Establishment and application method of an automatic recognition model of immunofixation electrophoresis

Abstract

The invention discloses a method for establishing and using an automatic recognition model of an immunofixation electrophoretic graph. The model needs to be established first, and the model is trained according to the following steps: a) data preparation; b) data cleaning; c) data preprocessing; d) Data segmentation; e) Building a model, CNN network extracts the image features of the protein electrophoresis zone, and the image features of each protein electrophoresis zone are stitched together to form a sequence feature; f) Data training, training the LSTM model; g) Saving the obtained model and establishing the depth After the model, use the trained LSTM model to predict the IFE classification result. The present invention provides an end-to-end deep learning method for automatic identification of IFE diagrams, with fast running speed and high accuracy.

Description

technical field [0001] The invention relates to the field of deep learning, in particular to a method for establishing and using an automatic recognition model of immunofixation electrophoresis. Background technique [0002] Immunofixation electrophoresis (IFE) is an operation including agar gel protein electrophoresis and immunoprecipitation. Serum IFE can detect IgG, IgM, IgA, etc. and κ light chain, λ light chain. The principle is to perform zone electrophoresis on the sample on the agar plate, and after separation, it is covered with antiserum, including anti-κ light chain, anti-λ light chain, anti-μ heavy chain, anti-δ heavy chain, anti-γ heavy chain, and anti-ε heavy chain chain and anti-α heavy chain antiserum, when the antibody combines with monoclonal Ig in a certain zone, an immune complex precipitation can be formed, which can be adsorbed and fixed, and then rinsed and stained to present a thick and narrow colored zone. The technique is immunofixation electropho...

AI Technical Summary

Problems solved by technology

This method has high requirements for personnel, is time-consuming and labor-intensive, and has large classification deviations, resulting in low classification accuracy and repeatability, that is, different personnel may give different classification results for the same picture.

[0004] The best method that has been published is to first use the convolutional neural network to extract the features of the IFE graph, and then use the machine learning method to classify. This method is not an end-to-end structural design, so the running speed is slow and the real-time performance cannot be guaranteed.

Images