A lipoprotein classification detection system based on electrophoretic image analysis

Abstract

This invention discloses a lipoprotein classification and detection system based on electrophoresis image analysis, specifically relating to the field of lipoprotein classification and detection. The system includes an intelligent sensing system comprising a preprocessing unit, an electrophoresis unit, a scanning unit, and a control host. The output of the preprocessing unit is connected to the input of the electrophoresis unit, and the output of the electrophoresis unit is connected to the input of the scanning unit. The preprocessing unit, electrophoresis unit, and scanning unit are all connected to the control host to form a detection execution path and a status acquisition path for the same sample. By associating the preprocessing state, electrophoresis state, and scanning state with corresponding band regions, and mapping the results back to candidate segments, reference criteria, and status sources after classification, the traceability and reliable discrimination ability of the lipoprotein classification and determination results can be relatively improved. This directly addresses the problem of difficulty in determining results when electrophoresis images are disturbed by multiple factors.

Description

Technical Field

[0001] This invention relates to the field of lipoprotein classification and detection technology, and more specifically, to a lipoprotein classification and detection system based on electrophoretic image analysis. Background Technology

[0002] In lipoprotein subtyping detection, existing technologies are mainly designed to address the problems of difficulty in directly distinguishing different lipoprotein components and the inability of conventional blood lipid indicators to reflect the sub-risks. The usual approach is to first preprocess the sample, then form lipoprotein migration bands through gradient gel electrophoresis, and then scan the electrophoresis results to obtain images. The analysis software is then used to complete band identification, subtype classification, cholesterol conversion, and report output. Some testing platforms also combine quality control materials, electrophoresis devices, scanning modules, and analysis modules into a continuous testing process. Taking batch lipoprotein classification testing in hospital laboratories or third-party medical laboratories as an example, under the condition that the same batch of samples needs to continuously complete pretreatment, electrophoretic separation, scanning imaging and automatic interpretation, and the test results are required to be generated directly within a limited time and cannot rely on manual verification of each sample for a long time, even if the overall testing process is running normally, the actual spectrum will still repeatedly show phenomena such as unclear band layering, local shadow interference, irregular band shape or boundary offset. In this case, if the subtype classification and bile determination results are still directly executed based on a single scan image, it is often difficult to distinguish whether the abnormal results come from the true lipoprotein distribution of the sample itself, or from pretreatment fluctuations, electrophoretic migration disturbances or scanning imaging deviations. In the end, although the results can be generated, the basis for their formation is difficult to trace and the reliability is difficult to determine. The existing processing method is difficult to meet the requirements of automatic interpretation and reliable release of classification test results under the condition of intelligent sensing system. The technical problem this application aims to solve is: how to, during the lipoprotein electrophoresis classification and detection process, when the electrophoresis image is simultaneously affected by the preprocessing state, the electrophoretic migration state, and the scanning imaging state, still be able to perform correlation perception, credibility judgment, and result verification of the basis for the formation of each lipoprotein classification and determination result, so as to output verifiable lipoprotein classification and detection results. Summary of the Invention

[0003] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a lipoprotein classification and detection system based on electrophoretic image analysis. By associating the preprocessing state, electrophoresis state, and scanning state with the band regions in the electrophoretic image, and combining this with image reference to complete band alignment, classification mapping, result verification, and credibility checks, the system solves the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a lipoprotein classification and detection system based on electrophoretic image analysis, comprising: The intelligent sensing system includes a preprocessing unit, an electrophoresis unit, a scanning unit, and a control host. The output of the preprocessing unit is connected to the input of the electrophoresis unit, and the output of the electrophoresis unit is connected to the input of the scanning unit. The preprocessing unit, electrophoresis unit, and scanning unit are all connected to the control host to form a detection execution path and a status acquisition path for the same sample. The reference component includes a quality control loading unit and a reference storage unit. The quality control loading unit works with the electrophoresis unit to form a reference lane for the same batch of tests. The reference storage unit is connected to the control host and is used to store the reference image, migration position and classification interval of the corresponding test batch. The state chain module is used to read the state data of the intelligent sensing system during the pretreatment, electrophoresis and scanning processes, and to match, sort and write the state data according to the sample identifier and detection batch, thereby outputting the state chain corresponding to the sample to be tested. The mapping module is used to read the electrophoresis image, state chain and reference image of the corresponding detection batch of the sample to be tested, perform position alignment on the electrophoresis image according to the migration position in the reference image, divide the aligned band image into multiple candidate segments, and then compare each candidate segment with the classification interval one by one, thereby outputting the lipoprotein classification record corresponding to each candidate segment; The verification module is used to read lipoprotein classification records, status chains, and bile index data of the sample to be tested. It maps each lipoprotein classification record to the band segment that formed the classification record, the reference basis, and the status source, and performs consistency verification between each lipoprotein classification record and bile index data, thereby outputting lipoprotein classification bile determination results and reliable records. The reporting module is used to read lipoprotein classification and determination results and reliable records. Based on the reliable records, it directly outputs the results, prompts for rescanning, and prompts for verification, and generates a test report containing lipoprotein classification map, lipoprotein classification and determination results, and verification information.

[0005] In a preferred embodiment, the execution of the intelligent sensing system includes: The preprocessing unit receives the sample to be tested and performs centrifugation and filtration. The electrophoresis unit receives the preprocessed sample to be tested and forms an electrophoretic band with a corresponding sample identifier. The scanning unit scans and captures images of the electrophoretic band. The control host receives the status data written by the preprocessing unit, electrophoresis unit and scanning unit at the corresponding execution time and generates a detection process record with a corresponding sample identifier. The control host matches the detection process record with the electrophoresis image output by the scanning unit. Based on the sample identifier and execution time, it associates the preprocessing state, electrophoresis state, and scanning state with the corresponding strip regions. When the same strip region corresponds to one preprocessing state, one electrophoresis state, and one scanning state, and the execution times of the three are arranged in the order of preprocessing, electrophoresis, and scanning, the strip region is marked as a valid area. When the same strip region corresponds to one preprocessing state, one electrophoresis state, and one scanning state, but two of the three have the same execution time, are in reverse order, or their corresponding ranges overlap with adjacent strip regions, the strip region is marked as an interference area. When the same strip region lacks any of the preprocessing state, electrophoresis state, or scanning state, or when the same state corresponds to two or more non-adjacent strip regions, the strip region is marked as an abnormal area, thus outputting an electrophoresis image record with status markings. The control host associates the detection process record with the electrophoretic image output by the scanning unit and stores the electrophoretic image record, thereby forming a correspondence between the detection execution path result and the status acquisition path result of the same sample.

[0006] In a preferred embodiment, the execution of the reference component includes: The quality control loading unit receives the quality control samples of the corresponding test batch and sends the quality control samples into the electrophoresis unit. The electrophoresis unit performs electrophoretic separation on the quality control samples and forms a reference lane. The scanning unit performs scanning and image acquisition on the reference lane. The reference storage unit receives and stores the reference image of the corresponding test batch. The control host reads the reference image, identifies the start and end positions of each band along the migration direction of the reference swimlane, writes the start and end positions of each band as the migration position of the corresponding band, and writes the image segments between adjacent bands and the corresponding segments of each band as the corresponding classification intervals. The reference storage unit receives and stores the migration positions and classification intervals. The control host will associate the reference image, migration location, and classification range according to the detection batch. The reference storage unit will perform batch storage on the associated reference image, migration location, and classification range, thereby outputting the reference basis for the corresponding detection batch.

[0007] In a preferred embodiment, the execution of the state chain module includes: Read the status data written by the intelligent sensing system during the pretreatment, electrophoresis and scanning processes, extract the sample identifier, detection batch, process category, execution time and status content from each status data, and identify the status data with the same sample identifier and the same detection batch as the data to be processed corresponding to the same sample. The data to be processed is arranged in order of execution time from first to last. When the execution time of two state data is the same, the order of preprocessing, electrophoresis and scanning is determined. The arranged state data is then written into the state sequence of the corresponding sample identifier to output the initial state chain.

[0008] In a preferred embodiment, the execution of the state chain module further includes: The process state fragment chain is output by comparing two adjacent state data in the initial state chain item by item. When the process categories and state contents are the same, the latter state data is merged into the former state data and the end time of the former state data is updated at the execution time of the latter state data. When the process categories and state contents are different, the latter state data is written after the former state data and the changes between the two are recorded. Read each process state segment in the process state segment chain in the order of pretreatment, electrophoresis and scanning. Connect the first and last process state segments in the complete sequence to form the state chain corresponding to the sample to be tested. Remove process state segments that are missing preceding or following process state segments from the state chain.

[0009] In a preferred embodiment, the execution of the mapping module includes: Read the electrophoresis image, state chain, and reference image of the corresponding detection batch of the sample to be tested. Extract the strip regions corresponding to the effective area marker, interference area marker, and abnormal area marker from the electrophoresis image. Extract the migration position in the reference image along the migration direction. Extract the position to be aligned in the strip region corresponding to the effective area marker. Calculate the position difference between each position to be aligned and each migration position. Perform position alignment according to the position correspondence relationship of the consistent migration direction order and the first position in the total position difference sorted first. Then output the aligned strip image. Pixel-wise grayscale integration is performed on the aligned strip image along the migration direction. In the strip region corresponding to the effective region, the position where the grayscale integration changes from rising to falling is written as the starting boundary, and the position where the grayscale integration changes from falling to rising is written as the ending boundary. The image segment between the adjacent starting boundary and ending boundary is written as the initial segment. Then, the boundary distance and grayscale continuity order between the strip region corresponding to the interference region and the adjacent initial segments are calculated. The strip region corresponding to the interference region with the highest boundary distance and the same grayscale continuity order is merged into the initial segment on the corresponding side. The strip region corresponding to the abnormal region is removed from the segment division, thus outputting the candidate segment.

[0010] In a preferred embodiment, the execution of the mapping module further includes: Each candidate segment is compared with the migration position and classification interval of the corresponding detection batch one by one. The boundary overlap length, center position difference, and gray-scale cumulative difference between each candidate segment and each classification interval are calculated. First, they are sorted in descending order by boundary overlap length, then in ascending order by center position difference, and finally in ascending order by gray-scale cumulative difference. The classification interval with the first position is determined as the classification interval corresponding to the candidate segment. When the same classification interval corresponds to multiple candidate segments, the candidate segment with the first position is retained, and the remaining candidate segments are moved out of the classification interval, thus outputting the initial classification record.

[0011] In a preferred embodiment, the execution of the mapping module further includes: The system reads the initial classification records, the band regions corresponding to interference region markers, and the band regions corresponding to abnormal region markers. For candidate segments corresponding to interference region markers, it performs merging or splitting processing based on the migration distance, boundary connection position, and gray-scale accumulation between the segment and its adjacent classification intervals. For candidate segments corresponding to abnormal region markers, it checks the corresponding time and position item by item according to the preprocessing state, electrophoresis state, and scanning state in the state chain. When the same candidate segment has an abnormal record with the same corresponding position in two or more states, the candidate segment is written as an abnormal classification segment and removed from the classification output. When the same candidate segment has an abnormal record with the same corresponding position in one state, the candidate segment is returned to the classification interval for comparison until there are no unprocessed candidate segments corresponding to interference region markers and candidate segments corresponding to abnormal region markers. Then, it outputs the lipoprotein classification records corresponding to each candidate segment.

[0012] In a preferred embodiment, the execution of the verification module includes: Read the lipoprotein classification records, state chains, and choleretic index data of the sample to be tested. Assign each lipoprotein classification record to the candidate segment that forms the lipoprotein classification record, the corresponding migration position in the reference image, and the corresponding state data in the state chain. Write the corresponding results into the same classification return sequence to output the return entry corresponding to the lipoprotein classification record. Read the candidate segments, corresponding migration positions, and criterion data from each return entry, calculate the proportion of the cumulative gray value of the candidate segment to the total cumulative gray value of each candidate segment, multiply the proportion by the criterion data to obtain the criterion value corresponding to each lipoprotein classification record, and then write each criterion value and the classification of the corresponding candidate segment into the record to output the lipoprotein classification criterion result. Read the lipoprotein classification and identification results and each return entry, check item by item whether the candidate segment corresponding to each lipoprotein classification and identification result exists, whether the corresponding migration position is unique, and whether the corresponding status data is complete, and calculate the difference between the sum of each classification and identification value and the identification index data. Write the lipoprotein classification and identification results within the predetermined range and where the candidate segment, corresponding migration position, and corresponding status data are all one-to-one as reliable records, and write the remaining lipoprotein classification and identification results as verification records.

[0013] In a preferred embodiment, the execution of the reporting module includes: Read the lipoprotein classification and betting results and the trusted records. Write the lipoprotein classification and betting results written as trusted records in the trusted records into the direct output results. Write the lipoprotein classification and betting results written as rescan records in the trusted records into the rescan prompts. Write the lipoprotein classification and betting results written as verification records in the trusted records into the verification prompts, and thus output the result set. Read the electrophoresis image, lipoprotein classification record and result set corresponding to the sample to be tested, write the corresponding position of the candidate segment in the lipoprotein classification record into the lipoprotein classification map, write the classification classification, classification value and record type in the result set into the result area and prompt area of ​​the test report respectively, and write the candidate segment, reference basis and status source in the lipoprotein classification record into the verification area of ​​the test report, thereby outputting the test report; Read the lipoprotein classification map, result area, prompt area, and verification area in the test report, and verify the corresponding sample identifier and test batch. Write the test reports with the same sample identifier and the same test batch as the output report, and write the test reports with different sample identifiers or different test batches as the review prompt.

[0014] The technical effects and advantages of this invention are as follows: By associating the preprocessing state, electrophoresis state, and scanning state with the corresponding band regions, and then mapping the results back to the candidate segments, reference criteria, and state sources after classification, the traceability and reliable discrimination ability of the basis for lipoprotein classification and determination results can be improved, thereby directly addressing the problem of difficulty in judging the results when electrophoresis images are disturbed by multiple factors. By introducing reference images, migration locations, and classification intervals from the same batch of detections, and performing position alignment, candidate segment division, and classification interval comparison on electrophoresis images accordingly, the impact of different detection batches and band position shifts on classification can be relatively reduced, thereby relatively improving the consistency of lipoprotein subtyping results. By performing incorporation, splitting, removal, and return comparison processes on the strip regions corresponding to the valid region marker, interference region marker, and abnormal region marker, the situation where local shadows, boundary intersections, and state missing directly enter the classification output can be relatively suppressed, thereby relatively improving the stability of automatic interpretation under abnormal map conditions. By calculating the classification and determination values ​​based on the cumulative grayscale values ​​of candidate segments and the determination index data, and performing consistency checks on the candidate segments, corresponding migration positions, corresponding status data, and total difference, a closed correspondence can be formed between the classification results and the determination results, which helps to output a verifiable lipoprotein classification test report. Attached Figure Description

[0015] Figure 1 This is a system block diagram of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Refer to the instruction manual appendix Figure 1 The present invention provides a lipoprotein classification and detection system based on electrophoretic image analysis, comprising: The intelligent sensing system includes a preprocessing unit, an electrophoresis unit, a scanning unit, and a control host. The output of the preprocessing unit is connected to the input of the electrophoresis unit, and the output of the electrophoresis unit is connected to the input of the scanning unit. The preprocessing unit, electrophoresis unit, and scanning unit are all connected to the control host to form a detection execution path and a status acquisition path for the same sample. In this implementation, the intelligent sensing system is used to synchronously associate the preprocessing, electrophoretic separation, and scanning imaging processes of the sample with the state data generated by the same sample at each execution time, so that the subsequently read electrophoretic images simultaneously carry band morphology information and process origin information; the preprocessing unit, electrophoresis unit, scanning unit, and control host sequentially execute detection tasks around the same sample identifier, and establish a correspondence between the detection execution path and the state acquisition path in the control host; this implementation process includes the following steps: The preprocessing unit outputs a sample ready for loading and generates a preprocessing status. Specifically, the control unit writes the sample identifier and batch number when the sample enters the preprocessing unit. The preprocessing unit centrifuges the sample, reads the liquid layer boundary, and takes a predetermined liquid layer as the loading liquid. It then filters the loading liquid to obtain the preprocessed sample. The preprocessing unit writes the centrifugation start time, centrifugation end time, filtration start time, filtration end time, and the volume of the filtered sample into the preprocessing status. If no identifiable liquid layer boundary is formed after centrifugation, or if the volume of the filtered sample is lower than the loading liquid, the preprocessing unit will not be considered a loading sample. If the sample volume is not specified, the sample is marked as having an abnormal preprocessing status and the sample delivery is stopped. For samples not marked as having an abnormal preprocessing status, the electrophoresis unit receives the preprocessed sample and forms an electrophoretic band with the corresponding sample identifier, while simultaneously writing the sample loading time, electrophoresis start time, electrophoresis end time, and lane identifier. The scanning unit scans the electrophoretic bands, outputs the electrophoretic image, and writes the scan start time, scan end time, lane position, and image storage address. The control host reads the preprocessing status, electrophoresis status, and scanning status, merges them according to sample identifier and detection batch, and generates a detection process record with the corresponding sample identifier. The control host is used to perform status marking on the band regions in the electrophoresis image based on the detection process record. Specifically, the control host first reads the electrophoresis image, extracts the band regions formed by grayscale changes along the lane direction, and numbers them sequentially according to the migration direction. Then, it reads the pre-processing state, electrophoresis state, and scanning state from the detection process record, associating the three states with the corresponding band regions according to the sample identifier and execution time. When the same band region corresponds to one pre-processing state, one electrophoresis state, and one scanning state, and the execution times of the three are arranged in the order of pre-processing, electrophoresis, and scanning, the band region is marked as a valid area. The image record contains three states: a preprocessing state, an electrophoresis state, and a scanning state. However, if two of these states are executed at the same time, in reverse order, or if the position range corresponding to either the electrophoresis state or the scanning state simultaneously covers the area of ​​the current strip and an adjacent strip, the strip is marked as an interference area. If the same strip lacks any of the preprocessing state, electrophoresis state, or scanning state, or if the same state record falls within two or more image ranges interspersed with other strip areas, the strip is marked as an abnormal area. The control host writes the strip area number, start pixel position, end pixel position, and marker type into the electrophoresis image record to obtain an electrophoresis image record with state markers. The control host is also used to establish a storage association between the detection process record and the electrophoresis image record. In specific execution, the control host matches the detection process record and the electrophoresis image record with the sample identifier as the primary matching field and the detection batch as the secondary matching field. Then, it checks whether the electrophoresis status and the lane position correspond with the lane identifier. If they correspond, the detection process record index is written into the electrophoresis image record, and the electrophoresis image record index is written into the detection process record, forming a correspondence between the detection execution path result and the status acquisition path result for the same sample. If the same sample identifier corresponds to multiple detection batches, or the same sample identifier in the same detection batch corresponds to multiple electrophoresis image records, then a set of records with a complete preprocessing, electrophoresis, and scanning sequence is retained, and the remaining records are moved into the set of records to be reviewed. After processing, the control host outputs the electrophoresis image record carrying the detection process record index, the strip region marking result, and the sample identifier for subsequent state chain module and mapping module to read. Through the above implementation process, the intelligent sensing system, while performing sample detection, stably associates the preprocessing state, electrophoresis state, and scanning state with the band regions in the electrophoresis image. This enables subsequent steps to distinguish the band regions corresponding to valid region markers, interference region markers, and abnormal region markers, avoiding the direct writing of process abnormalities into the lipoprotein classification results. In practical applications: after multiple samples from the same batch enter the preprocessing unit, the control host writes a sample identifier for each sample. The preprocessing unit completes centrifugation and filtering, the electrophoresis unit forms electrophoretic bands for each lane, and the scanning unit outputs the corresponding electrophoresis image. The control host then associates the preprocessing state, electrophoresis state, and scanning state with each band region one by one. Band regions that meet the sequential correspondence are written as valid region markers, band regions with reversed order or overlapping positions are written as interference region markers, and band regions lacking corresponding states or with one state spanning multiple non-adjacent band regions are written as abnormal region markers. Finally, an electrophoresis image record with state markers is output for direct reading in subsequent state chain generation and lipoprotein classification processing.

[0018] The reference component includes a quality control loading unit and a reference storage unit. The quality control loading unit works with the electrophoresis unit to form a reference lane for the same batch of tests. The reference storage unit is connected to the control host and is used to store the reference image, migration position and classification interval of the corresponding test batch. In this implementation, the reference component provides a unified reference for samples within the same testing batch, enabling subsequent mapping modules to use the reference image, migration position, and classification interval corresponding to that testing batch when performing position alignment, candidate segmentation, and classification comparison. The basic process is as follows: first, a reference swimlane is formed using quality control samples and a reference image is obtained; then, the migration position and classification interval of each band are extracted from the reference image; finally, the corresponding association and batch storage are completed according to the testing batch. This implementation process includes the following steps: The quality control loading unit is used to form a reference lane consistent with the corresponding test batch and output a reference image. In specific execution, the quality control loading unit receives the quality control sample of the corresponding test batch, writes the quality control sample into the test batch identifier, and sends it to the electrophoresis unit. The electrophoresis unit performs electrophoretic separation on the quality control sample to form a reference lane in the electrophoresis medium. The scanning unit performs scanning and image acquisition on the reference lane and outputs a reference image with the test batch identifier. The reference storage unit receives the reference image and writes it into the reference image record according to the test batch. If the same test batch corresponds to multiple reference images, the first reference image whose scanning end time is after the electrophoresis end time of the same batch is retained, and the remaining reference images are written into the reference image set to be reviewed. The control host extracts migration positions and classification intervals from the reference image, transforming the strip positions in the reference swimlane into readable classification criteria. Specifically, the control host reads the reference image, extracts a pixel-by-pixel grayscale sequence along the migration direction of the reference swimlane, writes the position where grayscale changes from increasing to decreasing as the start position, and the position where grayscale changes from decreasing to increasing as the end position, and writes the start and end positions into the migration positions of the corresponding stripes. Then, it reads each migration position according to the migration direction, sequentially writing the image segments between adjacent stripes and the image segments corresponding to each strip itself into the classification intervals. The reference storage unit receives the migration positions and classification intervals and stores them according to the detection batch. If no separable image segments are formed between two adjacent stripes, the control host writes the midpoint between them as the interval boundary position, and then generates the corresponding classification interval based on this interval boundary position. The control host is also used to close the reference image, migration position, and classification interval into a reference basis for the same detection batch. In specific execution, the control host reads the reference image record, migration position record, and classification interval record according to the detection batch and writes the three into the same reference index. The reference storage unit performs batch storage on the reference image, migration position, and classification interval under the same reference index and outputs the reference basis for the corresponding detection batch. If there are multiple sets of migration positions or classification intervals under the same detection batch, the set of records with the same number of stripes as the number of stripes identified in the reference image is retained, and the remaining records are written into the reference record set to be reviewed for subsequent manual verification or reconstruction of the reference basis. Through the above implementation process, the reference component stably converts the reference lanes formed by the quality control samples into reference images, migration positions, and classification intervals for the corresponding test batch. This ensures that subsequent steps have a unified positional basis when performing position alignment and classification comparison, avoiding classification shifts caused by directly mixing strip positions between different test batches. In practical applications: before the start of the same test batch, the quality control loading unit first sends the quality control samples into the electrophoresis unit to form reference lanes. The scanning unit outputs reference images, and the control host then extracts the start and end positions of each strip along the reference lanes to generate migration positions. The image segments corresponding to adjacent strips and each strip itself are written as classification intervals. Finally, the reference storage unit completes the same batch storage according to the test batch, obtaining a reference basis that can be used by all test samples in the test batch.

[0019] The state chain module is used to read the state data of the intelligent sensing system during the pretreatment, electrophoresis and scanning processes, and to match, sort and write the state data according to the sample identifier and detection batch, thereby outputting the state chain corresponding to the sample to be tested. In this implementation, the state chain module is used to organize the discrete state data generated by the intelligent sensing system during the preprocessing, electrophoresis, and scanning processes into a continuously readable state chain for the same sample. This allows the subsequent mapping and verification modules to read the state source according to a unified time sequence. The basic process is as follows: first, a unified field is extracted from the state data, and data corresponding to the same sample is filtered out. Then, an initial state chain is formed according to the execution time and process sequence. Subsequently, adjacent state data in the same process are processed by merging or segmenting. Finally, the state chain corresponding to the sample to be tested is obtained by splicing the data in the order of the preprocessing, electrophoresis, and scanning processes. This implementation process includes the following steps: The state chain module is first used to filter out the data to be processed corresponding to the same sample. Its mechanism is to first unify the state data fields, and then establish the belonging boundary by sample identifier and detection batch, so as to avoid the state data of different samples or different detection batches from being mixed in the same state chain. In specific execution, the state chain module reads the state data written by the intelligent sensing system, and extracts the sample identifier, detection batch, process category, execution time and state content from each state data. The process category is used to distinguish the pretreatment process, electrophoresis process and scanning process, and the state content is used to record the state item name and state item value in the corresponding process. Then, the state data is filtered according to the sample identifier and detection batch. The state data with the same sample identifier and the same detection batch are identified as the data to be processed corresponding to the same sample, and the data to be processed is written to the processing queue corresponding to the sample identifier. If a state data is missing any of the fields of sample identifier, detection batch, process category, execution time or state content, the state data is moved to the state data set to be reviewed and does not participate in the generation of the current state chain. The state chain module is then used to generate the initial state chain. Its mechanism is to first arrange the state data of the same sample into a unified time sequence, and then fix the process sequence relationship when the execution time is the same, so as to ensure that the subsequent fragment processing has a single sequential basis. In specific execution, the state chain module reads the data to be processed in the queue to be processed and arranges them in order of execution time. When the execution time of two state data is different, the order of execution time determines the position. When the execution time of two state data is the same, the order of preprocessing, electrophoresis and scanning processes determines the position. After sorting, the arranged state data is written into the state sequence of the corresponding sample identifier, and a start time field is written for each state data, where the start time is the execution time of the state data, so as to output the initial state chain. If there are more than two state data with the same process category at the same execution time, the state data with the earlier writing time is retained and the remaining state data is moved into the state data set to be reviewed. The state chain module further organizes the same process state data in the initial state chain into a process state fragment chain. Its mechanism involves merging unchanged state data into the same process state fragment, while retaining changed state data as adjacent process state fragments, thus providing fragment-level objects for connecting the beginning and end of subsequent state chains. Specifically, during execution, the state chain module reads two adjacent state data entries from the initial state chain, comparing their process categories and state content item by item. When both have the same process category and state content, the latter state data is merged into the former, and the execution time of the latter state data is written as the former's. The end time of a status data item; when two items have the same process category but different status content, the later status data item is written after the earlier status data item, and the end time of the earlier status data item is written as the execution time of the later status data item. At the same time, fields with the same status item name but different status item values ​​in the status content are written as changed content; when two items have different process categories, the order remains unchanged and they are written as different process status fragments into the process status fragment chain; if the end time of the earlier status data item has not yet been written and there is no subsequent status data item of the same process, the end time of the status data item is written as its own execution time. The state chain module is ultimately used to output the state chain corresponding to the sample to be tested. Its working mechanism is to screen out sequentially closed process state segments from the process state segment chain and connect them end-to-end according to the preprocessing process, electrophoresis process, and scanning process to avoid missing segments from entering subsequent calculations. In specific execution, the state chain module reads each process state segment in the process state segment chain in the order of preprocessing process, electrophoresis process, and scanning process. Process state segments that simultaneously have preprocessing process state segments, electrophoresis process state segments, and scanning process state segments, and whose end times and start times are in the order of preprocessing process before electrophoresis process and electrophoresis process before scanning process, are determined to be sequentially complete process state segments. Then, sequentially complete process state segments are connected end-to-end in the above order to generate the state chain corresponding to the sample to be tested, and this state chain is written into the state chain record corresponding to the sample identifier. Process state segments that lack preceding or following process state segments are removed from the state chain and written into the set of state segments to be reviewed, and do not participate in the mapping and verification processing of the current sample to be tested. Through the above implementation process, the state chain module organizes the originally scattered preprocessing state, electrophoresis state and scanning state into a state chain with unified fields, unified order and unified fragments, so that subsequent steps do not need to re-filter and sort the original state data when reading the state source, and also avoids different process states from being mixed or missing fragments directly entering the mapping calculation. In practical applications: the same sample to be tested forms a pre-processing state after centrifugation and filtration in the pre-processing unit, an electrophoresis state after loading and migration in the electrophoresis unit, and a scanning state after image acquisition in the scanning unit. The state chain module first filters out three types of state data based on sample identification and detection batch, and then forms an initial state chain according to execution time and process sequence. Subsequently, state data with consistent state content are merged into the same process state segment, and state data with changing state content are retained as adjacent process state segments. Finally, only process state segments with complete sequence of pre-processing, electrophoresis, and scanning processes are connected end to end to form a state chain, which is then used by the subsequent mapping module to read the corresponding state source.

[0020] The mapping module is used to read the electrophoresis image, state chain and reference image of the corresponding detection batch of the sample to be tested, perform position alignment on the electrophoresis image according to the migration position in the reference image, divide the aligned band image into multiple candidate segments, and then compare each candidate segment with the classification interval one by one, thereby outputting the lipoprotein classification record corresponding to each candidate segment; In this implementation, the mapping module is used to convert the electrophoresis image of the sample to be tested into a lipoprotein classification record that can be used for classification output. The basic idea is as follows: first, a unified migration coordinate system is established using the reference image of the corresponding batch; then, the band regions in the electrophoresis image are filtered and aligned using a state chain; subsequently, segment generation, classification assignment, and anomaly rollback are completed, so that the classification results entering the subsequent verification module simultaneously have location, reference, and state criteria. This implementation process includes the following steps: The mapping module first aligns the electrophoresis image of the sample to be tested with the reference image. Its mechanism involves projecting the effective band regions in the electrophoresis image of the sample to be tested onto the reference migration coordinates of the corresponding detection batch, and then performing subsequent segmentation under the same migration coordinates. Specifically, the mapping module reads the electrophoresis image, state chain, and reference image of the corresponding detection batch of the sample to be tested. It extracts the band regions corresponding to the effective region markers, interference region markers, and abnormal region markers from the electrophoresis image, where the band regions are arranged along the migration direction of the swim lanes. Then, it extracts the migration positions of each band along the reference swim lanes of the reference image and aligns them with the effective region markers. The alignment module extracts the positions to be aligned from the corresponding strip regions. The alignment positions are the start position, end position, and grayscale peak position of the strip region. The mapping module constructs a position correspondence scheme between each alignment position and each migration position, keeping the migration direction order unchanged. For each position correspondence scheme, the sum of position differences is calculated, and the position correspondence scheme with the first-ranked sum of position differences is taken as the position alignment result. Then, the electrophoresis image of the sample to be tested is mapped to the reference migration coordinates according to the position alignment result, and the aligned strip image is output. If there are fewer than two strip regions corresponding to the effective area marker, the position alignment of that lane is not performed, and it is directly written into the image set to be reviewed. The mapping module is then used to generate candidate segments from the aligned strip image. Its mechanism involves first extracting segments eligible for classification from the strip regions corresponding to the valid region markers, and then correcting the segment boundaries using interference and anomaly region markers. Specifically, the mapping module performs pixel-by-pixel gray-level integration on the aligned strip image along the migration direction. In the strip regions corresponding to the valid region markers, the position where the gray-level integration changes from increasing to decreasing is written as the starting boundary, and the position where the gray-level integration changes from decreasing to increasing is written as the ending boundary. The image segment between adjacent starting and ending boundaries is written as the initial segment. For the strip regions corresponding to the interference region markers, the distance between the preceding adjacent segments is calculated. The boundary spacing between the initial segment and the adjacent initial segment is calculated, and the direction of grayscale change at the boundary is read. When the side with the shorter boundary spacing has the same direction of grayscale change as the strip region corresponding to the interference area marker, the strip region corresponding to the interference area marker is merged into the initial segment on that side. When the boundary spacing on both sides is the same, the difference in the cumulative grayscale amount on both sides is compared, and the side with the smaller cumulative grayscale amount difference is taken as the merged side. For the strip region corresponding to the abnormal area marker, it is directly removed from the segment division and written into the abnormal segment set, and does not participate in the current candidate segment generation. After completing the above processing, the candidate segments are output, and the start position, end position, and lane of each candidate segment are written into the candidate segment record. The mapping module is further used to assign candidate segments to the corresponding classification intervals of the test batch. Its mechanism involves comparing and adjudicating candidate segments one by one using the migration positions and classification intervals provided by the reference image to obtain initial classification records. Specifically, the mapping module reads the migration positions and classification intervals of the corresponding test batch, where the classification intervals are the migration position ranges corresponding to each lipoprotein category in the reference image. Then, each candidate segment is projected onto the reference migration coordinates and compared one by one with each classification interval, calculating the overlap length between the candidate segment and each classification interval. The center position difference and the cumulative gray level difference are calculated. After calculation, the segments are first sorted in descending order by boundary overlap length, then in ascending order by center position difference, and finally in ascending order by cumulative gray level difference. The first sorted segment is determined as the corresponding segment. When the same segment corresponds to multiple segments, the first sorted segment is retained, and the remaining segments are moved to the set of segments to be reassigned. The retained segment, the corresponding segment, the corresponding migration position, and the sample identifier are then written into the initial classification record for subsequent error rollback steps. The mapping module is ultimately used to perform interference correction and anomaly rollback on the initial classification records. Its mechanism involves continuing to process candidate segments that have not yet been stably assigned until a lipoprotein classification record that can be directly output is formed. Specifically, the mapping module reads the initial classification records, the band regions corresponding to interference region markers, and the band regions corresponding to anomaly region markers. For candidate segments corresponding to interference region markers, it calculates the migration distance, boundary connection position, and cumulative grayscale value between the candidate segment and its adjacent classification intervals. When the boundary connection position of the candidate segment coincides with that of an adjacent classification interval, it performs incorporation processing. When the candidate segment simultaneously crosses two classification intervals, it performs splitting processing according to the classification interval boundaries, and the processing results are rewritten into the initial classification records. For candidate segments corresponding to anomaly region markers... The selection process involves the mapping module reading the preprocessing, electrophoresis, and scanning states from the state chain, and verifying the corresponding time and position of each candidate segment. If an abnormal record with overlapping positions exists in more than two states of the same candidate segment, the candidate segment is written as an abnormal classification segment and removed from the classification output. If an abnormal record with overlapping positions exists in only one state, the candidate segment is returned to the classification interval for comparison. After each round of incorporation processing, split processing, or return comparison processing is completed, the system re-checks whether there are still unprocessed records in the candidate segments corresponding to the interference region marker and the candidate segments corresponding to the abnormal region marker. When there are no unprocessed records, the system outputs the lipoprotein classification records corresponding to each candidate segment and writes the abnormal classification segments into the classification set to be reviewed. Through the above implementation process, the mapping module stably converts the electrophoresis image with state labels into lipoprotein classification records, so that the classification results no longer depend solely on the band morphology itself, but are simultaneously constrained by the reference image, migration position, classification interval, and state chain, thereby reducing the direct impact of interference region labels and abnormal region labels on the classification output. In practical applications: After aligning the electrophoresis image of a sample to be tested, the mapping module first extracts the alignment position from the band region corresponding to the effective region marker, and then unifies the migration coordinates according to the reference image of the corresponding detection batch; subsequently, candidate segments are generated from the aligned band image, and the band region corresponding to the interference region marker is merged into the adjacent initial segment, while the band region corresponding to the abnormal region marker is removed from the segment division; then, the classification interval of each candidate segment is determined according to the boundary overlap length, the difference in center position, and the difference in cumulative gray level; finally, the candidate segments corresponding to the interference region marker are merged or split according to the state chain, and the candidate segments corresponding to the abnormal region marker are reversed or removed according to the abnormal region marker, to obtain the lipoprotein classification record that can be directly read by the verification module.

[0021] The verification module is used to read lipoprotein classification records, status chains, and bile index data of the sample to be tested. It maps each lipoprotein classification record to the band segment that formed the classification record, the reference basis, and the status source, and performs consistency verification between each lipoprotein classification record and bile index data, thereby outputting lipoprotein classification bile determination results and reliable records. In this implementation process, the verification module is used to restore the lipoprotein classification record to the image basis, reference basis, and status basis that formed the record. Based on this, it completes the calculation of the classification and determination value and the reliability verification of the results. This ensures that the lipoprotein classification and determination results read by the subsequent reporting module not only have classification attribution but also a verifiable generation path. The basic process is as follows: first, a verification entry is created for each lipoprotein classification record; then, the classification and determination value is calculated based on the cumulative grayscale value of the candidate segment and the determination index data; finally, the candidate segment, corresponding migration position, corresponding status data, and total difference are verified to form a reliable record or a verification record. This implementation process includes the following steps: The verification module first establishes verification entries for lipoprotein classification records. Its mechanism is to bind the classification result with the segment, position, and state that formed the result, preventing subsequent cholesterol value calculations from deviating from the source basis. In specific execution, the verification module reads the lipoprotein classification record, state chain, and cholesterol index data of the sample to be tested. The cholesterol index data is the test result of the cholesterol-related test item corresponding to the sample to be tested, preferably the total cholesterol test value of the sample to be tested. Then, according to the sample identifier, it reads the candidate segment corresponding to each lipoprotein classification record and the corresponding migration position in the reference image, and extracts the corresponding state data from the state chain that is closest to the formation time of the candidate segment and belongs to the pretreatment process, electrophoresis process, and scanning process, respectively. The verification module writes the lipoprotein classification record, candidate segment, corresponding migration position, corresponding state data, and cholesterol index data into the same classification verification sequence to generate the verification entry corresponding to the lipoprotein classification record. If a lipoprotein classification record lacks a candidate segment or a corresponding migration position, the lipoprotein classification record will not enter the classification verification sequence and will be directly written into the set of records to be reviewed. The verification module then calculates the classification value corresponding to each lipoprotein classification record. Its mechanism involves converting the grayscale proportion of candidate segments in all classification output segments into a distribution ratio of the classification index data. Specifically, the verification module reads the candidate segments, corresponding migration positions, and classification index data from each verification entry. It sums the pixel grayscale values ​​of each candidate segment from its start position to its end position to obtain the cumulative grayscale value of that candidate segment. Then, it adds the cumulative grayscale values ​​of all candidate segments entering the current classification output to obtain the total cumulative grayscale value, where the values ​​of segments that have been moved out are included. Abnormal classification segments in the class output are not included in the total gray-scale accumulation. Then, the cumulative gray-scale of each candidate segment is divided by the total cumulative gray-scale to obtain the proportion value of the candidate segment. This proportion value is then multiplied by the criterion index data to obtain the classification criterion value corresponding to the lipoprotein classification record. After the calculation is completed, the verification module writes the classification criterion value, the classification of the corresponding candidate segment, and the sample identifier into the lipoprotein classification criterion result. If the total cumulative gray-scale is zero, the calculation of the classification criterion value for this round is stopped, and all lipoprotein classification records corresponding to the sample identifier are written into the verification record set. The verification module is ultimately used to perform reliability checks on the lipoprotein classification and cholesterol determination results. Its mechanism involves simultaneously verifying the integrity of the source and the closure of the numerical values, only recording results where both the source and total amount are closed as reliable records. Specifically, the verification module reads the lipoprotein classification and cholesterol determination results and each verification entry, checking item by item whether the candidate segment corresponding to each result exists, whether the corresponding migration position is unique, and whether the corresponding status data is complete. Complete status data means that the candidate segment corresponding to the lipoprotein classification and cholesterol determination result simultaneously has a pre-processing status, an electrophoresis status, and a scanning status. Then, all classification and cholesterol determination values ​​are added together to obtain the sum of the classification and cholesterol determination values, and then subtracted from the cholesterol determination index data. The sum of the classification and determination values ​​is used to obtain the total difference. The total difference is then divided by the determination index data to obtain the difference ratio. In this implementation, the predetermined range adopts the allowable difference ratio range written in the preset configuration of the detection system. The allowable difference ratio range is obtained by statistically analyzing the difference ratio distribution formed by repeated testing of quality control samples during the trial production stage, and is written into the configuration table of the corresponding test item by the control host. When the difference ratio falls into the allowable difference ratio range, and the candidate segment, the corresponding migration position, and the corresponding status data are all one-to-one, the lipoprotein classification and determination result is written as a reliable record. The remaining lipoprotein classification and determination results are written as verification records, and the corresponding sample identifier is written into the set of results to be verified for subsequent reporting module reading. Through the above implementation process, the verification module transforms lipoprotein classification records into verifiable results with candidate segments, corresponding migration positions, corresponding state data, and classification cholesterol values. It completes the screening of reliable records through dual checks of source integrity and total closure, ensuring that the cholesterol values ​​in subsequent test reports not only correspond to specific lipoprotein classifications but also allow for tracing back to their formation basis. In practical applications: after outputting lipoprotein classification records for a test sample, the verification module first maps each lipoprotein classification record to a candidate segment, the corresponding migration position in the reference image, and the preprocessing state, electrophoresis state, and scanning state in the state chain. Then, it allocates total cholesterol detection values ​​based on the proportion of the cumulative grayscale value of each candidate segment to the total cumulative grayscale value of all classification output segments, obtaining the classification cholesterol value corresponding to each lipoprotein classification record. Finally, it checks whether the candidate segments, corresponding migration positions, and corresponding state data correspond one-to-one, and performs a difference ratio check between the sum of the classification cholesterol values ​​and the total cholesterol detection value. Results that meet the allowable difference ratio range are written as reliable records, and the remaining results are written as verification records.

[0022] The report module is used to read lipoprotein classification and gallbladder determination results and reliable records. According to the reliable records, it directly outputs the results, rescan prompts and review prompts, and generates a test report containing lipoprotein classification map, lipoprotein classification and gallbladder determination results and verification content. In this implementation process, the reporting module is used to organize the lipoprotein classification and bile determination results, reliable records, and feedback content into a test report that can be directly output or prompts for review, ensuring that the test results, prompts, and basis for their formation correspond in a closed loop under the same sample and the same test batch. The basic process is as follows: first, the lipoprotein classification and bile determination results are sorted according to the reliable records; then, the electrophoresis images, lipoprotein classification records, and result sets are written into a unified report structure; finally, the sample identification and test batch in the test report are checked for consistency to determine whether to output a report or prompt for review. This implementation process includes the following steps: The report module is first used to generate a result set. Its mechanism is to first sort the lipoprotein classification and gallbladder determination results according to the record type to avoid mixing different processing conclusions in the report output stage. In specific execution, the report module reads the lipoprotein classification and gallbladder determination results and the reliable records, and checks the sample identification, test batch and classification attribution of each record. For records that match, the module reads the record type in the reliable record, writes the lipoprotein classification and gallbladder determination results written as reliable records into the direct output results, writes the lipoprotein classification and gallbladder determination results written as rescan records into the rescan prompt, writes the lipoprotein classification and gallbladder determination results written as verification records into the verification prompt, and writes the classification attribution, classification and gallbladder determination value, record type and corresponding sample identification into the result set. If the lipoprotein classification and gallbladder determination results are inconsistent with the reliable records in terms of sample identification, test batch or classification attribution, then the record is not included in the result set and is directly written into the verification prompt set. The report module is then used to generate the test report. Its mechanism is to write the image results, numerical results, and verification content into the same report structure, so that subsequent verification and output can be completed on a unified carrier. In specific execution, the report module reads the electrophoresis image, lipoprotein classification record, and result set corresponding to the sample to be tested. It plots the corresponding positions of the candidate segments in the lipoprotein classification record onto the electrophoresis image to generate a lipoprotein classification map. Then, it writes the classification, classification value, and record type in the result set into the result area and prompt area of ​​the test report, respectively. At the same time, it reads the candidate segments, reference basis, and status source in the lipoprotein classification record and writes the candidate segment position, corresponding migration position, and corresponding status data into the verification area of ​​the test report. After writing is completed, the sample identifier and test batch are simultaneously written into the lipoprotein classification map, result area, prompt area, and verification area, and the test report is output. If the result set is empty, the test report is still generated, but a review prompt is written in the prompt area and direct output is stopped. The report module is ultimately used to determine the output status of the test report. Its mechanism involves verifying the consistency of sample identifiers and test batches to prevent content from different samples or test batches from being included in the same test report. Specifically, the report module reads the lipoprotein classification map, result area, prompt area, and verification area from the test report, extracts the sample identifiers and test batches written therein, and compares them item by item. When the sample identifiers and test batches in the lipoprotein classification map, result area, prompt area, and verification area are the same, the test report is written as an output report and written to the report storage area. If any of the sample identifiers or test batches are different, a verification prompt is written to the test report, and the corresponding sample identifier is written to the set of reports to be verified. If a test report is prompted for verification, it is not included in the output report set. Through the above implementation process, the reporting module organizes the lipoprotein classification and determination results, prompts, and verification content into a test report with unified fields and closed sources. This ensures that the direct output results, rescan prompts, and verification prompts are placed separately under the same sample and the same test batch, and avoids mismatches between the graphs, values, and verification content. In practical application: after completing the lipoprotein classification and determination results and reliable record output for a certain test sample, the reporting module first writes the results written as reliable records into the direct output results, writes the results written as rescan records into the rescan prompts, and writes the results written as verification records into the verification prompts. Then, it plots the corresponding positions of the candidate segments as lipoprotein classification graphs and writes the classification, determination value, record type, candidate segment, reference basis, and status source into the test report. Finally, it checks the sample identification and test batch in the lipoprotein classification graph, result area, prompt area, and verification area. Test reports that match are written as the output report, and test reports that do not match are written as verification prompts.

[0023] Working Principle: This solution first centrifuges and filters the sample to be tested to obtain a sample-ready solution. Then, electrophoresis is used to form electrophoretic bands corresponding to the sample identifiers, and the electrophoretic bands are scanned and imaged. Simultaneously, the system organizes the state data generated in the three stages of pretreatment, electrophoresis, and scanning into a state chain according to the sample identifier and the test batch. The system also uses quality control samples to form a reference image, migration position, and classification interval for the same batch of tests. Subsequently, the system marks the effective area, interference area, and abnormal area of ​​the band region in the electrophoretic image according to the state chain. Then, using the reference image as a reference, the system completes position alignment, candidate segment division, and classification interval comparison to obtain lipoprotein classification records. On this basis, the system then maps each classification record back to the candidate segment, reference basis, and state source. Combined with the bile index data, the system calculates the classification bile value of each lipoprotein category and verifies whether the source of the results is complete and whether the total amount is closed. Finally, a test report containing classification map, classification bile value, and verification content is generated. For example, when performing lipoprotein classification tests on a batch of serum samples in a hospital laboratory, the system first performs pretreatment on each sample and assigns a sample label. Then, within the same batch, it simultaneously establishes quality control reference lanes and lanes for the samples to be tested. After scanning, the system not only sees what the bands look like, but also knows the pretreatment, electrophoresis, and scanning states in which these bands were formed. Thus, when a sample band is clear and its position is stable, the system will directly complete the classification and output the results. When a sample band has intersections, shadows, or missing states, the system will identify the corresponding band as an interference or abnormal area and continue to perform merging, splitting, or rollback verification, rather than directly outputting it as a normal classification result. The final report not only gives the result values ​​for each lipoprotein category, but also explains which band segments, reference positions, and process states these results were based on, making it more suitable for use in batch testing scenarios.

[0024] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A lipoprotein classification detection system based on electrophoretic image analysis, characterized by, include: The intelligent sensing system includes a preprocessing unit, an electrophoresis unit, a scanning unit, and a control host. The output of the preprocessing unit is connected to the input of the electrophoresis unit, and the output of the electrophoresis unit is connected to the input of the scanning unit. The preprocessing unit, electrophoresis unit, and scanning unit are all connected to the control host to form a detection execution path and a status acquisition path for the same sample. The reference component includes a quality control loading unit and a reference storage unit. The quality control loading unit works with the electrophoresis unit to form a reference lane for the same batch of tests. The reference storage unit is connected to the control host and is used to store the reference image, migration position and classification interval of the corresponding test batch. The state chain module is used to read the state data of the intelligent sensing system during the pretreatment, electrophoresis and scanning processes, and to match, sort and write the state data according to the sample identifier and detection batch, thereby outputting the state chain corresponding to the sample to be tested. The mapping module is used to read the electrophoresis image, state chain and reference image of the corresponding detection batch of the sample to be tested, perform position alignment on the electrophoresis image according to the migration position in the reference image, divide the aligned band image into multiple candidate segments, and then compare each candidate segment with the classification interval one by one, thereby outputting the lipoprotein classification record corresponding to each candidate segment; The verification module is used to read lipoprotein classification records, status chains, and bile index data of the sample to be tested. It maps each lipoprotein classification record to the band segment that formed the classification record, the reference basis, and the status source, and performs consistency verification between each lipoprotein classification record and bile index data, thereby outputting lipoprotein classification bile determination results and reliable records. The reporting module is used to read lipoprotein classification and determination results and reliable records. Based on the reliable records, it directly outputs the results, prompts for rescanning, and prompts for verification, and generates a test report containing lipoprotein classification map, lipoprotein classification and determination results, and verification information.

2. The lipoprotein classification and detection system based on electrophoretic image analysis according to claim 1, characterized in that: The execution of the intelligent sensing system includes: The preprocessing unit receives the sample to be tested and performs centrifugation and filtration. The electrophoresis unit receives the preprocessed sample to be tested and forms an electrophoretic band with a corresponding sample identifier. The scanning unit scans and captures images of the electrophoretic band. The control host receives the status data written by the preprocessing unit, electrophoresis unit and scanning unit at the corresponding execution time and generates a detection process record with a corresponding sample identifier. The control host matches the detection process record with the electrophoresis image output by the scanning unit. Based on the sample identifier and execution time, it associates the preprocessing state, electrophoresis state, and scanning state with the corresponding strip regions. When the same strip region corresponds to one preprocessing state, one electrophoresis state, and one scanning state, and the execution times of the three are arranged in the order of preprocessing, electrophoresis, and scanning, the strip region is marked as a valid area. When the same strip region corresponds to one preprocessing state, one electrophoresis state, and one scanning state, but two of the three have the same execution time, are in reverse order, or their corresponding ranges overlap with adjacent strip regions, the strip region is marked as an interference area. When the same strip region lacks any of the preprocessing state, electrophoresis state, or scanning state, or when the same state corresponds to two or more non-adjacent strip regions, the strip region is marked as an abnormal area, thus outputting an electrophoresis image record with status markings. The control host associates the detection process record with the electrophoretic image output by the scanning unit and stores the electrophoretic image record, thereby forming a correspondence between the detection execution path result and the status acquisition path result of the same sample.

3. The lipoprotein classification and detection system based on electrophoretic image analysis according to claim 2, characterized in that: The execution of the reference component includes: The quality control loading unit receives the quality control samples of the corresponding test batch and sends the quality control samples into the electrophoresis unit. The electrophoresis unit performs electrophoretic separation on the quality control samples and forms a reference lane. The scanning unit performs scanning and image acquisition on the reference lane. The reference storage unit receives and stores the reference image of the corresponding test batch. The control host reads the reference image, identifies the start and end positions of each band along the migration direction of the reference swimlane, writes the start and end positions of each band as the migration position of the corresponding band, and writes the image segments between adjacent bands and the corresponding segments of each band as the corresponding classification intervals. The reference storage unit receives and stores the migration positions and classification intervals. The control host will associate the reference image, migration location, and classification range according to the detection batch. The reference storage unit will perform batch storage on the associated reference image, migration location, and classification range, thereby outputting the reference basis for the corresponding detection batch.

4. The lipoprotein classification and detection system based on electrophoretic image analysis according to claim 3, characterized in that: The execution of the state chain module includes: Read the status data written by the intelligent sensing system during the pretreatment, electrophoresis and scanning processes, extract the sample identifier, detection batch, process category, execution time and status content from each status data, and identify the status data with the same sample identifier and the same detection batch as the data to be processed corresponding to the same sample. The data to be processed is arranged in order of execution time from first to last. When the execution time of two state data is the same, the order of preprocessing, electrophoresis and scanning is determined. The arranged state data is then written into the state sequence of the corresponding sample identifier to output the initial state chain.

5. The lipoprotein classification and detection system based on electrophoretic image analysis according to claim 4, characterized in that: The execution of the state chain module also includes: The process state fragment chain is output by comparing two adjacent state data in the initial state chain item by item. When the process categories and state contents are the same, the latter state data is merged into the former state data and the end time of the former state data is updated at the execution time of the latter state data. When the process categories and state contents are different, the latter state data is written after the former state data and the changes between the two are recorded. Read each process state segment in the process state segment chain in the order of pretreatment, electrophoresis and scanning. Connect the first and last process state segments in the complete sequence to form the state chain corresponding to the sample to be tested. Remove process state segments that are missing preceding or following process state segments from the state chain.

6. The lipoprotein classification and detection system based on electrophoretic image analysis according to claim 5, characterized in that: The execution of the mapping module includes: Read the electrophoresis image, state chain, and reference image of the corresponding detection batch of the sample to be tested. Extract the strip regions corresponding to the effective area marker, interference area marker, and abnormal area marker from the electrophoresis image. Extract the migration position in the reference image along the migration direction. Extract the position to be aligned in the strip region corresponding to the effective area marker. Calculate the position difference between each position to be aligned and each migration position. Perform position alignment according to the position correspondence relationship of the consistent migration direction order and the first position in the total position difference sorted first. Then output the aligned strip image. Pixel-wise grayscale integration is performed on the aligned strip image along the migration direction. In the strip region corresponding to the effective region, the position where the grayscale integration changes from rising to falling is written as the starting boundary, and the position where the grayscale integration changes from falling to rising is written as the ending boundary. The image segment between the adjacent starting boundary and ending boundary is written as the initial segment. Then, the boundary distance and grayscale continuity order between the strip region corresponding to the interference region and the adjacent initial segments are calculated. The strip region corresponding to the interference region with the highest boundary distance and the same grayscale continuity order is merged into the initial segment on the corresponding side. The strip region corresponding to the abnormal region is removed from the segment division, thus outputting the candidate segment.

7. The lipoprotein classification and detection system based on electrophoretic image analysis according to claim 6, characterized in that: The execution of the mapping module also includes: Each candidate segment is compared with the migration position and classification interval of the corresponding detection batch one by one. The boundary overlap length, center position difference, and gray-scale cumulative difference between each candidate segment and each classification interval are calculated. First, they are sorted in descending order by boundary overlap length, then in ascending order by center position difference, and finally in ascending order by gray-scale cumulative difference. The classification interval with the first position is determined as the classification interval corresponding to the candidate segment. When the same classification interval corresponds to multiple candidate segments, the candidate segment with the first position is retained, and the remaining candidate segments are moved out of the classification interval, thus outputting the initial classification record.

8. The lipoprotein classification and detection system based on electrophoretic image analysis according to claim 7, characterized in that: The execution of the mapping module also includes: The system reads the initial classification records, the band regions corresponding to interference region markers, and the band regions corresponding to abnormal region markers. For candidate segments corresponding to interference region markers, it performs merging or splitting processing based on the migration distance, boundary connection position, and gray-scale accumulation between the segment and its adjacent classification intervals. For candidate segments corresponding to abnormal region markers, it checks the corresponding time and position item by item according to the preprocessing state, electrophoresis state, and scanning state in the state chain. When the same candidate segment has an abnormal record with the same corresponding position in two or more states, the candidate segment is written as an abnormal classification segment and removed from the classification output. When the same candidate segment has an abnormal record with the same corresponding position in one state, the candidate segment is returned to the classification interval for comparison until there are no unprocessed candidate segments corresponding to interference region markers and candidate segments corresponding to abnormal region markers. Then, it outputs the lipoprotein classification records corresponding to each candidate segment.

9. A lipoprotein classification and detection system based on electrophoretic image analysis according to claim 8, characterized in that: The execution of the verification module includes: Read the lipoprotein classification records, state chains, and choleretic index data of the sample to be tested. Assign each lipoprotein classification record to the candidate segment that forms the lipoprotein classification record, the corresponding migration position in the reference image, and the corresponding state data in the state chain. Write the corresponding results into the same classification return sequence to output the return entry corresponding to the lipoprotein classification record. Read the candidate segments, corresponding migration positions, and criterion data from each return entry, calculate the proportion of the cumulative gray value of the candidate segment to the total cumulative gray value of each candidate segment, multiply the proportion by the criterion data to obtain the criterion value corresponding to each lipoprotein classification record, and then write each criterion value and the classification of the corresponding candidate segment into the record to output the lipoprotein classification criterion result. Read the lipoprotein classification and identification results and each return entry, check item by item whether the candidate segment corresponding to each lipoprotein classification and identification result exists, whether the corresponding migration position is unique, and whether the corresponding status data is complete, and calculate the difference between the sum of each classification and identification value and the identification index data. Write the lipoprotein classification and identification results within the predetermined range and where the candidate segment, corresponding migration position, and corresponding status data are all one-to-one as reliable records, and write the remaining lipoprotein classification and identification results as verification records.

10. A lipoprotein classification and detection system based on electrophoretic image analysis according to claim 9, characterized in that: The execution of the reporting module includes: Read the lipoprotein classification and betting results and the trusted records. Write the lipoprotein classification and betting results written as trusted records in the trusted records into the direct output results. Write the lipoprotein classification and betting results written as rescan records in the trusted records into the rescan prompts. Write the lipoprotein classification and betting results written as verification records in the trusted records into the verification prompts, and thus output the result set. Read the electrophoresis image, lipoprotein classification record and result set corresponding to the sample to be tested, write the corresponding position of the candidate segment in the lipoprotein classification record into the lipoprotein classification map, write the classification classification, classification value and record type in the result set into the result area and prompt area of ​​the test report respectively, and write the candidate segment, reference basis and status source in the lipoprotein classification record into the verification area of ​​the test report, thereby outputting the test report; Read the lipoprotein classification map, result area, prompt area, and verification area in the test report, and verify the corresponding sample identifier and test batch. Write the test reports with the same sample identifier and the same test batch as the output report, and write the test reports with different sample identifiers or different test batches as the review prompt.

Images