CN-122019810-A - Serum quality detection method through image parameterization and database comparison

CN122019810ACN 122019810 ACN122019810 ACN 122019810ACN-122019810-A

Abstract

The invention belongs to the field of clinical medicine and discloses a serum quality detection method by image parameterization and database comparison, which comprises the following steps of S1, randomly selecting an effective serum picture, extracting serum pixel points and storing RGB data, respectively establishing four sub-databases of a comprehensive serum pixel point database and normal serum, hemolytic serum, lipidemia serum and jaundice serum pixel point data, optimizing the RGB data into three intervals according to the RGB data of the pixel points in the four sub-databases, and forming a color cube in three dimensions, S2, shooting a plurality of pictures of serum to be detected, comparing the RGB data of the pixel points of each serum picture with the comprehensive serum pixel point database, wherein the most matching number is used as a serum window optimal picture for serum quality analysis, and S3, comparing the RGB data of the pixel points of the serum window optimal picture with the four sub-databases, wherein the corresponding serum quality of the sub-database with the most matching number of the pixel points is the quality of the serum to be detected.

Inventors

WANG MINGYANG
LI ZHUOHAN
GAO YUHUA

Assignees

郑州大学第一附属医院

Dates

Publication Date: 20260512
Application Date: 20251211

Claims (10)

1. A method for detecting serum quality by image parameterization and database comparison, comprising the steps of: S1, randomly selecting an effective serum picture containing normal serum, hemolyzed serum, lipidemia serum and jaundice serum, extracting serum pixel points, storing R data, G data and B data of the serum pixel points, respectively establishing four sub-databases of a comprehensive serum pixel point database and normal serum, hemolyzed serum, lipidemia serum and jaundice serum pixel point data, respectively extracting R data, G data and B data of the pixel points in the four sub-databases, solving target values and standard deviations of the R data, the G data and the B data, obtaining intervals of the R data, the G data and the B data, optimizing the range of the pixel points contained in the database into three intervals, and forming a color cube in three dimensions of the database; S2, serum picture analysis, namely shooting a plurality of pictures of serum to be detected, comparing R data, G data and B data of each serum picture pixel point with a comprehensive serum pixel point database, respectively recording the matching number of the serum pixel points, selecting one picture with the largest matching number as a serum window optimal picture, and outputting and using the picture for serum quality analysis; And S3, serum quality analysis, namely comparing R data, G data and B data of the pixel point of the optimal picture of the serum window with R data, G data and B data of four sub-databases, wherein the serum quality corresponding to the sub-database with the largest matching number of the pixel point of the picture is the quality output result of the serum to be detected.
2. The serum quality detection method according to claim 1, wherein in step S3, when R data, G data, and B data of a pixel point on the optimal picture of the serum window are all in a section of R data, G data, and B data of a certain database, the R data, the G data, and the B data are matched with the database.
3. The method according to claim 1, wherein when the number of serum pixels of the serum window optimal picture outputted by the serum picture auxiliary analysis module is greater than 200, the serum pixel is used as an effective mark when the serum picture is analyzed in step S2.
4. The serum quality detection method according to claim 1, wherein the serum picture analysis is performed by a serum picture auxiliary analysis module in the step S2, and the development process of the serum picture auxiliary analysis module is that the serum picture auxiliary analysis module is developed by a K-NN algorithm based on comparison of four given serum picture pixel points and the serum pixel point database in the step S1.
5. The serum quality detection method according to claim 4, wherein the step S3 is characterized in that a serum quality auxiliary analysis module is adopted to carry out serum quality analysis, a new set of serum pictures is provided in the development process of the serum quality auxiliary analysis module, a serum window optimal picture is selected based on the serum picture auxiliary analysis module, the selected serum window optimal picture pixel point is compared with four sub-databases of a serum pixel point database, and a K-NN algorithm is adopted to develop the serum quality detection method.
6. The method according to claim 1, wherein the serum type to which the serum picture belongs is determined by clinical judgment in step S1.
7. The method of claim 6, wherein the serum picture used to create the database of serum pixels in step S1 is a pre-processed picture taken by a TLA photographic system.
8. The serum quality detection system is characterized by comprising a TLA photographic system, a serum picture auxiliary analysis module and a serum quality auxiliary analysis module; the TLA photographing system is used for photographing serum to be tested and obtaining pictures of the serum to be tested; The serum picture auxiliary analysis module takes a serum picture shot by the TLA photographic system as an input value, compares pixel point R data, G data and B data of the serum picture with a comprehensive serum pixel point database, and outputs a serum window optimal picture; The serum quality auxiliary analysis module takes the optimal picture of the serum window as an input value, compares the pixel point R data, the G data and the B data of the optimal picture of the serum window with four sub-databases, and outputs a serum quality result.
9. An electronic device comprising a memory and a processor, characterized in that the memory has stored thereon a computer program, which when executed by the processor implements the detection method according to any of claims 1-7.
10. A computer readable storage medium having a computer program stored thereon, wherein the computer program when executed by a processor implements the detection method according to any one of claims 1 to 7.

Description

Serum quality detection method through image parameterization and database comparison Technical Field The invention belongs to the field of clinical medicine, and relates to the field of computer science and information technology, in particular to a serum quality detection method by image parameterization and database comparison. Background The current pre-serum analysis error rate is high, and 60% -70% of errors in laboratory diagnosis occur in the pre-serum analysis stage, possibly resulting in improper examination and unnecessary increase of cost, and even improper treatment or treatment modification in some cases, wherein hemolysis, jaundice and lipidemia samples account for the major proportion of pre-serum analysis errors. At present, the existing serum index examination (SIS) at home and abroad depends on manual judgment or reagent detection, and has the problems of low efficiency, high cost, strong subjectivity and the like. The standardized requirements of serum detection are urgent, and lack of uniform serum quality assessment standards leads to poor repeatability of detection results. In addition, shielding of serum by the barcode on the serum tube greatly interferes with the whole laboratory automation (TLA) selection and subsequent analysis of serum pictures. Therefore, there is an urgent need for a fully automated, efficient, low cost method for serum quality inspection. The content of bilirubin in jaundice serum is increased, the blood lipid concentration of triglyceride and the like in the lipidemia serum is abnormally increased, and a large amount of hemoglobin is released by the rupture of haemolytic serum erythrocytes, so that different degrees of jaundice, lipidemia and haemolysis can generate multi-level and multi-level influences on the serum color, and the RGB attribute (R is Red, G is Green and B is Blue) of pixel points on a serum picture can be influenced. At present, except for direct visual observation, the serum quality detection is mainly carried out by a biochemical analyzer method and a spectrophotometry method, the biochemical analysis method depends on equipment and matched reagents, the cost is high, and the spectrophotometry method has low batch processing efficiency and needs manual calibration. These methods all pay attention to the importance of serum color in serum quality detection, but the serum color cannot be further parameterized and combined with AI to achieve fully automatic, high-efficiency, low-cost serum quality detection. Patent CN 108562584A uses RGB data of serum pixel points, but the patent only uses the average value of R, G, B data of serum pixel points as a value for measuring the degree of jaundice, hemolysis, and lipidemia to compare with the values set by them to determine whether the serum is abnormal, which has a certain limitation. Therefore, developing a method for comparing and combining serum image parameterization with AI and database provides a brand new view angle for serum quality detection and reducing the pre-analysis error rate and has great significance. Disclosure of Invention Aiming at the problems and the shortcomings in the prior art, the invention aims to provide a serum quality detection method by comparing image parameterization with a database. In order to achieve the purpose of the invention, the technical scheme adopted by the invention is as follows: the first aspect of the invention provides a serum quality detection method by image parameterization and database comparison, comprising the following steps: S1, randomly selecting an effective serum picture containing normal serum, hemolyzed serum, lipidemia serum and jaundice serum, extracting serum pixel points, storing R data, G data and B data of the serum pixel points, respectively establishing four sub-databases of a comprehensive serum pixel point database and normal serum, hemolyzed serum, lipidemia serum and jaundice serum pixel point data, respectively extracting R data, G data and B data of the pixel points in the four sub-databases, solving target values and standard deviations of the R data, the G data and the B data, obtaining intervals of the R data, the G data and the B data, optimizing the range of the pixel points contained in the database into three intervals, and forming a color cube in three dimensions of the database; S2, serum picture analysis, namely shooting a plurality of pictures of serum to be detected, comparing R data, G data and B data of each serum picture pixel point with a comprehensive serum pixel point database, respectively recording the matching number of the serum pixel points, selecting one picture with the largest matching number as a serum window optimal picture, and outputting and using the picture for serum quality analysis; And S3, serum quality analysis, namely comparing R data, G data and B data of the pixel point of the optimal picture of the serum window with R data, G data and B data of four sub-databases, wherein the serum q