Multi-channel contrast analysis method for drainage liquid property image and system thereof

Abstract

This invention belongs to the field of medical monitoring image processing technology, and discloses a method and system for comparative analysis of multi-channel drainage fluid characteristics images. The method includes: a multi-channel synchronous acquisition step, in which multiple drainage fluid acquisition terminals are controlled by a central processing unit to synchronously acquire image data and perform time alignment; a color feature extraction step, in which image color features are mapped to a standardized chromaticity coordinate system; a multi-channel comparative analysis step, in which chromaticity differences are calculated and color change trends are compared; a dual-mode anomaly detection step, in which single-channel longitudinal change detection and multi-channel lateral correlation detection are performed; and a report generation and output step, in which a comprehensive monitoring report is generated. This invention solves the problems of existing technologies that cannot simultaneously monitor drainage fluid from multiple tubes and lack cross-channel correlation analysis capabilities.

Description

Technical Field

[0001] This invention belongs to the field of medical monitoring image processing technology, and particularly relates to a method and system for comparative analysis of multi-channel drainage fluid characteristics images. Background Technology

[0002] Patients with multiple drainage tubes in the intensive care unit are a key focus of clinical monitoring. These patients typically have multiple drainage tubes in place simultaneously, including chest tubes, abdominal tubes, gastric tubes, and urinary catheters. Changes in the characteristics of the drainage fluids from each tube can reflect the pathological state of different parts of the patient's body. In clinical practice, nursing staff need to regularly observe and record the color, transparency, and other characteristics of each drainage fluid. This information is of significant clinical value for the early detection of complications such as postoperative bleeding, infection, and anastomotic leakage.

[0003] Chinese patent CN111612742A discloses a method and system for detecting mold growth in traditional Chinese medicine. This technical solution acquires raw image data of the traditional Chinese medicine to be tested using a high-definition camera, utilizes a convolutional neural network to obtain image feature and color information, and combines multiple detection methods such as temperature sensors, humidity sensors, odor scanners, and microscopes to achieve comprehensive detection and grade identification of mold growth in traditional Chinese medicine. This technical solution improves detection accuracy by employing image processing and multi-source information fusion, and provides timely alerts for abnormal conditions through an early warning module.

[0004] However, the aforementioned existing technologies have the following problems and shortcomings: First, the technical solution is designed for a single detection object and does not support the synchronous acquisition and time alignment processing of multi-channel data, thus failing to meet the clinical needs of patients with multiple drainage tubes in the intensive care unit who require simultaneous monitoring of multiple drainage channels; Second, the technical solution lacks cross-channel correlation analysis capabilities and cannot detect the temporal correlation between changes in drainage fluids from different sites, which often has important implications in clinical diagnosis. For example, the simultaneous appearance of bloody changes in abdominal drainage fluid and gastric tube drainage fluid may indicate gastrointestinal bleeding; Third, the color analysis method used in this technical solution lacks standardized colorimetric methods, making it difficult to achieve accurate comparison of color characteristics between different time points and different channels; Fourth, the abnormality detection mode of this technical solution is singular, focusing only on the longitudinal changes of a single detection object and lacking the ability to detect cross-channel lateral correlation, thus failing to identify cross-tube collaborative abnormal patterns.

[0005] Therefore, current technologies cannot effectively address the clinical monitoring challenges faced by patients in intensive care units with multiple drainage lines. These challenges include the need for separate observation and recording of the characteristics of each drainage fluid, low efficiency in overall assessment, and difficulty in identifying cross-line correlations in clinical monitoring. In clinical practice, nurses often need to observe each drainage line individually and record data manually. This is not only labor-intensive and prone to omissions, but also makes it difficult to grasp the overall trend of the patient's condition and identify potential correlations between lesions in different locations because the data from each line are independent. When abnormal changes occur simultaneously in different drainage lines, this pattern often indicates a more serious pathological condition, but current technologies lack effective means of identification.

[0006] In summary, there is an urgent need for an intelligent monitoring method and system for drainage fluid that supports multi-channel synchronous acquisition and comparative analysis. This system should be able to achieve synchronous acquisition of data from multiple drainage channels, standardized color quantification, cross-channel comparative analysis, and dual-mode abnormality detection, thereby improving the monitoring efficiency and quality of patients with multi-channel drainage in intensive care units and providing technical support for the early identification of complications in clinical practice. Summary of the Invention

[0007] To address the problems existing in the prior art, this invention provides a method and system for comparative analysis of multi-channel drainage fluid characteristics images.

[0008] This invention is implemented as follows: a multi-channel drainage fluid property image comparison and analysis method, which includes the following steps:

[0009] The multi-channel synchronous acquisition step involves the central processing unit sending a unified sampling time point to multiple connected drainage fluid acquisition terminals. Each drainage fluid acquisition terminal synchronously acquires image data of the corresponding drainage channel according to the unified sampling time point and adds acquisition timestamp information to each frame of image. The central processing unit receives the image data uploaded by each drainage fluid acquisition terminal and performs time alignment processing according to the acquisition timestamp information to generate multi-channel drainage fluid image data.

[0010] The color feature extraction step involves performing color feature extraction operations on the images of each channel in the multi-channel drainage fluid image data, mapping the extracted color features to a standardized chromaticity coordinate system, generating standardized chromaticity coordinate parameters for each channel, and generating time-series curves of color features for each channel based on the standardized chromaticity coordinate parameters arranged in a time series.

[0011] The multi-channel comparison analysis step aligns and displays the color feature time-series curves of each channel on the same time axis, calculates the chromaticity difference value between any two or more channels, compares and analyzes the color change trends of each channel, and generates multi-channel comparison analysis results.

[0012] The dual-mode anomaly detection step performs anomaly detection operations in two modes based on the results of multi-channel comparative analysis: single-channel longitudinal change detection and multi-channel lateral correlation detection. Single-channel longitudinal change detection monitors the temporal changes in the properties of the drainage fluid in each channel and identifies abrupt and gradual anomalies. Multi-channel lateral correlation detection analyzes the temporal correlation of changes in drainage fluid in different channels and identifies cross-channel collaborative anomalies, generating single-channel longitudinal change detection results, multi-channel lateral correlation detection results, and anomaly event record data.

[0013] Another objective of this invention is to provide a multi-channel drainage fluid characteristic image comparison analysis system for implementing the aforementioned multi-channel drainage fluid characteristic image comparison analysis method. The multi-channel drainage fluid characteristic image comparison analysis system includes: a multi-channel image acquisition module, a color feature extraction module, a multi-channel comparison analysis module, a dual-mode anomaly detection module, and a report generation and output module.

[0014] The multi-channel image acquisition module, connected to the central processing unit, is used to synchronously acquire image data of each drainage channel through multiple drainage fluid acquisition terminals according to a unified sampling time point, and to add acquisition timestamp information to each frame of image. After time alignment processing, multi-channel drainage fluid image data is generated.

[0015] The color feature extraction module, connected to the central processing unit, is used to perform color feature extraction operations on the images of each channel in the multi-channel drainage fluid image data, and generate standardized chromaticity coordinate parameters and color feature time-series curves for each channel.

[0016] The multi-channel contrast analysis module, connected to the central processing unit, is used to align and display the color feature time-series curves of each channel on the same time axis, calculate the color difference value and compare the color change trend, and generate multi-channel contrast analysis results.

[0017] The dual-mode anomaly detection module, connected to the central processing unit, is used to perform single-channel longitudinal change detection and multi-channel lateral correlation detection based on the results of multi-channel comparative analysis, and generate anomaly event record data.

[0018] The report generation and output module, connected to the central processing unit, is used to generate comprehensive monitoring reports based on the test results and output them through bedside display terminals and mobile applications.

[0019] In summary, the advantages and positive effects of this invention are as follows: This invention achieves time-consistent data acquisition from multiple drainage channels through a multi-channel synchronous acquisition mechanism, solving the problem of existing technologies being unable to simultaneously monitor drainage fluid from multiple channels; it achieves precise quantification of drainage fluid color characteristics through standardized chromaticity coordinate mapping, solving the problem of existing technologies relying on subjective judgment for color assessment; it achieves joint display and difference calculation of cross-channel data through multi-channel comparative analysis, improving overall assessment efficiency; and it simultaneously achieves single-channel longitudinal change detection and multi-channel lateral correlation detection through a dual-mode anomaly detection architecture, enabling the identification of cross-channel collaborative abnormal patterns and providing technical support for the early clinical detection of serious complications such as gastrointestinal bleeding. The technical solution of this invention can improve the efficiency of nursing staff's observation and recording of drainage fluid by approximately 40% to 60%, and the anomaly detection accuracy rate reaches approximately 92% or higher. Attached Figure Description

[0020] Figure 1 This is a flowchart of the multi-channel drainage fluid property image comparison and analysis method provided in the embodiments of the present invention.

[0021] Figure 2 This is a schematic diagram of the structure of the multi-channel drainage fluid property image comparison and analysis system provided in the embodiment of the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0023] To address the problems existing in the prior art, this invention provides a method and system for comparative analysis of multi-channel drainage fluid characteristics images. The invention will be described in detail below with reference to the accompanying drawings.

[0024] like Figure 1 As shown, the multi-channel drainage fluid property image comparison and analysis method provided in this embodiment of the invention includes the following steps:

[0025] Step S1, multi-channel synchronous acquisition step.

[0026] The technical objective of this step is to achieve time-synchronized acquisition and alignment of image data from multiple drainage channels, ensuring temporal consistency in the data foundation for subsequent multi-channel comparative analysis. In the clinical setting of the intensive care unit, patients with multiple drainage lines are typically connected to four to eight different types of drainage lines simultaneously. The characteristics of the drainage fluid from each line need to be collected and compared at the same time point to have clinical significance.

[0027] The specific implementation of this step is as follows: First, the central processing unit, as the control core of the entire system, is responsible for generating a unified sampling time point and broadcasting sampling instructions to all connected drainage fluid collection terminals via a wireless communication network. The central processing unit is equipped with a high-precision real-time clock module with a clock accuracy of ±10ms, ensuring that the time base of sampling instructions from each channel is consistent. The unified sampling time point is generated periodically according to preset sampling cycle parameters. The preset sampling cycle parameters can be configured within the range of 5 to 60 minutes according to clinical needs; a setting of 15 minutes or 30 minutes is recommended for routine monitoring scenarios.

[0028] Upon receiving the sampling command, each drainage fluid collection terminal immediately initiates the image acquisition process. The drainage fluid collection terminal adopts a modular design, consisting of two main components: a transmissive optical sensor and a clamping and fixing device. The transmissive optical sensor includes a visible light LED light source and an image sensor. The visible light LED light source uses white LEDs with a color temperature of 5000K to 6500K, with an output luminous flux of 100lm to 300lm, capable of forming uniform transmissive illumination in the drainage fluid sample area. The image sensor uses a CMOS image sensor with a resolution of no less than 2 million pixels, supporting an image acquisition rate of 30 frames per second. The clamping and fixing device is designed with an adjustable structure, adaptable to four common types of drainage containers: chest drainage bottles, abdominal drainage bottles, gastric tube drainage bags, and urine collection bags. Flexible clamping arms ensure stable fixation of drainage tubes or bags of different diameters.

[0029] After image acquisition, a timestamp is appended to each frame. The timestamp is encoded using the ISO 8601 date and time format, accurate to the millisecond, and is a continuous string of numbers representing year, month, day, hour, minute, second, and millisecond. Each drainage fluid acquisition terminal packages the image data and timestamp information and uploads it to the central processing unit via a wireless communication module. The wireless communication module supports both WiFi and Bluetooth Low Energy protocols, with a data transmission rate of at least 1 Mbps, meeting the real-time transmission requirements of high-resolution image data.

[0030] After receiving image data uploaded from each drainage fluid acquisition terminal, the central processing unit first performs time alignment processing. Time alignment is determined using a time deviation tolerance threshold, which can be configured from 0 to 5 seconds, with a default setting of 2 seconds. The specific processing flow is as follows: using a unified sampling time point as a reference, the absolute value of the time deviation between the timestamp of each channel's image acquisition and the reference time point is calculated. When the absolute value of the time deviation does not exceed the time deviation tolerance threshold, the image frame is determined to be valid synchronization data and included in the multi-channel drainage fluid image dataset. When the absolute value of the time deviation exceeds the time deviation tolerance threshold, the image frame is marked as time-abnormal data, and the reason for the abnormality is recorded. Time-abnormal data does not participate in subsequent multi-channel comparative analysis but is retained for system operation status monitoring and fault diagnosis.

[0031] After time alignment, the generated multi-channel drainage fluid image data has a unified time label. Image data from each channel under the same time label can be directly used for subsequent comparative analysis. The data structure of the multi-channel drainage fluid image data includes four components: a time label field, a channel identifier field, an image data field, and a metadata field. The time label field stores the unified sampling time point; the channel identifier field stores the drainage channel number corresponding to the image, with the number ranging from 1 to 8; the image data field stores the pixel matrix data of the original image; and the metadata field stores auxiliary information such as the acquisition terminal model, light source parameters, and exposure time.

[0032] Step S2, color feature extraction step.

[0033] The technical objective of this step is to convert the color information in multi-channel drainage fluid image data into standardized chromaticity coordinate parameters, achieving accurate quantification and cross-channel comparability of drainage fluid color characteristics. Existing methods for assessing drainage fluid color mainly rely on the subjective visual judgment of nursing staff, leading to significant differences in judgment among different observers and making it difficult to accurately describe subtle color changes. This step, by introducing the CIE standard chromaticity coordinate system developed by the International Commission on Illumination, maps the drainage fluid color onto a two-dimensional chromaticity plane, achieving an objective quantitative expression of color characteristics.

[0034] The specific implementation of this step is as follows: First, the region of interest (ROI) segmentation is performed on the images of each channel in the multi-channel drainage fluid image data. The purpose of ROI segmentation is to extract the effective region containing drainage fluid from the complete image and eliminate interference factors such as background, container edges, and air bubbles. The segmentation algorithm adopts a method based on a combination of color thresholding and morphological operations: First, the original RGB image is converted to the HSV color space, where H represents the hue component, S represents the saturation component, and V represents the lightness component; Second, initial segmentation is performed based on preset saturation and lightness thresholds, with the saturation threshold ranging from 0.15 to 0.85 and the lightness threshold ranging from 0.2 to 0.9, segmenting candidate regions that may contain drainage fluid; Third, morphological opening and closing operations are performed on the candidate regions. The opening operation uses a 5×5 pixel circular structural element, and the closing operation uses a 7×7 pixel circular structural element to remove noise spots and fill small holes; Fourth, the area and roundness parameters of each candidate region are calculated, and the region with the largest area and a roundness greater than 0.7 is retained as the final ROI.

[0035] After segmenting the region of interest (ROI), color statistical analysis is performed on the pixels within the ROI. The color statistical analysis includes the following steps: First, iterate through all pixels within the ROI and extract the RGB three-channel values ​​for each pixel. The RGB values ​​are integers ranging from 0 to 255. Second, calculate the arithmetic mean of the RGB three-channel values ​​for all pixels, and denote this as the average red component. Average green content Average blue component The third step is to normalize the average RGB components, mapping the numerical range from 0 to 255 to a floating-point range of 0 to 1, thus obtaining the normalized red component. Normalized green component Normalized blue component .

[0036] Normalized RGB components are converted to CIE 1931 XYZ tristimulus values ​​using a chromaticity coordinate transformation matrix. The chromaticity coordinate transformation matrix adopts the sRGB to XYZ transformation matrix recommended by the International Commission on Illumination (ICI), and the transformation formula is as follows:

[0037] ,

[0038] ,

[0039] ,

[0040] in: , , These are the CIE 1931 XYZ tristimulus values, which are dimensionless values ​​ranging from 0 to 1. These three parameters together describe the position of the color in the three-dimensional color space. , , These represent the normalized red, green, and blue components, respectively, with dimensionless values ​​ranging from 0 to 1, obtained by dividing the original RGB pixel values ​​by 255. The matrix coefficients 0.4124, 0.3576, 0.1805, 0.2126, 0.7152, 0.0722, 0.0193, 0.1192, and 0.9505 are the standard sRGB color space conversion coefficients recommended by the International Commission on Illumination (ICI). These coefficients are determined based on the human eye's perception of different wavelengths of light, ensuring that image data acquired by different devices can be converted to a unified chromaticity coordinate system. The technical effect of using this conversion matrix is ​​to achieve device-independent color representation and eliminate color deviations between different acquisition terminals.

[0041] The chromaticity coordinates were further calculated from the CIE 1931 XYZ tristimulus values. Values ​​and chromaticity coordinates The value is calculated using the following formula:

[0042] ,

[0043] ,

[0044] in: The x-axis component of the chromaticity coordinate system is a dimensionless value ranging from 0 to 1. It represents the position of the color on the x-axis of the CIE1931 chromaticity diagram and reflects the relative proportion of red and blue components in the color. The vertical component of the chromaticity coordinate system is a dimensionless value ranging from 0 to 1, representing the position of the color on the vertical axis of the CIE 1931 chromaticity diagram and reflecting the relative content of the green component in the color. , , The CIE 1931 XYZ tristimulus values ​​calculated using the aforementioned formula. Chromaticity coordinates. Values ​​and chromaticity coordinates The values ​​together constitute the standardized chromaticity coordinate parameters, which are represented in the form of ordered pairs. The standardized chromaticity coordinate parameters have a good correspondence with human eye's perception of color, and can intuitively reflect the hue characteristics of the drainage fluid. The technical effect of using CIE chromaticity coordinates instead of directly using RGB values ​​is to achieve uniform color quantification, that is, the same Euclidean distance in the chromaticity coordinate space corresponds to the degree of difference in the same color perceived by the human eye.

[0045] Based on the standardized chromaticity coordinate parameters calculated at each sampling time point, color feature time-series curves for each channel are generated by arranging them in a time series. The data structure of the color feature time-series curves is a time series array, where each element contains a sampling time point and its corresponding standardized chromaticity coordinate parameters. For channels with... The time series curve for each sampling point can be represented as:

[0046] ,

[0047] in: Indicates the first The color feature timing curve of the channel. The value range is from 1 to the total number of channels. , The typical value is 4 to 8; Indicates the first Each sampling time point The value range is from 1 to the total number of sampling points. The unit is minutes or ISO 8601 format timestamp; and They represent the first Chromaticity coordinates corresponding to each sampling time point Values ​​and chromaticity coordinates The values ​​are dimensionless values ​​ranging from 0 to 1. The color characteristic time series curves completely record the trajectory of color change of drainage fluid in each channel over time, providing a data foundation for subsequent multi-channel comparative analysis and anomaly detection.

[0048] Step S3, multi-channel comparison analysis step.

[0049] The technical objective of this step is to achieve joint analysis and comparison of color characteristic data of multi-channel drainage fluid, including three main functions: time axis aligned display, color difference calculation, and trend comparison analysis, providing clinical medical staff with an intuitive comprehensive view of multi-channel data.

[0050] The specific implementation method for this step is as follows: First, align and display the color feature time-series curves of each channel on the same time axis. The time axis alignment uses a multi-curve overlay binding method, with the horizontal axis representing a unified time coordinate and the vertical axis optionally displaying chromaticity coordinates. Values, chromaticity coordinates The timeline can be set to a single value or a combination of both. Different colors and line types are used to distinguish the color characteristics of each channel's timeline curves; for example, the first channel uses a solid red line, the second channel uses a dashed blue line, etc. The timeline display range can be scaled as needed, supporting viewing data from the most recent 1 hour, 6 hours, 12 hours, 24 hours, or a custom time range.

[0051] The chromaticity difference value is calculated using the Euclidean distance formula, based on the standardized chromaticity coordinate parameters of the two channels at the same time. Let the... Channel at time The standardized chromaticity coordinate parameters are , No. Channel at time The standardized chromaticity coordinate parameters are Then the two channels at time Color difference value The calculation formula is as follows:

[0052] ,

[0053] in: For the first Channel and the Channel at time The chromaticity difference value ranges from 0 to The dimensionless value indicates that the larger the value, the greater the difference in color between the two channels of drainage fluid; a value of 0 indicates that the two channels are exactly the same color. and The first Channel at time chromaticity coordinates Values ​​and chromaticity coordinates The value ranges from 0 to 1; and The first Channel at time chromaticity coordinates Values ​​and chromaticity coordinates The value ranges from 0 to 1; and This is the channel index, with a value range of 1 to... and Based on clinical experience, when the color difference value is less than 0.05, the drainage fluid from the two channels is considered to be basically the same color; when the color difference value is between 0.05 and 0.15, the drainage fluid from the two channels is considered to have a slight color difference; when the color difference value is greater than 0.15, the drainage fluid from the two channels is considered to have a significant color difference, and attention should be paid to whether there is an abnormality.

[0054] Trend comparison analysis is achieved by calculating the first derivative of the time-series curves of color characteristics for each channel. The first derivative reflects the rate and direction of color parameter change over time, and can quantitatively describe the trend of color change in the drainage fluid. For discretely sampled color characteristic time-series curves, the first derivative is numerically approximated using the central difference method, and the calculation formula is as follows:

[0055] ,

[0056] ,

[0057] in: and The first Channel chromaticity coordinates Value and The first derivative of the value with respect to time, after conversion, is h⁻¹, representing the change in chromaticity coordinates per hour; This is the current sampling time point, in minutes; and These are the previous and next sampling time points, respectively; , The chromaticity coordinates of the previous and next sampling time points are respectively. value; , The chromaticity coordinates of the previous and next sampling time points are respectively. The sign of the first derivative indicates the direction of change: a positive value indicates an increase in the color parameter, and a negative value indicates a decrease. The absolute value of the first derivative indicates the rate of change; a larger absolute value indicates a more drastic color change. By comparing the sign and magnitude of the first derivatives for each channel, it is possible to determine whether the color of the drainage fluid in each channel changes in the same direction and whether the rate of change is consistent, providing a basis for identifying cross-channel synergistic anomalies.

[0058] The data structure of the multi-channel comparative analysis results includes three components: time-axis aligned display data, a chromaticity difference matrix, and trend comparison data. The time-axis aligned display data is the time-series curve data used for visualization; the chromaticity difference matrix is... A symmetric matrix, the matrix elements are the time series data of the chromaticity difference between each channel pair, and the diagonal elements of the matrix are zero; the trend comparison data stores the calculation results and comparison conclusions of the first derivative of each channel.

[0059] Step S4, dual-mode anomaly detection step.

[0060] The technical objective of this step is to perform two modes of anomaly detection operations based on the results of multi-channel comparative analysis. This identifies longitudinal abnormal changes within a single channel and lateral correlations between multiple channels, enabling comprehensive detection and classification of abnormal drainage fluid characteristics. The dual-mode anomaly detection architecture is designed based on two main manifestations of abnormal drainage fluid in clinical practice: one is abnormal changes in the characteristics of a single drainage channel, and the other is abnormal correlations between changes in multiple drainage channels.

[0061] The specific implementation of the single-channel longitudinal change detection mode is as follows: This mode is executed independently for each drainage channel, monitoring the temporal changes in the properties of the drainage fluid in each channel and identifying two types of abnormalities, namely abrupt abnormalities and gradual abnormalities.

[0062] Abrupt changes in drainage color refer to drastic changes within a short period, typically indicating an acute event such as postoperative bleeding or recanalization of a blocked drainage tube. The identification method for abrupt changes employs a standardized deviation detection algorithm, with the specific calculation process as follows: First, define the historical window length parameter. The value ranges from 6 to 24 sampling points, with a default setting of 12 sampling points, corresponding to a 3-hour history window with a sampling period of 15 minutes. Next, the mean value of the chromaticity coordinate parameters within the history window is calculated. with standard deviation The calculation formula is:

[0063] ,

[0064] ,

[0065] in: Chromaticity coordinates within the history window The mean value, a dimensionless value ranging from 0 to 1, represents the average state of the drainage fluid color in that channel over the recent period; Chromaticity coordinates within the history window The standard deviation of the value, a dimensionless value ranging from 0 to above, indicates the degree of recent fluctuation in the color of the drainage fluid in that channel; This is the historical window length parameter, expressed in the number of sampling points, with a typical value of 12. For the first time in the history window Chromaticity coordinates of each sampling point value, The value range is 1 to ; This is the current sampling time point. Chromaticity coordinates. mean of values with standard deviation The same formula is used for calculation.

[0066] Then, calculate the standardized deviation value of the current sampling point. The calculation formula is:

[0067] ,

[0068] in: Standardized deviation is a dimensionless value ranging from 0 to above, representing the degree of deviation of the color parameter of the current sampling point from the historical mean, measured in units of standard deviation. and These are the chromaticity coordinates of the current sampling point. Value and value; and These are the chromaticity coordinates within the history window. Value and The mean of the values; and These are the chromaticity coordinates within the history window. Value and The standard deviation of the value. When the standardized deviation value Exceeding the preset threshold for mutant anomalies When this occurs, it is identified as a mutant anomaly. A preset mutant anomaly threshold is set. The configurable range is 2 to 5, with a default setting of 3. This setting corresponds to an anomaly detection standard with approximately 99.7% confidence in statistics, meaning that an anomaly is considered to have occurred when the color parameter deviates from the historical mean by more than three standard deviations. The technical advantage of using the standardized deviation detection algorithm is that it can adaptively set the anomaly detection threshold based on the historical fluctuation characteristics of the drainage fluid in each channel, avoiding false alarms or missed alarms caused by fixed thresholds.

[0069] Progressive abnormalities refer to a slow, sustained change in the color of drainage fluid over a prolonged period, typically indicating the development of a chronic pathological process, such as gradually worsening infection or a gradual decrease in drainage volume. The identification method for progressive abnormalities employs a linear regression slope detection algorithm, with the specific calculation process as follows: First, define the sliding window length parameter. The value range is 12 to 48 sampling points, with a default setting of 24 sampling points, corresponding to a 6-hour sliding window with a sampling period of 15 minutes. Next, linear regression is performed on the color feature time-series curve within the sliding window, and the regression slope is calculated as the gradual change rate value. The linear regression uses the least squares method, and the slope calculation formula is:

[0070] ,

[0071] in: Chromaticity coordinates The gradual rate of change of the value, after conversion, is in h. -1 , represents the change in chromaticity coordinates per hour, with positive values ​​indicating that the color parameter increases over time and negative values ​​indicating that the color parameter decreases over time; For the first in the sliding window Each sampling time point is represented in minutes. This represents the average of the sampling time points within the sliding window; For the first in the sliding window Chromaticity coordinates of each sampling point value; Chromaticity coordinates within the sliding window The mean of the values; This is the sliding window length parameter, expressed in the number of sampling points. Chromaticity coordinates. The gradual rate of change of value The same formula is used for calculation. When the absolute value of the gradual rate exceeds the preset gradual anomaly threshold... When this occurs, it is classified as a progressive anomaly. A preset progressive anomaly threshold is set. The configurable range is 0.1 to 0.5 (in terms of hourly chromaticity coordinate change), with a default setting of 0.2. This setting corresponds to a gradual rate of change of approximately 0.05 to 0.1 chromaticity coordinates per 24 hours, a rate of change that is generally considered clinically significant. The technical advantage of using a linear regression slope detection algorithm is its ability to identify slow but continuous color change trends, which may be difficult to detect in a single observation but have important clinical implications.

[0072] The specific implementation of the multi-channel lateral correlation detection mode is as follows: This mode is executed jointly for multiple drainage channels, analyzing the temporal correlation of changes in drainage fluid in different channels and identifying cross-channel synergistic anomalies. Cross-channel synergistic anomalies refer to the specific temporal correlation pattern of color changes in multiple drainage channels. This correlation pattern often indicates a more serious pathological condition or a specific type of complication.

[0073] The quantification of temporal correlation is achieved using the cross-correlation coefficient. The cross-correlation coefficient measures the degree of correlation between two time series at different time lags, and its calculation formula is as follows:

[0074] ,

[0075] in: The cross-correlation coefficient is a dimensionless value ranging from -1 to +1. The closer the absolute value is to 1, the stronger the correlation. A positive value indicates a positive correlation (both channels change in the same direction), a negative value indicates a negative correlation (both channels change in opposite directions), and a value close to 0 indicates no significant correlation. This is a time lag parameter, expressed in units of sampling points, with a value range of [value range missing]. to Positive value indicates the first Changes in the channel lag behind the first Channel, negative value indicates the first Changes in the channel precede the first aisle; The maximum lag parameter can be configured from 1 to 20 sampling points, with a default setting of 5 sampling points, corresponding to a maximum time lag of 75 minutes when the sampling period is 15 minutes. For the first Channel at time chromaticity coordinate parameters; For the first Channel at time chromaticity coordinate parameters; and The first Mean and standard deviation of channel chromaticity coordinate parameters; and The first Mean and standard deviation of channel chromaticity coordinate parameters; This represents the total number of sampling points in the time series. The technique of using cross-correlation coefficients instead of simple correlation coefficients is effective in detecting association patterns with time lags, which is clinically significant. For example, upper gastrointestinal bleeding may cause bloody changes in gastric tube drainage fluid before those in abdominal drainage fluid.

[0076] When the absolute value of the cross-correlation coefficient exceeds the preset abnormal correlation threshold When this occurs, it is determined to be a cross-channel collaboration anomaly. A preset correlation anomaly threshold is set. The configurable range is 0.6 to 0.9, with a default setting of 0.75. This setting corresponds to the statistical criterion for significant correlation, which states that when the absolute value of the cross-correlation coefficient between two channels exceeds 0.75, a significant cross-channel association is considered to exist.

[0077] Cross-channel co-occurrence anomalies are further classified into two types based on their specific manifestations: synchronous blood color change anomalies and amplified difference anomalies. The criteria for synchronous blood color change anomalies are that two or more channels exhibit a shift in chromaticity coordinates towards the blood color region within the same time window, and the cross-correlation coefficient exceeds a preset correlation anomaly threshold. The blood color region is defined in the CIE 1931 chromaticity coordinate system as chromaticity coordinates... Value greater than 0.45 and chromaticity coordinates A value less than 0.35 corresponds to a red to dark red color range, representing the typical color characteristic of blood mixed with drainage fluid. When a synchronous abnormal change in blood color is detected, the system generates a high-priority alarm, indicating potential bleeding from multiple sites or gastrointestinal bleeding. The criterion for abnormally widening differences is that the color difference between two similar drainage channels (e.g., left and right pleural drainage) continuously increases within a preset observation period, and the increase exceeds a preset difference amplification threshold. The preset observation period is configurable from 6 hours to 24 hours, with a default setting of 12 hours; the preset difference amplification threshold is configurable from 0.05 to 0.2, with a default setting of 0.1. This setting corresponds to a 0.1 increase in color difference from the baseline level, indicating that the colors of the drainage fluids on both sides change from similar to significantly different. When an abnormally widening difference is detected, the system generates a medium-priority alarm, indicating potential unilateral disease progression.

[0078] The output of dual-mode anomaly detection includes single-channel longitudinal change detection results, multi-channel lateral correlation detection results, and anomaly event recording data. Single-channel longitudinal change detection results include abrupt anomaly markers and progressive anomaly markers for each channel; multi-channel lateral correlation detection results include cross-correlation matrix and cross-channel collaborative anomaly classification; anomaly event recording data includes detailed information such as anomaly occurrence time, anomaly type, involved channels, and severity.

[0079] Step S5, report generation and output steps.

[0080] The technical objective of this step is to summarize and organize all the test results generated in the preceding steps, generate a structured comprehensive monitoring report, and output and display it through various terminal devices to meet the information access needs in different clinical scenarios.

[0081] The specific implementation method of this step is as follows: First, data is summarized and integrated based on the single-channel longitudinal change detection results, multi-channel lateral correlation detection results, and abnormal event record data. The data summarization process extracts the output data scattered across various detection modules and organizes it according to the format requirements of the report template. The report template adopts a structured design, including four main parts: report header information, a summary table of multi-channel drainage fluid property changes, an abnormal event timeline, and a trend comparison chart.

[0082] The report header information includes basic information such as patient identification, bed number, monitoring time range, report generation time, and attending nurse. The multi-channel drainage fluid characteristic change summary table displays the color status changes of each drainage channel within the monitoring period in tabular form. Rows correspond to each drainage channel, columns to each sampling time point, and cell content is a description of the channel's color status at that moment. The color status description is automatically generated based on standardized chromaticity coordinate parameters, using commonly used clinical descriptive language, such as pale yellow, bloody, and purulent. The abnormal event timeline displays all abnormal events occurring within the monitoring period in Gantt chart form. The horizontal axis is the time axis, and each abnormal event is represented by a different colored bar. The start and end positions of the bars correspond to the occurrence time and duration of the abnormal event, and the bar color corresponds to the abnormality type; for example, red indicates abrupt abnormalities, orange indicates progressive abnormalities, and purple indicates cross-channel synergistic abnormalities. The trend comparison chart displays a comparison view of the color characteristic time series curves of each channel in multi-curve line chart form, supporting interactive operations such as channel filtering, time range scaling, and parameter switching.

[0083] The comprehensive monitoring report is displayed via both a bedside display terminal and a mobile application. The bedside display terminal uses a 10- to 15-inch touchscreen display mounted on a device bracket beside the patient's bed, displaying the current status and trend arrows of each drainage fluid in real time in a multi-channel dashboard format. The multi-channel dashboard interface is designed with a partitioned layout, with each partition corresponding to a drainage channel. Each partition displays the channel name, current color sample, chromaticity coordinate value, recent trend arrow, and anomaly marker icon. The trend arrow is automatically generated based on the first derivative calculation results: an upward arrow indicates an increase in color parameters, a downward arrow indicates a decrease in color parameters, and a horizontal line indicates relative stability. Anomaly marker icons are displayed when an anomaly is detected; the icon shape and color correspond to the anomaly type and severity.

[0084] The mobile application allows nursing staff to remotely view multi-channel drainage monitoring data of patients and receive push notifications for abnormalities. The mobile application features a responsive interface design, adapting to both smartphones and tablets. Application functions include viewing patient lists, displaying real-time multi-channel data, reviewing historical data, viewing comprehensive monitoring reports, and setting up push notifications for abnormalities. The push notification system employs a tiered mechanism: high-priority abnormalities, such as synchronous changes in blood quality, receive immediate push notifications accompanied by vibration and a ringtone; medium-priority abnormalities, such as widening differences, receive regular push notifications; and low-priority abnormalities, such as slight parameter fluctuations, receive summary push notifications. Nursing staff can configure the scope and notification method of push notifications according to their personal preferences.

[0085] The comprehensive monitoring report supports exporting to two standard formats: nursing record format and medical record archiving format. The nursing record format is formatted according to the nursing record writing standards of medical institutions, including necessary elements such as observation time, observation content, abnormal conditions, and treatment measures, and can be directly used for nursing handover records and nursing document archiving. The medical record archiving format uses structured coding according to electronic medical record archiving standards and data encapsulation in the HL7 FHIR format, a medical information exchange standard, supporting data interface with hospital information systems and long-term storage.

[0086] like Figure 2 As shown, the multi-channel drainage fluid property image comparison and analysis system provided in this embodiment of the invention includes: a multi-channel image acquisition module 1, a color feature extraction module 2, a multi-channel comparison and analysis module 3, a dual-mode anomaly detection module 4, a report generation and output module 5, and a central processing unit 6.

[0087] The multi-channel image acquisition module 1, connected to the central processing unit 6, is used to synchronously acquire image data of each drainage channel through multiple drainage fluid acquisition terminals at a unified sampling time point, and to attach acquisition timestamp information to each frame of image. After time alignment processing, multi-channel drainage fluid image data is generated. The hardware components of the multi-channel image acquisition module 1 include four to eight drainage fluid acquisition terminals, a wireless communication gateway, and a clock synchronization server. Each drainage fluid acquisition terminal interacts with the central processing unit 6 through the wireless communication gateway, and the clock synchronization server is responsible for maintaining the time base of the entire system. The software functions of the multi-channel image acquisition module 1 include sampling command broadcasting, image acquisition control, timestamp attachment, data uploading, and time alignment processing. The specific algorithm flow is as described in step S1 above.

[0088] Color feature extraction module 2, connected to the central processing unit 6, performs color feature extraction operations on the images of each channel in the multi-channel drainage fluid image data, generating standardized chromaticity coordinate parameters and color feature time-series curves for each channel. Color feature extraction module 2 is implemented using an embedded image processing chip, supporting parallel processing of multi-channel images. The software functions of color feature extraction module 2 include region of interest segmentation, pixel color statistics, color space conversion, chromaticity coordinate calculation, and time-series curve generation; the specific algorithm flow is as described in step S2 above.

[0089] The multi-channel contrast analysis module 3, connected to the central processing unit 6, is used to align and display the color feature time-series curves of each channel on the same time axis, calculate the chromaticity difference value, compare the color change trend, and generate multi-channel contrast analysis results. The software functions of the multi-channel contrast analysis module 3 include time axis alignment display data preparation, chromaticity difference matrix calculation, first derivative calculation, and trend contrast analysis. The specific algorithm flow is as described in step S3 above.

[0090] The dual-mode anomaly detection module 4, connected to the central processing unit 6, is used to perform single-channel vertical change detection and multi-channel horizontal correlation detection based on the results of multi-channel comparative analysis, generating anomaly event record data. The dual-mode anomaly detection module 4 includes two functional units: a single-channel vertical change detection submodule and a multi-channel horizontal correlation detection submodule. The single-channel vertical change detection submodule implements abrupt anomaly detection algorithm and a progressive anomaly detection algorithm; the multi-channel horizontal correlation detection submodule implements a cross-correlation coefficient calculation algorithm and a cross-channel collaborative anomaly classification algorithm, with the specific algorithm flow described in step S4 above.

[0091] The report generation and output module 5, connected to the central processing unit 6, is used to generate a comprehensive monitoring report based on the test results and output it through a bedside display terminal and a mobile application. The software functions of the report generation and output module 5 include data aggregation and integration, report template rendering, multi-channel dashboard interface generation, mobile data synchronization, and abnormal push notification sending, as described in step S5 above. The report generation and output module 5 also includes a data export function, supporting the export of the comprehensive monitoring report to nursing record format and medical record archiving format.

[0092] The central processing unit (CPU) 6 is implemented using an industrial-grade embedded computer, with a processor clock speed of no less than 2GHz, a memory capacity of no less than 8GB, and a storage capacity of no less than 256GB solid-state drive. CPU 6 runs a Linux operating system, and the upper-layer application software is developed using Python. Image processing algorithms are implemented using the OpenCV image processing library, and data storage uses an embedded SQLite database. CPU 6, along with the multi-channel image acquisition module 1, color feature extraction module 2, multi-channel comparison analysis module 3, dual-mode anomaly detection module 4, and report generation and output module 5, adopts a modular software architecture. Each module interacts with data through standardized interfaces, facilitating system expansion and maintenance.

[0093] The multi-channel drainage fluid characteristic image comparison and analysis system provided in this invention has achieved good results in clinical application testing. The test was conducted in the intensive care unit of a tertiary-level hospital, with the subjects being postoperative patients with multiple drainage tubes. The test period was 30 days, and a total of 87 patients were monitored. Test results show that: the multi-channel synchronous acquisition mechanism of this invention can achieve synchronous acquisition with a time deviation of less than 2 seconds, and the acquisition success rate reaches approximately 98.5%; the color feature extraction algorithm of this invention can stably map the drainage fluid color to a standardized chromaticity coordinate system, with color recognition consistency reaching over 96%; the dual-mode anomaly detection algorithm of this invention can effectively identify single-channel anomalies and cross-channel collaborative anomalies, with an anomaly detection accuracy rate of approximately 92%, including approximately 94% accuracy for abrupt anomalies, approximately 89% accuracy for progressive anomalies, and approximately 91% accuracy for cross-channel collaborative anomalies; compared with traditional manual observation and recording methods, this invention improves the efficiency of drainage fluid monitoring for nursing staff by approximately 40% to 60%.

[0094] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, an optical medium, or a semiconductor medium, etc.

[0095] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for comparative analysis of multi-channel drainage fluid characteristics images, characterized in that, The multi-channel drainage fluid property image comparison and analysis method includes the following steps: The multi-channel synchronous acquisition step involves sending a unified sampling time point to multiple connected drainage fluid acquisition terminals through a central processing unit. Each drainage fluid acquisition terminal synchronously acquires image data of the corresponding drainage channel according to the unified sampling time point and adds acquisition timestamp information to each frame of image. The central processing unit receives the image data uploaded by each drainage fluid acquisition terminal and performs time alignment processing according to the acquisition timestamp information to generate multi-channel drainage fluid image data. The color feature extraction step involves performing color feature extraction operations on the images of each channel in the multi-channel drainage fluid image data, mapping the extracted color features to a standardized chromaticity coordinate system, generating standardized chromaticity coordinate parameters for each channel, and generating a time-series curve of color features for each channel based on the standardized chromaticity coordinate parameters arranged in a time series. The multi-channel comparison analysis step involves aligning and displaying the color feature time-series curves of each channel on the same time axis, calculating the chromaticity difference value between any two or more channels, comparing and analyzing the color change trends of each channel, and generating multi-channel comparison analysis results. The dual-mode anomaly detection step performs anomaly detection operations in two modes based on the multi-channel comparative analysis results: single-channel longitudinal change detection and multi-channel lateral correlation detection. The single-channel longitudinal change detection monitors the temporal changes in the properties of the drainage fluid in each channel and identifies abrupt and gradual anomalies. The multi-channel lateral correlation detection analyzes the temporal correlation of changes in drainage fluid in different channels and identifies cross-channel collaborative anomalies, generating single-channel longitudinal change detection results, multi-channel lateral correlation detection results, and anomaly event record data.

2. The multi-channel drainage fluid property image comparison and analysis method as described in claim 1, characterized in that, In the multi-channel synchronous acquisition step, the central processing unit supports simultaneous connection of four to eight drainage fluid acquisition terminals. The unified sampling time point is generated according to a preset sampling period parameter, the value range of which is five to sixty minutes. The time alignment processing uses a time deviation tolerance threshold for alignment judgment, the value range of which is zero to five seconds. When the deviation between the acquisition timestamp of any channel image and the unified sampling time point exceeds the time deviation tolerance threshold, the frame image is marked as time abnormal data.

3. The multi-channel drainage fluid property image comparison and analysis method as described in claim 1, characterized in that, In the color feature extraction step, the standardized chromaticity coordinate system adopts the CIE 1931 chromaticity coordinate system. The color feature extraction operation includes segmenting the image into regions of interest, performing pixel color statistics on the regions of interest, and converting the statistical results into CIE 1931 XYZ tristimulus values ​​based on the chromaticity coordinate transformation matrix. Then, the chromaticity coordinate x-values ​​and chromaticity coordinate y-values ​​are calculated from the CIE 1931 XYZ tristimulus values ​​as the standardized chromaticity coordinate parameters.

4. The multi-channel drainage fluid property image comparison and analysis method as described in claim 1, characterized in that, In the multi-channel comparison analysis step, the chromaticity difference value is calculated using the Euclidean distance formula, based on the standardized chromaticity coordinate parameters of the two channels at the same time. The trend comparison analysis includes calculating the first derivative of the color feature time series curve of each channel and comparing the direction and rate of color change of each channel.

5. The multi-channel drainage fluid property image comparison and analysis method as described in claim 1, characterized in that, In the multi-channel lateral correlation detection, the temporal correlation is quantified by calculating the cross-correlation coefficient between the time-series curves of the color features of different channels. The cross-correlation coefficient is calculated based on the time lag parameter, which ranges from the negative maximum lag to the positive maximum lag. The maximum lag ranges from one to twenty sampling points. When the absolute value of the cross-correlation coefficient exceeds a preset correlation anomaly threshold, it is determined to be a cross-channel collaborative anomaly. The preset correlation anomaly threshold ranges from 0.6 to 0.

9.

6. The multi-channel drainage fluid property image comparison and analysis method as described in claim 1, characterized in that, In the single-channel longitudinal change detection, the method for identifying abrupt anomalies is to calculate the deviation between the standardized chromaticity coordinate parameters of the current sampling point and the mean within the historical window, and divide it by the standard deviation within the historical window to obtain the standardized deviation value. When the standardized deviation value exceeds a preset abrupt anomaly threshold, it is determined to be an abrupt anomaly. The preset abrupt anomaly threshold ranges from two to five. The method for identifying gradual anomalies is to calculate the linear regression slope of the color feature time-series curve within the sliding window as the gradual change rate value. When the gradual change rate value exceeds a preset gradual anomaly threshold, it is determined to be a gradual anomaly. The preset gradual anomaly threshold ranges from 0.1 to 0.

5.

7. The multi-channel drainage fluid property image comparison and analysis method as described in claim 5, characterized in that, The cross-channel collaborative anomalies include two types: synchronous blood color change anomalies and difference expansion anomalies. The criteria for determining synchronous blood color change anomalies are that two or more channels show a shift in chromaticity coordinates toward the blood color region within the same time window, and the absolute value of the cross-correlation coefficient exceeds the preset correlation anomaly threshold. The criteria for determining difference expansion anomalies are that the chromaticity difference value between two drainage channels of the same type continues to increase within a preset observation period, and the increase exceeds the preset difference increase threshold.

8. The multi-channel drainage fluid property image comparison and analysis method as described in claim 1, characterized in that, The drainage fluid collection terminal adopts a modular design, including a transmissive optical sensor and a clamping and fixing device. The transmissive optical sensor includes a visible light LED light source and an image sensor. The visible light LED light source uses a white LED with a color temperature of 5,000 to 6,500 Kelvin. The clamping and fixing device is compatible with four types of drainage containers: chest drainage bottle, abdominal drainage bottle, gastric tube drainage bag, and urine collection bag.

9. The multi-channel drainage fluid property image comparison and analysis method as described in claim 1, characterized in that, It also includes report generation and output steps, which generate a comprehensive monitoring report based on the single-channel longitudinal change detection results, the multi-channel lateral correlation detection results, and the abnormal event record data. The report includes a summary table of changes in the properties of the multi-channel drainage fluid, an abnormal event timeline, and a trend comparison chart. The report is then displayed and output through a bedside display terminal and a mobile application. The bedside display terminal displays the current status and trend arrows of each drainage fluid in real time in the form of a multi-channel dashboard. The mobile application allows nursing staff to remotely view the patient's multi-channel drainage monitoring data and receive abnormal push notifications. The comprehensive monitoring report can be exported as a nursing record format and a medical record archive format.

10. A multi-channel drainage fluid property image comparison and analysis system, used to implement the multi-channel drainage fluid property image comparison and analysis method according to any one of claims 1-9, characterized in that, The multi-channel drainage fluid property image comparison and analysis system includes: The multi-channel image acquisition module is connected to the central processing unit. It is used to synchronously acquire image data of each drainage channel through multiple drainage fluid acquisition terminals according to a unified sampling time point, and to add acquisition timestamp information to each frame of image. After time alignment processing, multi-channel drainage fluid image data is generated. The color feature extraction module is connected to the central processing unit and is used to perform color feature extraction operations on the images of each channel in the multi-channel drainage fluid image data to generate standardized chromaticity coordinate parameters and color feature time-series curves for each channel. The multi-channel comparison analysis module, connected to the central processing unit, is used to align and display the color feature time-series curves of each channel on the same time axis, calculate the color difference value and compare the color change trend, and generate multi-channel comparison analysis results. A dual-mode anomaly detection module, connected to the central processing unit, is used to perform single-channel longitudinal change detection and multi-channel lateral correlation detection based on the multi-channel comparative analysis results, and generate anomaly event record data. The report generation and output module, connected to the central processing unit, is used to generate a comprehensive monitoring report based on the detection results and output and display it through a bedside display terminal and a mobile application.

Images