CN111444941A

CN111444941A - Method for diagnosing early lung cancer by combining electrolyte in serum and proteomic data

Info

Publication number: CN111444941A
Application number: CN202010137369.2A
Authority: CN
Inventors: 李艳坤; 董汝南; 马昕鹏; 庞佳烽; 景璟
Original assignee: North China Electric Power University
Current assignee: North China Electric Power University
Priority date: 2020-02-24
Filing date: 2020-02-24
Publication date: 2020-07-24

Abstract

The present invention discloses a method for diagnosing early lung cancer by combining electrolytes in serum and proteomics data. The method comprises the steps of firstly, carrying out wavelet transformation processing on five traditional electrolytes and eight protein content data contained in the serum of a known lung cancer sample and a normal (non-lung cancer) sample, and taking the data as a training set. And then modeling the training set by adopting a linear discriminant analysis algorithm to obtain a characteristic vector E value corresponding to each variable. And taking the sample to be diagnosed which is processed by the same wavelet transform method as a prediction set, and multiplying the prediction set by the E to obtain a discrimination vector value, thereby carrying out the identification diagnosis of the lung cancer sample and the normal sample.

Description

Method for diagnosing early lung cancer by combining electrolyte in serum and proteomic data

Technical Field

The present invention relates to a method for diagnosing early lung cancer by combining electrolytes in serum and proteomic data.

Background

Early diagnosis and timely treatment of cancer are the most effective ways to improve survival rate of cancer patients. However, early diagnosis of cancer has been a problem that plagues medical staff and related researchers. Lung cancer is one of the most common and most lethal malignant tumors in China, and the incidence rate and the growth rate of the lung cancer are high and the lung cancer is the first of all malignant tumors. Early symptoms of lung cancer are not obvious, the incubation period is long, and early diagnosis and identification of lung cancer are always difficult problems in the medical field.

Tumor markers are substances that undergo qualitative or quantitative changes in the body fluids or tissues of a tumor patient, and may reflect the presence or characteristics of a tumor. Traditional tumor markers include alpha-fetoprotein (AFP), carcinoembryonic antigen (CEA), neuron-specific enolase (NSE), and the like. The detection of tumor markers is an important means for tumor diagnosis and treatment in clinical practice at home and abroad at present, but the tumor diagnosis method based on a single traditional protein marker still has many defects, and particularly, the sensitivity and specificity of diagnosis need to be improved. Therefore, the method has important significance for diagnosing by combining a plurality of tumor markers and searching for a new tumor marker with clinical significance.

Human serum contains abundant information and can provide important clues for diseases. Compared with tissue or cell detection, the method is convenient and fast. In addition, serum diagnosis is less harmful to the human body. The electrolyte content in serum can indicate various diseases. Electrolytes are involved in many important functions and metabolic activities in the body, play an important role in maintaining normal physiological activities, and many diseases cause or accompany disorders of serum electrolytes. At present, a great deal of research aiming at the traditional protein tumor marker is carried out, and the research on the relation between electrolyte and tumor is expected to be carried out systematically.

Disclosure of Invention

The invention aims to provide a method for diagnosing early lung cancer by combining electrolyte and proteomics data in serum, and provides a new way for quickly, efficiently and accurately diagnosing early lung cancer. More accurate results are obtained relative to identification by independent electrolyte or proteomic data sets.

The method comprises the steps of taking a patient group with confirmed lung cancer and a normal (contrast) group as training sets, modeling the training sets by using the content data of five traditional electrolytes and eight proteins in serum of the patient group as discrete wavelet transform processing, then adopting a linear discriminant analysis (L DA) method to obtain a characteristic vector value (Eigenvetor, E) corresponding to each variable, wherein the dimension of the E is n × 1, and n is the number of the variables.

The method comprises the following specific steps:

1. five conventional electrolytes (K) in the serum of a patient group and a normal group with known confirmed diagnosis of lung cancer⁺，Na⁺，Cl^-，Ca²⁺，CO₂Binding force) and eight protein (NSE, Cyfra21-1, CA153, CEA, CA199, CA125, AFP, HCG) content data are processed by discrete wavelet transform (db 10' wavelet base, decomposition scale is 3) to be used as a training set;

2. performing linear discriminant analysis on the training set data obtained in the step 1 to obtain a characteristic vector (E) value corresponding to each variable;

3. for unknown sample data (prediction set) X to be diagnosed containing the same five traditional electrolytes and eight proteins_newPerforming discrete wavelet transform (DCT) similar to the training set, and Determining Vector (DV) corresponding to the transformed low-dimensional matrix by using DV ═ X_newAnd E, calculating. And (5) performing identification diagnosis of the cancer group and the normal group according to the difference of the DV values and the clustering result.

Drawings

FIG. 1 is a graph of the results of the identification and diagnosis of the combined electrolyte and protein data of lung cancer and normal samples of examples.

FIG. 2 is a graph showing the results of identification and diagnosis of electrolyte data of lung cancer and normal samples of examples.

FIG. 3 is a graph showing the results of identification and diagnosis of protein data of lung cancer and normal samples according to the examples.

Detailed Description

Example (b):

(1) the 252 Hospital of the people's liberation force of China provided 94 cases of serum of lung cancer patients and 58 cases of normal (non-lung cancer) serum, and five traditional electrolytes (K) contained in the serum were performed⁺、Na⁺、Cl^-、Ca²⁺、CO₂Binding force) and the content of eight proteins (NSE, Cyfra21-1, CA153, CEA, CA199, CA125, AFP, HCG).

(2) The serum index data of the lung cancer samples and the normal samples are randomly divided into a training set, a testing set and a prediction set, and are processed by discrete wavelet transform (db 10' wavelet basis, decomposition scale is 3).

(3) And (5) carrying out linear discriminant analysis on the training set samples to calculate a characteristic vector (E).

(4) Calculating DV values of the test set and the prediction set according to the E value, and performing discriminant diagnosis on the cancer group and the normal group according to the difference of the DV values and the clustering result, wherein the result is shown in figure 1. Fig. 2 and 3 are graphs showing the results of using the electrolyte and protein data alone, respectively, and it is apparent that the diagnostic result is best recognized using the combined data, and the accuracy reaches 100%, when compared with fig. 1.

Claims

1. A method for diagnosing early stage lung cancer by combining electrolyte and proteomic data in serum, comprising the steps of:

step one, five traditional electrolytes (K) contained in a known lung cancer serum sample and a normal (non-lung cancer) serum sample are used⁺，Na⁺，Cl^-，Ca²⁺，CO₂Binding force) and eight kinds of protein (NSE, Cyfra21-1, CA153, CEA, CA199, CA125, AFP, HCG) are combined to perform discrete wavelet transform ("db 10" wavelet base, decomposition scale is 3) processing, and the discrete wavelet transform is used as a training set;

performing linear discriminant analysis on the training set to calculate a feature vector E value corresponding to the variable;

and step three, processing the sample to be diagnosed by adopting a wavelet transformation method which is the same as the training set to serve as a prediction set, multiplying the prediction set by the E to obtain a discrimination vector value corresponding to the sample, and identifying and diagnosing the lung cancer sample and the normal sample according to the size difference of the discrimination vector value.