood labyrinth barrier. Furthermore, they suffer from problems such as complex operation, high invasiveness, and insufficient timeliness.

[0004] With the development of optical detection and sensor technologies, in vivo detection methods based on fluorescent tracers are increasingly being applied to the study of biological barrier permeability. By introducing fluorescent tracers into the target area and collecting changes in fluorescence signals, the distribution, penetration, and clearance processes of the tracers within the local area can be reflected to some extent. Furthermore, with the development of microcirculation detection, local flow velocity measurement, and multi-source data fusion processing technologies, combining fluorescence detection signals with dynamic information such as local blood flow and lymphatic flow to analyze tracer concentration changes under complex physiological environments has become an important technological development direction for improving the accuracy of biological barrier permeability assessment.

[0005] However, within the blood-labyrinth barrier region, fluorescence signals are not only affected by changes in the true concentration of the tracer, but are also susceptible to interference from various factors such as changes in local blood flow velocity, alterations in lymphatic flow patterns, boundary segment transition disturbances, transient fluctuation noise, baseline drift, and signal abrupt changes. Particularly under different flow conditions, fluorescence signals exhibit different characteristics: high flow rates are prone to spikes, rapid fluctuations, and baseline disturbances, while low flow rates are prone to weak signals, slow changes, local abrupt changes, and integral distortion. In existing techniques, uniformly processing fluorescence signals often fails to adequately account for the signal characteristics under different flow conditions, leading to insufficient signal compensation or over-correction, which in turn affects the reliability of subsequent concentration change analysis and permeability calculations.

[0006] Furthermore, existing technologies for the co-processing of fluorescence signals and flow velocity data typically lack differentiated processing mechanisms for high-velocity and low-velocity regions, as well as processing pathways for effectively separating the contributions of flow factors. Since local flow variations significantly impact fluorescence signals, if an accurate temporal correspondence between fluorescence signals and flow velocity data cannot be established, and the flow-induced signal contribution components cannot be separated from this data, it will be difficult to obtain an effective signal that truly characterizes changes in tracer concentration. Ultimately, this will affect the accuracy of calculating blood labyrinth barrier permeability values ​​and the real-time nature of monitoring results.

[0007] Therefore, how to establish a real-time monitoring method that can differentiate fluorescence signals under different flow states, effectively separate flow contributions, and accurately calculate the permeability value of the blood labyrinth barrier, based on the coupling characteristics of fluorescence signals and flow velocity data within the blood labyrinth barrier region, has become a key technical problem that urgently needs to be solved in this field. Summary of the Invention

[0008] To address the problems in existing technologies, such as the difficulty in achieving real-time continuous monitoring of blood labyrinth barrier permeability, the susceptibility of fluorescence signals to interference from changes in local flow states, the lack of targeted signal processing under different flow states, and the difficulty in effectively separating flow factors from actual concentration changes, this application provides a method, system, and storage medium for real-time monitoring of blood labyrinth barrier permeability, in order to improve the accuracy, real-time performance, and stability of blood labyrinth barrier permeability monitoring.

[0009] In a first aspect, this application provides a method for real-time monitoring of blood labyrinth barrier permeability, the method comprising: S1. Obtain the raw signal data and flow rate measurement data of the fluorescent tracer in the blood labyrinth barrier region, and obtain the fluorescence signal sequence and flow rate data sequence after preprocessing; S2. Based on the flow velocity data sequence, the sampling segments corresponding to the fluorescence signal sequence are divided into regions to determine the high flow velocity segment and the low flow velocity segment; S3. For the high-velocity and low-velocity sections, extract the corresponding fluorescence signal sequences and perform differential processing to obtain the high-velocity compensation sequence and the low-velocity integral sequence. S4. Based on the region type division results, integrate the high-velocity compensation sequence and the low-velocity integral sequence to obtain the overall compensation sequence; S5. Based on the temporal correspondence between the overall compensation sequence and the velocity data sequence, separate the flow contribution and obtain the pure concentration change sequence; S6. Calculate the blood labyrinth barrier permeability value based on the pure concentration change sequence and output the permeability monitoring results.

[0010] Secondly, this application provides a real-time monitoring system for blood labyrinth barrier permeability, the system comprising: The data acquisition module is used to acquire the raw signal data and flow rate measurement data of the fluorescent tracer in the blood labyrinth barrier region, and to acquire the fluorescence signal sequence and flow rate data sequence after preprocessing. The region segmentation module is used to divide the sampling segments corresponding to the fluorescence signal sequence into region types based on the flow velocity data sequence, and to determine the high flow velocity segment and the low flow velocity segment; The dual-flow compensation module is used to extract the corresponding fluorescence signal sequences for high-flow-velocity and low-flow-velocity sections respectively and perform differential processing to obtain high-flow-velocity compensation sequences and low-flow-velocity integral sequences. The signal integration module is used to integrate the high-velocity compensation sequence and the low-velocity integral sequence according to the region type division results to obtain the overall compensation sequence; The flow separation module is used to separate the flow contribution based on the temporal correspondence between the overall compensation sequence and the flow velocity data sequence, and to obtain the pure concentration change sequence. The permeability calculation module is used to calculate the blood labyrinth barrier permeability value based on the pure concentration change sequence and output the permeability monitoring results.

[0011] Thirdly, this application provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the above-described method for real-time monitoring of blood labyrinth barrier permeability.

[0012] The beneficial effects of this application are at least as follows: Combining fluorescence signals with flow velocity data for monitoring blood-labyrinth barrier permeability significantly enhances the characterization of tracer dynamic distribution by introducing local blood flow and lymph flow information to provide a flow state reference for fluorescence signal changes. Sampling sections are categorized based on flow velocity data, with differentiated processing mechanisms applied to high-velocity and low-velocity sections. For high-velocity sections, spike interference, baseline drift, and high-frequency noise are effectively suppressed; for low-velocity sections, weak signal expression is enhanced, local mutations are suppressed, and a slow change trend is maintained, significantly improving the targeting and effectiveness of signal processing. By aligning the high-velocity compensation sequence with the time axis, adjusting the amplitude, splicing multiple segments, and smoothing the boundaries of the low-velocity integral sequence, a more continuous and complete overall compensation sequence is constructed, effectively reducing splicing distortion and boundary mutations caused by branching. By establishing a temporal correspondence between the overall compensation sequence and the flow velocity data sequence, the signal contribution components caused by flow factors are separated, and pure concentration change sequences are extracted from the mixed signal, significantly reducing the interference of flow disturbances on permeability assessment. Based on the extraction of signal intensity change rate and long-term trend from pure concentration change sequences, the permeability value of the blood labyrinth barrier is calculated and the monitoring results are output in real time, which significantly improves the accuracy and stability of permeability detection and provides a reliable technical means for inner ear disease research, drug delivery evaluation and related pathological mechanism analysis. Attached Figure Description

[0013] To more clearly illustrate the technical solutions of 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 some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a flowchart of the method for real-time monitoring of blood labyrinth barrier permeability according to this application; Figure 2 This is a schematic diagram of the flow contribution separation results in an embodiment of this application; Figure 3 This is a schematic diagram representing the real-time monitoring results of blood labyrinth barrier permeability in an embodiment of this application; Figure 4 This is a schematic diagram of the structure of the real-time monitoring system for blood labyrinth barrier permeability of this application. Detailed Implementation

[0015] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms “comprising” or “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0016] For ease of understanding, the specific process of the embodiments of this application is described below. Figure 1 The diagram shows a flowchart of the real-time monitoring method for blood labyrinth barrier permeability provided by the present invention. The flowchart specifically includes the following steps: S1. Obtain the raw signal data and flow rate measurement data of the fluorescent tracer within the blood labyrinth barrier region, and obtain the fluorescence signal sequence and flow rate data sequence after preprocessing.

[0017] In one specific embodiment, the process of performing step S1 may specifically include the following steps: The sensor probe simultaneously acquires raw signal data of the fluorescent tracer, local blood flow velocity measurement data, and lymph flow velocity measurement data. The original signal data were arranged into an initial fluorescence signal sequence according to the sampling time order; Local blood flow velocity measurement data and lymph flow velocity measurement data were integrated into an initial flow velocity data sequence according to the sampling time sequence; The initial fluorescence signal sequence and the initial flow velocity data sequence were time-synchronized preprocessed to obtain signal-flow velocity data pairs. The integrity of the signal and flow velocity data pairs is verified. If the verification result meets the preset standard, the fluorescence signal sequence and flow velocity data sequence are output.

[0018] Specifically, the sensor probe is positioned near the target monitoring area of ​​the blood-labyrinth barrier and includes at least a fluorescence acquisition channel, a blood flow velocity measurement channel, a lymphatic fluid velocity measurement channel, and a buffer unit. The fluorescence acquisition channel receives the optical response generated after the fluorescent tracer is excited and converts the optical response into an electrical signal to output the raw signal data. The blood flow velocity measurement channel outputs local blood flow velocity measurement data, and the lymphatic fluid velocity measurement channel outputs lymphatic fluid flow velocity measurement data. The buffer unit temporarily stores the data output by each channel according to the sampling time. The fluorescence acquisition channel can acquire the fluorescence response by combining an excitation light source with a light-receiving device. The analog signal output by the light-receiving device is amplified and converted from analog to digital to form fluorescence amplitudes arranged discretely over time. The blood flow velocity measurement channel and the lymphatic fluid velocity measurement channel can output velocity measurement values ​​using a Doppler velocimetry method suitable for microcirculation scenarios. By setting the three acquisition channels to a common clock trigger mode, the fluorescence amplitude, local blood flow velocity value, and lymphatic fluid flow velocity value within the same sampling period correspond to the same sampling time, avoiding time offsets caused by independent timing of each channel.

[0019] During the sequencing phase, the fluorescence amplitude values ​​of each sampling point output from the fluorescence acquisition channel are written into a one-dimensional time series according to the sampling time sequence to form an initial fluorescence signal sequence. Simultaneously, the local blood flow velocity values ​​and lymph flow velocity values ​​corresponding to the same sampling time are written into a composite flow velocity sequence according to the same time sequence to form an initial flow velocity data sequence. Each sampling location in the initial flow velocity data sequence contains local blood flow velocity values ​​and lymph flow velocity values ​​at the same time.

[0020] The initial fluorescence signal sequence and initial flow velocity data sequence undergo background subtraction, abnormal null value marking, dimensional unification, and timestamp standardization. Background subtraction removes the fixed bias caused by dark current of the light-receiving device and the background response of the probe, preventing baseline offset from being directly carried over to subsequent processing. Abnormal null value marking identifies empty sampling locations caused by analog-to-digital conversion failure, velocity measurement failure, or communication buffer packet loss. Dimensional unification converts local blood flow velocity values ​​and lymph flow velocity values ​​to the same velocity unit. Timestamp standardization maps each sampling point to a unified time axis. If there is a difference in the actual sampling period between the fluorescence acquisition channel and the flow velocity acquisition channel, the time axis of the higher sampling frequency channel is used as the reference time axis, and resampling is performed on the lower sampling frequency channel. Resampling can be performed using linear interpolation, i.e., the interpolation result is calculated between two adjacent valid sampling points according to the relative position at the target time. Time synchronization is performed using the timestamp-standardized initial fluorescence signal sequence and initial flow velocity data sequence as input. Time synchronization includes coarse alignment and fine alignment. Coarse alignment maps data from each channel to the same reference time axis by unifying the hardware trigger time. Fine alignment is used to correct residual time deviations caused by probe response delays and signal processing delays. Fine alignment can be achieved using a cross-correlation search method. Within a preset time offset range, the cross-correlation value between the initial fluorescence signal sequence and the initial flow velocity data sequence at different time offsets is calculated. The time offset corresponding to the peak cross-correlation value is used as a compensation amount to correct the time shift of the initial fluorescence signal sequence or the initial flow velocity data sequence. This processing is adopted because the influence of local flow changes on the fluorescence response usually has a short-term hysteresis. If this hysteresis is not eliminated first, when subsequent determination of high-velocity and low-velocity segments based on the flow velocity data sequence, the extracted fluorescence fragments will be misaligned with the actual flow state, leading to misjudgment of time misalignment as flow contribution. After alignment, each sampling time corresponds to a fluorescence value, a local blood flow velocity value, and a lymph flow velocity value. The fluorescence signal sequence obtained after background subtraction, time synchronization, anomaly handling, and necessary resampling or missing data filling is the basic corrected fluorescence signal sequence.

[0021] Integrity verification includes sampling integrity verification, continuity verification, effective range verification, and synchronization consistency verification. Sampling integrity verification determines whether the proportion of null points, duplicate points, and packet loss points within the current time window is lower than a preset threshold. Continuity verification determines whether the time interval between adjacent sampling points is within a preset tolerance range. Effective range verification determines whether the fluorescence value is within the sensor's linear response range and whether the local blood flow velocity and lymph flow velocity values ​​are within the physically measurable range of the corresponding velocity measurement channel. Synchronization consistency verification determines whether fluorescence values, local blood flow velocity values, and lymph flow velocity values ​​exist simultaneously under the same time index. The preset thresholds, tolerance ranges, and measurable ranges are pre-set based on sensor calibration parameters, sampling frequency, and the physiological range of the target object. For time windows with fewer missing points than a preset proportion and discontinuous missing locations, neighborhood interpolation is used to fill in the gaps. Continuously missing, discontinuous, or abnormal segments exceeding the effective range are discarded. When the verification and anomaly handling results meet the preset standards, the fluorescence signal sequence and flow velocity data sequence are output.

[0022] S2. Based on the flow velocity data sequence, the sampling segments corresponding to the fluorescence signal sequence are divided into region types to determine the high flow velocity segment and the low flow velocity segment.

[0023] In one specific embodiment, the process of performing step S2 may specifically include the following steps: For the flow velocity data sequence, extract the local blood flow velocity data and lymph flow velocity data corresponding to each sampling time point; The local blood flow velocity data is compared with the preset blood flow velocity threshold. The continuous sampling time points where the local blood flow velocity data is greater than the preset blood flow velocity threshold are determined as high flow velocity time windows. The sequence segments in the fluorescence signal sequence corresponding to the high flow velocity time windows are determined as high flow velocity segments. For sampling time points that are not identified as high-flow-rate segments, the lymph flow velocity data is compared with a preset lymph flow velocity threshold. Continuous sampling time points where the lymph flow velocity data is less than the preset lymph flow velocity threshold are identified as low-flow-rate time windows. Sequence segments in the fluorescence signal sequence that correspond to the low-flow-rate time windows are identified as low-flow-rate segments. The judgment results are summarized to generate region type classification results aligned with the time axis of the fluorescence signal sequence.

[0024] Specifically, after obtaining the local blood flow velocity data and lymph flow velocity data corresponding to each sampling time point, the local blood flow velocity data is first compared with a preset blood flow velocity threshold. When the local blood flow velocity value at a certain sampling time point is greater than the preset blood flow velocity threshold, the sampling time point is marked as a high flow velocity point; when the local blood flow velocity value is not greater than the preset blood flow velocity threshold, the sampling time point is retained as a candidate point for subsequent low flow velocity determination. The local blood flow velocity threshold comparison is performed first because when the local blood flow velocity increases, the enhanced transport of fluorescent tracers in the blood labyrinth barrier region more directly affects the fluorescence signal morphology, manifested as increased peak value, rapid local fluctuations, and baseline perturbation. If such sampling points are not identified first and are directly judged alongside the low flow velocity condition, sampling points where the local blood flow velocity has increased but the lymph flow velocity is also low will be mistakenly classified into the low flow velocity segment, causing the subsequent low flow velocity integration processing path to inherit high flow velocity interference signals.

[0025] All high-flow-rate points are sequentially merged according to their time index to form high-flow-rate time windows. During merging, high-flow-rate points that are temporally adjacent and have no other valid sampling points in between are included in the same high-flow-rate time window. For cases where the sampling periods are not strictly equal, the time interval between adjacent high-flow-rate points can be compared with a preset continuity tolerance. If the time interval does not exceed the preset continuity tolerance, they are still included in the same high-flow-rate time window; if it exceeds the preset continuity tolerance, the current high-flow-rate time window ends and a new high-flow-rate time window begins. For cases where only a single sampling time point satisfies the condition that the local blood flow velocity value is greater than a preset blood flow velocity threshold, this single sampling time point is considered a high-flow-rate time window of length one. After forming the high-flow-rate time windows, the sequence segments in the fluorescence signal sequence that correspond one-to-one with the time index of each high-flow-rate time window are identified as high-flow-rate segments.

[0026] For sampling time points not identified as high-flow-rate segments, if the lymph flow velocity value at a given sampling time point is less than a preset lymph flow velocity threshold, that sampling time point is marked as a low-flow-rate point; if the lymph flow velocity value is not less than the preset lymph flow velocity threshold, that sampling time point is not included in the low-flow-rate segment. The preset lymph flow velocity threshold is used to characterize the state of slowed local lymph flow; under this state, the tracer is more likely to exhibit enhanced retention, slowed clearance, and low-amplitude slow changes in the local area. All low-flow-rate points are merged according to the same continuity merging rules as high-flow-rate time windows to form low-flow-rate time windows; for cases where only a single sampling time point meets the condition that the lymph flow velocity value is less than the preset lymph flow velocity threshold, that single sampling time point is considered a low-flow-rate time window of length one. After forming the low-flow-rate time windows, the sequence segments in the fluorescence signal sequence that correspond one-to-one with the time indices of each low-flow-rate time window are identified as low-flow-rate segments.

[0027] The preset blood flow velocity threshold and preset lymph flow velocity threshold can be set as fixed thresholds, or determined based on the calibration samples, the device's velocity measurement range, or statistical results within the baseline sampling window. For example, a high flow velocity threshold can be obtained by calculating the quantile threshold or the mean plus offset from the local blood flow velocity data within the baseline sampling window; similarly, a low flow velocity threshold can be obtained by calculating the low quantile threshold from the lymph flow velocity data within the baseline sampling window.

[0028] The determination results of high-flow-rate and low-flow-rate sections are summarized according to the time index of the fluorescence signal sequence to generate a region type classification result aligned with the time axis of the fluorescence signal sequence. The sampling points whose time index corresponds to the high-flow-rate section are marked as high-flow-rate type, the sampling points whose time index corresponds to the low-flow-rate section are marked as low-flow-rate type, and the remaining sampling points are marked as regular type.

[0029] Preferably, to avoid isolated points being caused by occasional noise, a neighborhood consistency check can be performed on the single-point window after forming a high-velocity or low-velocity time window of length one. The neighborhood consistency check can be based on the velocity change direction and fluorescence change amplitude of adjacent sampling points before and after the single point. When the velocity change direction between the single point and its adjacent sampling points is discontinuous, and the fluorescence change amplitude does not support the corresponding abrupt change in flow state, the single-point window can be marked as a window to be checked for subsequent anomalies, instead of being directly removed in this step. This processing method does not change the generation rules of the segmentation results; it is only used to avoid mixing instantaneous valid events with occasional noise during the segmentation stage.

[0030] S3. For the high-velocity and low-velocity sections, extract the corresponding fluorescence signal sequences and perform differential processing to obtain the high-velocity compensation sequence and the low-velocity integral sequence.

[0031] In one specific embodiment, in S3, for the high-flow-rate section, the fluorescence signal sequence is processed to obtain a high-flow-rate compensation sequence, including: Extract sequence fragments corresponding to each high-flow-rate region from the fluorescence signal sequence; For each extracted sequence segment, signal peak sharpness suppression processing and baseline drift dynamic tracking correction are performed sequentially to obtain the corresponding preliminary corrected segment; The fluctuation amplitude of each preliminary correction segment is sequentially determined to exceed the preset amplitude threshold. If so, local smoothing interpolation processing is performed on the corresponding preliminary correction segment for amplitude anomaly points. A high-frequency noise segmentation filtering mechanism is adopted to perform noise reduction processing on the preliminary correction segments after smooth interpolation and the preliminary correction segments that do not exceed the preset amplitude threshold. Based on the processed segments, a high flow rate compensation sequence is generated.

[0032] Specifically, for high-flow-rate and low-flow-rate segments, corresponding fluorescence signal sequences are extracted and processed differently according to different signal processing mechanisms. This is because under high-flow-rate conditions, the fluorescent tracer is affected by rapid blood flow transport and instantaneous scouring, making it more prone to spikes, jumps, baseline drift, and high-frequency perturbations in the signal. Under low-flow-rate conditions, the fluorescence signal is more characterized by stagnation enhancement, slow diffusion, and low-amplitude slow changes. If a uniform processing mechanism is used, the instantaneous anomalies in the high-flow-rate segment and the effective slow-change information in the low-flow-rate segment will be weakened together, which cannot correspond to the processing goals of subsequent segment integration and flow contribution separation.

[0033] Scan all sampling points in the time index order of the fluorescence signal sequence, merge consecutive sampling points marked as high flow rate into the same high flow rate segment, and record the start and end positions of the segment in the fluorescence signal sequence; for high flow rate segments that are not continuous on the time axis, they are treated as independent segments for subsequent processing.

[0034] Signal peak sharpness suppression is used to remove isolated spikes and narrow pulse-like anomalies caused by localized rapid scouring under high-velocity conditions, preventing a single extreme point from affecting subsequent baseline estimation and fluctuation amplitude judgment. In implementation, a local sliding window can be set within the current high-velocity segment, using the median or weighted average within the window as a neighborhood reference value. When the deviation of the current sampling point from the neighborhood reference value exceeds a preset spike determination threshold, the sampling point is identified as a sharp peak point, and the median, mean, or local linear interpolation result of non-abnormal sampling points in the neighborhood is used to replace or limit the sampling point. The spike determination threshold can be determined based on the local statistical results of the absolute value of the signal difference between adjacent sampling points within the current high-velocity segment, and can be corrected in conjunction with calibration experimental results. The sliding window length can be set according to the sampling frequency and the expected spike duration, preferably an odd number of lengths to keep the current sampling point at the center of the window. Spike suppression weakens transient anomalies within the high-velocity segment without changing the temporal order of the main changing trends within that segment. After peak suppression processing, baseline drift dynamic tracking correction is performed on the same high-flow-rate segment to remove slow baseline rises or falls caused by background scattering changes, local tissue micro-displacements, or increased blood flow under high-flow-rate conditions. During implementation, a slowly changing baseline trajectory is constructed along the time axis for the segment after peak suppression, and the baseline trajectory is subtracted from the segment signal to obtain the initially corrected segment. The baseline trajectory can be obtained through long-window smoothing, local low-quantile tracking, or sliding lower envelope extraction. For example, local low-quantile values ​​can be extracted within a preset-length sliding window and connected in chronological order to form a baseline sequence, which is then subtracted point-by-point from the segment signal. In this way, high-frequency peak anomalies and slowly varying baseline shifts in the initially corrected segment are weakened, and the remaining signal within the segment is closer to the effective fluctuations that need to be retained.

[0035] The fluctuation amplitude is used to measure the intensity of local abnormal fluctuations that still exist within a segment after the aforementioned two steps of processing, in order to distinguish between normal high-velocity changes and residual abnormal fluctuations. The fluctuation amplitude is calculated based on the statistical measures of the difference between the maximum and minimum values ​​within the current segment, the absolute values ​​of the differences between adjacent sampling points, or the statistical measures of the dispersion within the segment. When the fluctuation amplitude exceeds a preset amplitude threshold, it indicates that there are still local abnormal points or clusters of abnormal points in the segment, requiring further local smoothing interpolation processing for amplitude abnormal points. The preset amplitude threshold can be set based on the device noise baseline, the normal high-velocity fluctuation range in the calibration samples, or the steady-state noise statistics within the baseline sampling window. The local smoothing interpolation processing for amplitude abnormal points only corrects the identified local abnormal locations within the segment, without smoothing the entire segment as a whole, to avoid excessively weakening the still meaningful rapid changes in the high-velocity section. First, abnormal points are detected point by point within the current segment. The detection criterion can be set to whether the deviation between the current sampling point and the mean, median, or local trend line of its preceding and following neighbors exceeds a preset abnormal threshold. For the identified abnormal points, the nearest non-abnormal sampling points before and after the abnormal point are used as constraint points, and smoothing interpolation is performed within a local window. Smoothing interpolation can employ linear interpolation, piecewise cubic interpolation, or locally weighted regression interpolation. If linear interpolation is used, the local trend is determined by the nearest valid point on the left and right sides of the anomaly window, and the value of this trend at the anomaly sampling time replaces the original outlier value.

[0036] After completing local smoothing interpolation, a high-frequency noise segmented filtering mechanism is uniformly applied to the pre-corrected segments after smoothing interpolation and those that do not exceed the preset amplitude threshold for noise reduction. This mechanism removes fast random noise superimposed on the effective signal within high-velocity segments while avoiding boundary information aliasing caused by cross-segment filtering. Each high-velocity segment is treated as an independent processing unit for filtering, without performing continuous filtering across segment boundaries. Filtering methods can include short-window moving average, low-pass filtering, or discrete wavelet thresholding. If low-pass filtering is used, the cutoff frequency can be set based on the sampling frequency and the target signal bandwidth to suppress noise components exceeding the target bandwidth without weakening the main trend within the segment. If short-window moving average is used, boundary shrinking or mirror extension is applied at segment boundaries to avoid edge distortion.

[0037] After all high-flow-rate segments have undergone the above processing, each segment is backfilled according to its original time position in the fluorescence signal sequence to generate a high-flow-rate compensated sequence. Specifically, the sampling time points corresponding to high-flow-rate segments are written with the processed fluorescence signal values, while the sampling time points corresponding to non-high-flow-rate segments are either not written with values ​​or remain as placeholders.

[0038] Preferably, for high-velocity segments of length one, since they lack independent intra-segment neighborhood information, baseline trajectory estimation conditions, and segment-level fluctuation amplitude calculation conditions, they are not processed according to the conventional segment processing chain. Instead, the sampling point corresponding to the single-point segment is mapped back to the original fluorescence signal sequence, and an extended neighborhood is formed by combining it with the adjacent valid sampling points before and after it. Single-point consistency verification, local baseline reference correction, and single-point anomaly replacement processing are performed. When the single point is consistent with the adjacent valid sampling points in terms of fluorescence change direction and velocity change trend, the corrected single-point result is retained. When the single point lacks neighborhood support or is significantly inconsistent with the neighborhood trend, it is marked as an abnormal single point and removed or replaced with neighborhood interpolation results during subsequent sequence integration.

[0039] In one specific embodiment, in S3, for the low-flow-rate section, the fluorescence signal sequence is processed to obtain a low-flow-rate integral sequence, including: Sequence fragments corresponding to each low-flow-rate segment were extracted from the fluorescence signal sequence; For each extracted sequence segment, signal envelope reconstruction adjustment and adjacent sampling point slope constraint processing are performed sequentially to obtain the corresponding preliminary adjusted segment; Sequentially determine whether the signal strength of each preliminary adjustment segment is lower than the preset strength threshold. If so, perform delayed sampling compensation and instantaneous strength change suppression on the corresponding preliminary adjustment segment. A threshold following mechanism is adopted to perform threshold following adjustment on the preliminary adjustment segments that have undergone delayed sampling compensation and instantaneous intensity mutation suppression, as well as the preliminary adjustment segments that are not lower than the preset intensity threshold. A low flow rate integral sequence is generated based on each adjusted segment.

[0040] Specifically, all sampling points are scanned sequentially along the time index of the fluorescence signal sequence. Continuous sampling points marked as low flow rate types are merged into the same low flow rate segment, and the start and end positions of each low flow rate segment in the original fluorescence signal sequence are recorded. If multiple low flow rate segments are separated from each other on the time axis, they form multiple independent segments.

[0041] Signal envelope reconstruction adjustment is used to correct envelope discontinuities caused by weak signal attenuation, local sampling jitter, or short-term drop-offs under low flow velocity conditions, making the overall trend of the segment more closely resemble the slow retention and gradual exchange of tracers in local regions. A sliding window is set within the current low flow velocity segment, and local peaks and valleys are extracted within each window to form upper and lower envelope point sets. These sets are then connected by interpolation to form upper and lower envelope lines. The central trend between the upper and lower envelope lines, or a local low envelope line, is used as a reconstruction reference to reshape the trend of the original segment. For weak signal segments, to avoid individual sudden increases causing an upward shift in the overall trend, a local low envelope line or the intermediate trend between the upper and lower envelope lines is used as the adjustment benchmark. After envelope reconstruction adjustment, local gaps, local breaks, or sharp fluctuations that do not conform to the slow change characteristics of low flow velocities are weakened.

[0042] After signal envelope reconstruction and adjustment, slope constraint processing is applied to adjacent sampling points for the same low-flow-rate segment to limit abrupt changes between adjacent sampling points that do not conform to the characteristics of low-flow-rate diffusion or slow clearance, thus matching the rate of change within the segment to the low-flow-rate scenario. The slope between adjacent sampling points is calculated point-by-point along the time axis and compared with a preset slope constraint range. When the slope between adjacent sampling points exceeds the allowable rise range or falls below the allowable fall range, the signal value of the current sampling point is amplitude-limited to bring the rate of change between adjacent sampling points back to the preset slope constraint range. The slope can be calculated by dividing the signal difference between two adjacent points by the sampling interval. The preset slope constraint range can be set based on the tracer diffusion rate, sampling period, and the rate of change of normal low-flow-rate segments in the calibration sample under low-flow-rate conditions. If a point-by-point correction method is used, when an excessive slope is detected, the signal value of the previous sampling point can be used as a reference to recalculate the signal value of the current sampling point according to the maximum allowable rise slope or maximum fall slope.

[0043] Signal intensity is used to characterize whether the overall effective signal level of the fragment is in a weak signal state, distinguishing between the normal slow-changing situation in low-flow-rate fragments and the weak response situation with low amplitude, significant delay, and susceptibility to noise masking. The fragment mean, fragment median, local integral average, or fragment envelope mean can be used as the signal intensity characterization quantity and compared with a preset intensity threshold. When the signal intensity is lower than the preset intensity threshold, it indicates that the effective fluorescence response in the fragment is weak and easily affected by response lag and transient perturbations, requiring further delay sampling compensation and transient intensity mutation suppression. When the signal intensity is not lower than the preset intensity threshold, the initially adjusted fragment can directly proceed to subsequent threshold follow-up adjustments. The preset intensity threshold can be set based on the baseline noise level, the effective detection limit of the tracer, or the weak response boundary in the calibration sample.

[0044] Delayed sampling compensation does not add new sampling points. Instead, it corrects local distortion caused by delayed initial response within the time index range corresponding to the existing low-velocity segment. Specifically, the initial effective response point is first determined within the current low-velocity segment. Based on the connection relationship between this initial effective response point and the local trend at the end of the previous segment, the number of delayed sampling points is determined. The number of delayed sampling points can be obtained by dividing the time offset of the initial effective response point relative to the expected response position by the sampling period and rounding down. Subsequently, the initial effective response point and the subsequent preset length signal segment are mapped forward to an earlier time index position within the segment according to the number of delayed sampling points. For the gaps formed at the end after mapping, local interpolation between adjacent effective sampling points is used to fill in the gaps. After compensation, the total number of time indices for the low-velocity segments remains unchanged, the segment boundaries remain unchanged, and only the response pattern of the initial segment within the segment is corrected to reduce delayed distortion caused by tracer arrival lag, local exchange delay, or detection response offset.

[0045] After completing the delayed sampling compensation, instantaneous intensity mutation suppression is performed on the corresponding preliminary adjustment segments to remove occasional single-point increases, single-point decreases, or local short-term jumps in weak signal segments, preventing such non-continuous changes from being misidentified as effective concentration changes in the subsequent threshold following mechanism. The amplitude difference between the current sampling point and its adjacent sampling points is calculated point by point. If the amplitude difference exceeds a preset mutation threshold, and the adjacent sampling points do not support continuous unidirectional changes, the sampling point or the sampling point within a local short window is identified as an instantaneous mutation point, and suppression is achieved using neighborhood interpolation, neighborhood mean replacement, or local trend substitution. This step targets low-intensity segments, and the preset mutation threshold should be set in conjunction with the local background level of the segment to avoid normal small changes in weak signal segments being misjudged as abnormal mutations.

[0046] The threshold following mechanism is used to dynamically adjust the response judgment benchmark within low-flow-rate segments. Specifically, an analysis window is set by sliding along the time axis of the low-flow-rate segment with a preset step size. Local low quantile values ​​are extracted within each analysis window, and a preset offset is superimposed on these values ​​to obtain the local following threshold corresponding to that analysis window. The local following thresholds of adjacent analysis windows are then smoothly connected to form a continuously varying following threshold curve along the time axis. The preset offset is determined based on the local noise amplitude within the low-flow-rate segment, and the preset step size is smaller than the analysis window length to ensure the following threshold curve changes slowly. Subsequently, the signal value of each sampling point is compared with the following threshold at the corresponding time. Sampling points higher than the following threshold and meeting the continuity condition are retained; sampling points close to the following threshold and consistent with the local change trend are retained according to the smoothing trend; and sampling points deviating from the following threshold and not meeting the continuity condition are reverted according to the local trend. After threshold following adjustment, spurious fluctuations caused by instantaneous noise can be suppressed, and the continuity of the effective slow-varying response within the low-flow-rate segment can be maintained.

[0047] After all low-flow-rate segments have undergone the above processing, each segment is backfilled into its corresponding time index according to its original time position in the fluorescence signal sequence to generate a low-flow-rate integral sequence. For sampling time points belonging to the low-flow-rate segment, the corresponding adjusted fluorescence signal value is written; for sampling time points not belonging to the low-flow-rate segment, no value is written or a placeholder status is retained.

[0048] Preferably, for low-velocity segments of length one, since they lack independent intra-segment envelope construction conditions, slope calculation conditions, and segment-level intensity statistics conditions, it is not advisable to directly perform envelope reconstruction adjustment based on continuous changes within the segment, slope constraint processing of adjacent sampling points, segment-level signal intensity judgment, and delayed sampling compensation as with conventional low-velocity segments. The sampling points corresponding to this single-point segment can be mapped back to the original fluorescence signal sequence, and an extended neighborhood can be formed by combining the adjacent effective sampling points before and after it for single-point low-velocity verification. When the single point is classified as low-velocity in terms of velocity, and its fluorescence value is consistent with the change direction, local background level, and weak response trend of the adjacent effective sampling points before and after it, the single point is retained, and its single-point threshold is adjusted based on the local background reference value formed by the adjacent effective sampling points before and after it. When the trend of the single point is significantly inconsistent with the neighborhood, or its amplitude deviates too much from the neighborhood background, it is identified as an abnormal single point, and the nearest effective sampling point before and after it is used for neighborhood interpolation replacement, or it is removed in the subsequent integration stage when there is insufficient neighborhood support.

[0049] S4. Based on the regional type division results, integrate the high-velocity compensation sequence and the low-velocity integral sequence to obtain the overall compensation sequence.

[0050] In one specific embodiment, the process of performing step S4 may specifically include the following steps: For high-velocity compensation sequences and low-velocity integral sequences, time axis alignment is performed through synchronous interpolation; An amplitude scale normalization unified mapping mechanism is adopted to adjust the amplitude range of the time-axis aligned high-velocity compensation sequence and low-velocity integral sequence. Based on the regional classification results, the high velocity compensation sequence and low velocity integral sequence after amplitude range adjustment are spliced ​​together using multi-segment sequence splicing technology to obtain the spliced ​​sequence; Smooth the transition segments of the region boundaries in the spliced ​​sequence to obtain the initial overall sequence; The integrity of the initial overall sequence is checked. If the signal continuity meets the requirements, the initial overall sequence is determined as the overall compensation sequence.

[0051] Specifically, the time axis alignment process for synchronous interpolation of the high-velocity compensation sequence and the low-velocity integral sequence does not redefine the global sampling time of the two sequences. Instead, it maps the effective sampling values ​​of the two sequences within their respective effective segments, which are generated by local interpolation correction, single-point substitution or boundary backfilling, back to the time index position corresponding to the target time axis.

[0052] Specifically, the target time axis can be determined by the original fluorescence sampling time axis used in S1 or the time index sequence corresponding to the region type division result in S2. Then, synchronous interpolation mapping is performed on the sampling positions in the high-flow-rate compensated sequence and the low-flow-rate integral sequence where valid values ​​have already been written. For positions where there are corresponding sampling points on the target time axis for the two sequences, the original values ​​are kept unchanged. For missing points, boundary interpolation points, or single-point substitution points formed after local processing, linear interpolation, piecewise spline interpolation, or local preservation interpolation are used to map adjacent valid sampling points to the target time axis. In cases where there are valid values ​​only in one sequence, while the other sequence is empty or vacant at the corresponding position, mapping is only performed on the sequence with valid values, and values ​​are not forcibly generated for empty positions.

[0053] Since high-velocity compensated sequences primarily undergo peak suppression, baseline correction, and denoising, while low-velocity integral sequences primarily undergo envelope reconstruction, slope constraint, and threshold following adjustment, although both are based on fluorescence signals, their amplitude distributions, local dynamic ranges, and background reference levels may differ. Direct splicing can easily lead to discontinuous amplitude steps when switching between high-velocity and low-velocity segments. To address this, the reference amplitude ranges of the two sequences within their respective effective segments can be calculated separately. These ranges can be determined by local mean, local median, local upper and lower quantile differences, or effective minimum and maximum values ​​within a segment. Then, the two sequences can be mapped to a unified target amplitude scale. For example, based on local reference segments near the boundary, the local scale ratio and local offset between the high-velocity compensated sequence and the low-velocity integral sequence can be calculated. The amplitude can then be uniformly mapped to one or both sequences simultaneously according to this local scale ratio and local offset, ensuring that the local background level and effective variation range on both sides of the boundary remain consistent. If there are not enough local reference segments near the boundary, the statistical range of each effective segment can be used for normalization and then mapped to a unified amplitude range, but only if the relative change relationship within each segment is not changed.

[0054] Using the region type classification results as the basis for splicing control, the sequence is scanned point by point along a unified time axis. When the segment type corresponding to the current sampling time point is high-velocity, the effective value of the high-velocity compensation sequence at that time point is selected and written into the splicing sequence. When the segment type corresponding to the current sampling time point is low-velocity, the effective value of the low-velocity integral sequence at that time point is selected and written into the splicing sequence. When the current sampling time point belongs to a regular segment, an unclassified segment, or a segment with undetermined boundaries in the region type classification results, the extended value of the adjacent effective segment, the interpolated value of the local trend on both sides, or the signal value corresponding to the basic correction in the original fluorescence signal sequence is used as the supplementary written value. The splicing in this step is not a connection of two complete sequences end to end, but rather the selection of the source sequence point by point on the unified time axis according to the region type classification results to form a single continuous output sequence.

[0055] The regional boundary transition section refers to the boundary area between high-velocity and low-velocity sections, or the boundary area between effective and conventional sections. Since the two types of sequences are generated using different processing mechanisms, even after time axis alignment and amplitude range adjustment, local slope discontinuities, amplitude jumps, or phase abrupt changes may still exist at the boundary. Therefore, a transition window of preset length is set on both sides of each regional boundary, and smooth connection is only performed within the transition window without changing the content of the segments already spliced ​​outside the window. A weighted mixing method can be used to transition the sequences on both sides of the boundary, that is, gradually decreasing the weight of the previous source sequence on one side of the boundary while gradually increasing the weight of the next source sequence, so that the output value within the transition window smoothly transitions from the previous sequence to the next sequence. For example, within the transition window, sampling points closer to the previous segment can have a larger weight for the previous source sequence and a smaller weight for the next source sequence; when the weights are close at the center of the boundary, the weight distribution is gradually reversed on the side closer to the next segment. Alternatively, local curve fitting, local spline bridging, or a smooth connection based on boundary endpoint constraints can be used to maintain the continuity of the first-order variation trend on both sides of the boundary. If there is only a single valid value on one side of the boundary, then that single point is used as the endpoint constraint, and the shortest transition connection segment is constructed by combining it with multiple adjacent valid points on the other side.

[0056] Integrity checks are used to confirm that the S4 output sequence meets the requirements of subsequent transparency analysis in terms of time coverage, amplitude continuity, and boundary connectivity. It includes at least the following: 1. Time coverage check: Checking for any unacceptable gaps in the initial overall sequence on the target time axis; if some positions should have valid output in the region type division results but are still missing, they are considered incomplete and filled in by local interpolation if adjacent valid points exist. 2. Boundary continuity check: Checking for amplitude jumps or slope abrupt changes exceeding a preset continuity threshold before and after the transition segments of each region boundary; if so, returning to the boundary smoothing step to readjust the transition window length or weight distribution. 3. Segment consistency check: Checking whether high-velocity segment positions are mistakenly written into low-velocity integral results, or whether low-velocity segment positions are mistakenly written into high-velocity compensation results, to prevent segment source mismatch during splicing. 4. Physical rationality verification: Check the overall compensated sequence for any abnormal isolated peaks, abnormal collapses, or discontinuous reverse abrupt changes that clearly violate the local transport and exchange processes of the fluorescent tracer. If local anomalies are found, local interpolation, local smoothing, or local amplitude mapping can be re-executed within the corresponding boundary window or missing window, without having to roll back the entire processing flow. When all verifications meet the preset requirements, the initial overall sequence is determined as the overall compensated sequence.

[0057] S5. Based on the temporal correspondence between the overall compensation sequence and the velocity data sequence, separate the flow contribution and obtain the pure concentration change sequence.

[0058] In one specific embodiment, the process of performing step S5 may specifically include the following steps: By using phase delay compensation, time shift correction processing is performed on the overall compensation sequence and the flow velocity data sequence to establish the time sequence correspondence between the two. A trend consistency directional constraint verification mechanism is adopted to perform correlation analysis on the overall compensation sequence and the velocity data sequence after time shift correction to determine the direction and degree of influence of velocity changes on the overall compensation sequence changes. Based on the direction and degree of influence, a flow contribution separation process is performed to separate the signal contribution components caused by flow velocity changes in the overall compensation sequence after time shift correction processing. Based on the separation results, the signal contributing components are removed to generate a pure concentration change sequence.

[0059] Specifically, since local blood flow velocity and lymph flow velocity jointly affect the tracer transport process within the detection area, and although the overall compensated fluorescence signal sequence (i.e., the overall compensated sequence) has reduced segmentation errors and local disturbances after the aforementioned differential compensation and integration processing, it may still contain residual signal components caused by flow velocity changes. Therefore, the flow velocity data sequence is first converted into a flow velocity characterization sequence corresponding to the overall compensated fluorescence signal sequence, and then time-shift correction and flow contribution separation are performed in conjunction with the overall compensated fluorescence signal sequence. Specifically, flow velocity characterization values ​​can be constructed according to preset rules based on local blood flow velocity values ​​and lymph flow velocity values. The preset rules can be weighted summation, normalized weighted combination, or priority selection rules based on segment type. Preferably, the flow velocity characterization sequence is constructed using a normalized weighted combination based on segment type variations, with high-flow-velocity segments dominated by local blood flow velocity values ​​and low-flow-velocity segments dominated by lymph flow velocity values, thus obtaining a flow velocity characterization sequence arranged along a unified time axis. Subsequently, the overall compensated fluorescence signal sequence and the flow velocity characterization sequence are mapped to a unified sampling time axis, and one of the sequences is gradually shifted within a preset time shift search range. Under each candidate time shift condition, the correlation between the two is calculated, with cross-correlation value being preferred. The candidate time shift corresponding to the maximum cross-correlation value is determined as the phase delay compensation amount to establish the temporal correspondence between flow velocity changes and overall compensated fluorescence signal changes. Then, the flow velocity characterization sequence or overall compensated fluorescence signal sequence is time-shift corrected according to the phase delay compensation amount, resulting in a time-shift corrected overall compensated fluorescence signal sequence and a time-shift corrected flow velocity characterization sequence, which further separates the signal contribution component caused by flow velocity changes in the overall compensated fluorescence signal sequence. The preset time shift search range is limited to the physiological response delay range allowed by the detection system to avoid mismatches caused by unreasonable time shifts.

[0060] After time-shift correction is completed, a sliding analysis window is set along a unified time axis. Within each analysis window, the local variation trends of the time-shift corrected overall compensated fluorescence signal sequence and the time-shift corrected flow velocity characterization sequence are calculated. Preferably, the slope of the linear fitting within the window is used as the characterization quantity of the local variation trend. Within the same analysis window, if the local variation trends of the overall compensated fluorescence signal sequence and the flow velocity characterization sequence have the same sign, then a positive flow contribution is determined to exist within that analysis window; if their local variation trends have opposite signs, then a reverse flow contribution is determined to exist within that analysis window. To avoid misjudgment due to isolated noise points, the analysis window is only confirmed as a valid contribution interval when the number of sampling points satisfying the same or opposite relationship within the window reaches a preset proportion, and the corresponding change amplitude exceeds the local noise threshold. Furthermore, based on the proportional relationship between the local change in the overall compensated fluorescence signal and the local change in the flow velocity characterization sequence within the analysis window, the influence of flow velocity changes on fluorescence signal changes within the analysis window is determined and recorded as the window contribution coefficient. The local change in the overall compensated fluorescence signal is preferably represented by the difference in overall compensated values ​​between the first and last sampling points within the analysis window, and the local change in the flow velocity characterization sequence is preferably represented by the difference in flow velocity characterization values ​​between the first and last sampling points within the analysis window. When the absolute value of the local change in the flow velocity characterization sequence is less than a preset lower limit, the window contribution coefficient for that analysis window is set to zero to avoid abnormal amplification of the window contribution coefficient due to an excessively small denominator. When the absolute value of the local change in the flow velocity characterization sequence is not less than the preset lower limit, the ratio of the local change in the overall compensated fluorescence signal to the local change in the flow velocity characterization sequence is used as the window contribution coefficient.

[0061] After determining the influence direction and window contribution coefficient of each analysis window, the flow contribution sequence is further calculated. Specifically, the flow velocity change corresponding to each sampling point within each analysis window is first calculated. The flow velocity change is preferably represented by the difference between the flow velocity characterization value of the current sampling point and the flow velocity characterization value of the previous sampling point. Then, the window contribution coefficient of the corresponding analysis window is assigned a sign according to the influence direction: positive for forward flow contribution and negative for reverse flow contribution. Subsequently, the signed window contribution coefficient is multiplied point by point by the flow velocity change corresponding to each sampling point within the analysis window to obtain the flow contribution value of each sampling point. The flow contribution values ​​obtained in all analysis windows are arranged in a uniform time order to form a flow contribution sequence consistent with the time axis of the overall compensated fluorescence signal sequence. For analysis windows that do not meet the conditions of directional consistency, sampling point ratio, local amplitude, or lower limit of denominator, their window contribution coefficients are set to zero, and the corresponding flow contribution values ​​are all set to zero to avoid misidentifying the true concentration change as a flow contribution.

[0062] The time-shift corrected overall compensated fluorescence signal sequence is used as the object to be corrected. The corresponding flow contribution value is subtracted point-by-point according to a unified time index to obtain the pure concentration change value at each sampling point, thus forming a pure concentration change sequence. For sampling points with positive flow contribution, the corresponding positive additional component is subtracted; for sampling points with negative flow contribution, the subtraction is performed according to their sign to retain the signal components caused by the intrinsic concentration change of the tracer and remove the additional effects introduced by the flow rate change. If, after subtraction, an isolated anomalous spiking point or anomalous negative jump that clearly violates the gradual concentration change characteristic appears within a certain local window, the window contribution coefficient of that local window is corrected. Specifically, the window contribution coefficient of the local window is first reduced by a preset ratio, and the flow contribution value and subtraction result of each sampling point within the local window are recalculated accordingly. If the anomalous spiking point or anomalous negative jump still exists after recalculation, the window contribution coefficient is further reduced by the preset ratio, and the above process is repeated. When the subtraction result after reduction satisfies the gradual concentration change characteristic, or the window contribution coefficient decreases to a preset lower limit, the correction is terminated. If the anomaly still exists after the window contribution coefficient is reduced to the preset lower limit, the window contribution coefficient of the local window is set to zero, and the flow contribution value calculation and deduction process in the local window is re-executed with zero contribution, thereby obtaining the pure concentration change sequence.

[0063] Figure 2 This figure illustrates the flow contribution separation results. As shown, this embodiment displays the overall compensation sequence, flow velocity characterization sequence, flow contribution sequence, and pure concentration change sequence during a monitoring period of 0 to 10 seconds. The 2-4 second interval can be considered a region with strong flow contribution; the overall compensation sequence and flow velocity characterization sequence show relatively consistent trends, and the separated flow contribution sequence changes with the flow velocity characterization sequence. After deducting the flow contribution, the pure concentration change sequence maintains a relatively smooth upward trend. In contrast, in subsequent monitoring intervals, the changes in the overall compensation sequence mainly reflect concentration changes, while the flow contribution weakens relatively. This figure illustrates the flow contribution separation process and results proposed in this invention, demonstrating that the signal component caused by flow velocity changes can be separated and subtracted from the overall compensation sequence, thereby obtaining the pure concentration change sequence for subsequent permeability calculations.

[0064] S6. Calculate the blood labyrinth barrier permeability value based on the pure concentration change sequence and output the permeability monitoring results.

[0065] In one specific embodiment, the process of performing step S6 may specifically include the following steps: Calculate the rate of change of signal intensity based on pure concentration change sequences; By integrating the cumulative deviations, the long-term trend of the pure concentration change sequence was determined. The permeability value of the blood labyrinth barrier is calculated based on the rate of change of signal intensity and the long-term trend. The permeability value of the blood labyrinth barrier is calibrated within a certain range. When the permeability value of the blood labyrinth barrier meets the preset reasonable range, the permeability monitoring results are output.

[0066] Specifically, the pure concentration change sequence is a time series after deducting the signal contributions introduced by local blood flow and lymphatic flow changes, representing the intrinsic concentration change of the fluorescent tracer within the detection area corresponding to each sampling point. Since the blood labyrinth barrier permeability during continuous monitoring is simultaneously reflected in the rate of change of tracer crossing the barrier into the detection area per unit time, and the cumulative shift caused by the imbalance between net tracer penetration and net clearance within the continuous monitoring window, the signal intensity change rate and long-term trend are extracted from the pure concentration change sequence, and the permeability value is calculated based on both.

[0067] When calculating the rate of change of signal intensity based on the pure concentration change sequence, the pure concentration values ​​of two adjacent valid sampling points are read sequentially according to a unified sampling time axis, and the concentration difference between the two points is calculated. This concentration difference is then divided by the corresponding time interval to obtain the instantaneous rate of change within the current sampling interval. This process is repeated for continuous sampling points throughout the monitoring window to form a rate of change sequence. If the sampling time axis is an equally spaced time axis, the instantaneous rate of change within each sampling interval is obtained by dividing the difference between adjacent sampling points by a fixed sampling period. If the sampling time axis has unequal intervals, the rate of change is calculated segment by segment according to the corresponding actual time interval to avoid including sampling jitter in the rate result. For sampling points at the beginning of the monitoring window that cannot form a difference between the preceding and following points, they can be excluded from the rate representative value calculation, or the result of the next valid difference can be used to complete the calculation. Considering that residual high-frequency disturbances may still exist during volume detection, the obtained rate of change sequence can be processed by moving average, median smoothing, or amplitude limiting without changing the time index to suppress isolated differential spikes. Preferably, the representative rate is represented by the mean of the smoothed rate sequence within the current monitoring window; when it is necessary to enhance the sensitivity to the continuous rising process, the mean of the positive rate can be used as a supplementary representation to reflect the change level of the tracer crossing the blood labyrinth barrier into the detection area within the current time period.

[0068] To determine the long-term trend of the pure concentration variation sequence, a reference concentration level is first determined within the current monitoring window. When the detection process has a stable baseline period, the reference concentration level is preferably the average value of the baseline period before the tracer enters the system; when a stable baseline period is lacking or unreliable, the reference concentration level is the average value of the pure concentration variation sequence within the current monitoring window. The pure concentration value at each sampling point within the current monitoring window is subtracted from the reference concentration level to obtain the concentration deviation at each sampling point; then, the concentration deviations are accumulated in chronological order and combined with the time interval between adjacent sampling points to form a cumulative deviation integral sequence. If the pure concentration value within the current monitoring window is consistently higher than the reference concentration level, the cumulative deviation integral sequence continuously increases; if the pure concentration value within the current monitoring window is consistently lower than the reference concentration level, the cumulative deviation integral sequence continuously decreases. Based on the real-time monitoring scenario, the cumulative deviation integral is preferably executed recursively within the sliding monitoring window. This means that while receiving new sampling points, the earliest sampling point in the current window is removed, and the concentration deviation integral result is recalculated or recursively updated for the updated window. This ensures that the long-term trend reflects the net penetration status within the current period, avoiding excessive tailing of the current permeability status caused by accumulating data across all time periods. Preferably, the long-term trend is represented by the final value of the cumulative deviation integral corresponding to the current monitoring window. When it is necessary to characterize the overall offset within the window, the integral mean or integral slope can be used as supplementary characterization to reflect the changes in the tracer's continuous penetration, retention, or slow clearance within the current window.

[0069] After obtaining the representative rate and long-term trend, the blood labyrinth barrier permeability value is calculated. The representative rate is used as the short-term penetration capacity component, and the long-term trend is used as the continuous penetration accumulation component, and they are combined after unifying the dimensions. Dimension unification can be achieved through equipment calibration coefficients, tracer response conversion coefficients, or window duration normalization coefficients, so that the short-term penetration capacity component and the continuous penetration accumulation component can be calculated on the same permeability characterization scale. Preferably, the permeability value is calculated by a linear combination of the short-term penetration capacity component and the continuous penetration accumulation component, that is, the representative rate is multiplied by a first conversion coefficient, the long-term trend is multiplied by a second conversion coefficient, and the two parts are added together to obtain the permeability value corresponding to the current monitoring window; the first and second conversion coefficients are established based on equipment calibration results, tracer dosage conditions, detection depth, and normal control samples. For application scenarios that focus more on the acute barrier opening process, the proportion of the short-term penetration capacity component in the combined calculation can be increased; for application scenarios that focus more on slow leakage, continuous drug entry, or recovery processes, the proportion of the continuous penetration accumulation component in the combined calculation can be increased. If the pure concentration change sequence within the current monitoring window is generally stable and the rate of change is close to zero, while the long-term trend is also close to zero, the calculated permeability value will remain at a low level; if the pure concentration change sequence within the current monitoring window continues to rise, and the rate of change and the long-term trend increase synchronously, the calculated permeability value will increase synchronously.

[0070] During range verification, the currently calculated permeability value is compared with the lower and upper limits of a preset reasonable range. The preset reasonable range is established based on equipment calibration experiments, statistical intervals of normal control samples, tracer dosage conditions, and probe detection depth. If the permeability value is lower than the lower limit, the permeability value is output at the lower limit, and a low-value anomaly marker is recorded; if the permeability value is higher than the upper limit, the permeability value is output at the upper limit, and a high-value anomaly marker is recorded. Preferably, when multiple consecutive monitoring windows show anomaly markers exceeding the preset reasonable range, the corresponding monitoring window is recalculated or reviewed. For permeability values ​​within the preset reasonable range, they are directly output as the permeability monitoring result corresponding to the current sampling time or the current monitoring window, and are associated and stored with the current timestamp, the current pure concentration change sequence segment, and the current flow rate status label. In continuous real-time monitoring mode, a new permeability monitoring result is output each time a sliding monitoring window is updated, thus forming a permeability monitoring curve that changes continuously over time.

[0071] Figure 3This diagram illustrates the real-time monitoring results of blood-labyrinth barrier permeability. As shown, this embodiment displays the pure concentration change sequence, signal intensity change rate sequence, cumulative deviation integral sequence, and the finally calculated permeability value curve within a monitoring period of 0 to 10 seconds. The 4-6 second interval represents an example sliding monitoring window. Within this window, the pure concentration change sequence shows a continuous upward trend, the signal intensity change rate remains positive, the cumulative deviation integral continuously increases, and the permeability value calculated through linear combination increases synchronously, reflecting the enhanced permeability of the blood-labyrinth barrier. When the monitoring window slides to a relatively flat period, the pure concentration change tends to level off, the signal intensity change rate decreases or approaches zero, the cumulative deviation integral continues to change, and the permeability value is mainly affected by the long-term trend component. This figure illustrates the process of extracting the short-term penetration capacity component and the continuous cumulative penetration component from the pure concentration change sequence and calculating the permeability monitoring results, demonstrating that the proposed method can achieve real-time and continuous output of permeability values.

[0072] The above describes the method for real-time monitoring of blood labyrinth barrier permeability in the embodiments of this application. The following describes the system for real-time monitoring of blood labyrinth barrier permeability in the embodiments of this application. Please refer to [link to relevant documentation]. Figure 4 The present application provides a schematic diagram of the structure of a real-time monitoring system for blood labyrinth barrier permeability, which includes: The data acquisition module 10 is used to acquire the raw signal data and flow rate measurement data of the fluorescent tracer in the blood labyrinth barrier region, and to acquire the fluorescence signal sequence and flow rate data sequence after preprocessing.

[0073] The region segmentation module 20 is used to divide the sampling segments corresponding to the fluorescence signal sequence into region types based on the flow velocity data sequence, and to determine the high flow velocity segment and the low flow velocity segment.

[0074] The dual-flow compensation module 30 is used to extract the corresponding fluorescence signal sequences for the high-flow-velocity section and the low-flow-velocity section respectively and perform differential processing to obtain the high-flow-velocity compensation sequence and the low-flow-velocity integral sequence.

[0075] The signal integration module 40 is used to integrate the high-velocity compensation sequence and the low-velocity integral sequence according to the region type division results to obtain the overall compensation sequence.

[0076] The flow separation module 50 is used to separate the flow contribution based on the time-series correspondence between the overall compensation sequence and the flow velocity data sequence, and to obtain the pure concentration change sequence.

[0077] The permeability calculation module 60 is used to calculate the blood labyrinth barrier permeability value based on the pure concentration change sequence and output the permeability monitoring results.

[0078] This application also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform the steps of the real-time monitoring method for blood labyrinth barrier permeability.

[0079] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for real-time monitoring of blood labyrinth barrier permeability, characterized in that, The method includes: S1. Obtain the raw signal data and flow rate measurement data of the fluorescent tracer in the blood labyrinth barrier region, and obtain the fluorescence signal sequence and flow rate data sequence after preprocessing; S2. Based on the flow velocity data sequence, the sampling segments corresponding to the fluorescence signal sequence are divided into regions to determine high flow velocity segments and low flow velocity segments; S3. For the high-velocity section and the low-velocity section, extract the corresponding fluorescence signal sequences and perform differential processing to obtain the high-velocity compensation sequence and the low-velocity integral sequence. S4. Based on the region type division results, integrate the high-velocity compensation sequence and the low-velocity integral sequence to obtain the overall compensation sequence; S5. Separate the flow contribution based on the temporal correspondence between the overall compensation sequence and the flow velocity data sequence to obtain the pure concentration change sequence; S6. Calculate the blood labyrinth barrier permeability value based on the pure concentration change sequence and output the permeability monitoring results.

2. The method according to claim 1, characterized in that, S1 includes: The sensor probe simultaneously acquires raw signal data of the fluorescent tracer, local blood flow velocity measurement data, and lymph flow velocity measurement data. The original signal data were arranged into an initial fluorescence signal sequence according to the sampling time order; The local blood flow velocity measurement data and the lymph flow velocity measurement data are integrated into an initial flow velocity data sequence according to the sampling time sequence; The initial fluorescence signal sequence and the initial flow velocity data sequence are subjected to time synchronization preprocessing to obtain signal and flow velocity data pairs; The integrity of the signal and flow velocity data pair is checked. If the check result meets the preset standard, the fluorescence signal sequence and the flow velocity data sequence are output.

3. The method according to claim 1, characterized in that, S2 include: For the flow velocity data sequence, extract the local blood flow velocity data and lymph flow velocity data corresponding to each sampling time point; The local blood flow velocity data is compared with a preset blood flow velocity threshold. The continuous sampling time points where the local blood flow velocity data is greater than the preset blood flow velocity threshold are determined as high flow velocity time windows. The sequence segments in the fluorescence signal sequence corresponding to the high flow velocity time windows are determined as high flow velocity segments. For sampling time points that are not identified as high-flow-rate segments, the lymph flow velocity data is compared with a preset lymph flow velocity threshold. Continuous sampling time points where the lymph flow velocity data is less than the preset lymph flow velocity threshold are identified as low-flow-rate time windows. Sequence segments in the fluorescence signal sequence corresponding to the low-flow-rate time windows are identified as low-flow-rate segments. The judgment results are summarized to generate a region type classification result aligned with the time axis of the fluorescence signal sequence.

4. The method according to claim 1, characterized in that, In S3, for the high-flow-rate section, the fluorescence signal sequence is processed to obtain the high-flow-rate compensation sequence, including: Sequence fragments corresponding to each high-flow-rate segment are extracted from the fluorescence signal sequence; For each extracted sequence segment, signal peak sharpness suppression processing and baseline drift dynamic tracking correction are performed sequentially to obtain the corresponding preliminary corrected segment; The fluctuation amplitude of each preliminary correction segment is sequentially determined to exceed the preset amplitude threshold. If so, local smoothing interpolation processing is performed on the corresponding preliminary correction segment for amplitude anomaly points. A high-frequency noise segmentation filtering mechanism is adopted to perform noise reduction processing on the preliminary correction segments that have undergone smooth interpolation and the preliminary correction segments that do not exceed the preset amplitude threshold. The high flow rate compensation sequence is generated based on the processed segments.

5. The method according to claim 1, characterized in that, In S3, for the low-flow-rate section, the fluorescence signal sequence is processed to obtain the low-flow-rate integral sequence, including: Sequence fragments corresponding to each low-flow-rate segment are extracted from the fluorescence signal sequence; For each extracted sequence segment, signal envelope reconstruction adjustment and adjacent sampling point slope constraint processing are performed sequentially to obtain the corresponding preliminary adjusted segment; Sequentially determine whether the signal strength of each preliminary adjustment segment is lower than the preset strength threshold. If so, perform delayed sampling compensation and instantaneous strength change suppression on the corresponding preliminary adjustment segment. A threshold following mechanism is adopted to perform threshold following adjustment on the preliminary adjustment segments that have undergone delay sampling compensation and instantaneous intensity mutation suppression, as well as the preliminary adjustment segments that are not lower than the preset intensity threshold, and generate the low flow rate integral sequence based on the adjusted segments.

6. The method according to claim 1, characterized in that, S4 include: The high-velocity compensation sequence and the low-velocity integral sequence are time-axis aligned using synchronous interpolation. An amplitude scale normalization unified mapping mechanism is adopted to adjust the amplitude range of the time-axis aligned high-velocity compensation sequence and low-velocity integral sequence. Based on the region type division results, the high velocity compensation sequence and low velocity integral sequence after amplitude range adjustment are spliced ​​together using multi-segment sequence splicing technology to obtain a spliced ​​sequence. The transition segments of the region boundaries in the spliced ​​sequence are smoothly connected to obtain the initial overall sequence; The initial overall sequence is subjected to integrity verification. If the signal continuity meets the requirements, the initial overall sequence is determined as the overall compensation sequence.

7. The method according to claim 1, characterized in that, S5 include: By using phase delay compensation, time shift correction processing is performed on the overall compensation sequence and the flow velocity data sequence to establish a time sequence correspondence between the two. A trend consistency directional constraint verification mechanism is adopted to perform correlation analysis on the overall compensation sequence and the velocity data sequence after time shift correction to determine the direction and degree of influence of velocity changes on the overall compensation sequence changes. Based on the direction and degree of influence, a flow contribution separation process is performed to separate the signal contribution components caused by flow velocity changes in the overall compensation sequence after time-shift correction processing. Based on the separation results, the signal contributing components are removed to generate the pure concentration change sequence.

8. The method according to claim 1, characterized in that, S6 include: The rate of change of signal intensity is calculated based on the pure concentration change sequence. The long-term trend of the pure concentration change sequence is determined by cumulative deviation integration. The blood labyrinth barrier permeability value is calculated based on the rate of change of the signal intensity and the long-term trend. The permeability value of the blood labyrinth barrier is checked for range, and when the permeability value of the blood labyrinth barrier meets the preset reasonable range, the permeability monitoring result is output.

9. A real-time monitoring system for blood labyrinth barrier permeability, used to implement the method as described in any one of claims 1 to 8, characterized in that, The system includes: The data acquisition module is used to acquire the raw signal data and flow rate measurement data of the fluorescent tracer in the blood labyrinth barrier region, and to acquire the fluorescence signal sequence and flow rate data sequence after preprocessing. The region segmentation module is used to segment the sampling segments corresponding to the fluorescence signal sequence into regions based on the flow velocity data sequence, and to determine high flow velocity segments and low flow velocity segments. The dual-flow compensation module is used to extract the corresponding fluorescence signal sequences for the high-flow-rate section and the low-flow-rate section respectively and perform differential processing to obtain the high-flow-rate compensation sequence and the low-flow-rate integral sequence. The signal integration module is used to integrate the high-velocity compensation sequence and the low-velocity integral sequence according to the region type division results to obtain the overall compensation sequence; The flow separation module is used to separate the flow contribution based on the temporal correspondence between the overall compensation sequence and the flow velocity data sequence, and to obtain the pure concentration change sequence. The permeability calculation module is used to calculate the blood labyrinth barrier permeability value based on the pure concentration change sequence and output the permeability monitoring results.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When the instruction is executed by the processor, it implements the method for real-time monitoring of blood labyrinth barrier permeability as described in any one of claims 1 to 8.

Images