Electrocardio image recognition method under weak supervision

Abstract

The invention discloses an electrocardio image recognition method under weak supervision, and belongs to the field of image recognition. The method is characterized by comprising the following steps of 1) using a denoising algorithm to remove electrocardio signal noise; 2) locating each heart beat in an electrocardio signal through a locating algorithm, then cutting the electrocardio signal into single heart beats, wherein it is ensured that each heart beat includes all information of one heartbeat; 3) converting the one-dimensional heart beats into heart beat pictures, and then dividing the heart beat pictures into three parts including training pictures, verification pictures and test pictures; 4) inputting the heart beat training pictures into a convolutional neural network for training, and constructing a heart beat picture recognition model; 5) inputting the heart beat verification pictures into the recognition model in step 4 for verifying the heart beat picture recognition accuracy of the model and adjusting various key parameter values; 6) finally, inputting the heart beat test pictures into the heart beat picture recognition model after parameter adjustment in step 5, andconducting classification. By means of the method, the accuracy of classifying the heart beat pictures is high, only a small amount of picture data is needed for model construction, and the method hasgreat significance for the accurate identification of the electrocardio signal.

Description

technical field [0001] The invention belongs to the field of image recognition and relates to an electrocardiographic image recognition method under weak supervision. Background technique [0002] Deep learning is a method based on representation learning of data in machine learning. The concept of deep learning was proposed by Hinton et al. in 2006. The concept of deep learning originated from the study of artificial neural networks. A multi-layer perceptron with multiple hidden layers is an example of a deep learning architecture. Deep learning combines low-level features to form more abstract high-level features to discover distributed feature representations of data. [0003] Convolutional neural network is an efficient recognition method developed in recent years. In the 1960s, Hubel and Wiese found that its unique network structure can effectively reduce the complexity of the feedback neural network when studying the neurons used for local sensitivity and direction...

AI Technical Summary

Problems solved by technology

However, since the heart beats of the same type of different people have various forms, different types of heart beats may also be similar in shape to each other, resulting in difficulties in feature extraction and low classification accuracy of heart beats.

[0005] With the development of deep learning, a large number of researchers have applied the one-dimensional convolutional neural network (1-D CNN) to the field of automatic ECG signal classification, and achieved good results, but there are still problems with low accuracy.

Images