Arhythmia testing method for aiming at electrocardiogram data by means of binary neural network

A binary nerve and arrhythmia technology, applied in the field of deep learning, can solve the problem of high computing cost and achieve the effect of reducing computing memory, low computing power, and low energy consumption

CN110379506AActive Publication Date: 2019-10-25HANGZHOU DIANZI UNIV
10 Cites 18 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Publication Date
2019-10-25

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention discloses an arhythmia testing method for aiming at electrocardiogram data by means of a binary neural network. The method comprises the steps of firstly acquiring training model data, and preprocessing the data before training; constructing a set of full-precision convolutional network model, and inputting data for training, adjusting a network parameter for obtaining a relatively high effect; referring to the obtained full-precision models for constructing a binary network model, inputting the data for training, performing fine adjustment on a model parameter, and utilizing a Stop-BN training method in training for improving a training effect; using the trained full-precision model as a teacher model and using the untrained binary model as a student model, and performing distillation training on the student model by means of the teacher model, thereby obtaining a training effect better than that in directly training the binary network. The method performs discriminationon atrial fibrillation and can effectively reduce operation memory and operation time. The trained network model realizes relatively high effect for reducing precision loss caused by binarization.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The invention belongs to the field of deep learning, and is an arrhythmia detection application based on a convolutional neural network, which performs binary compression on its parameters and is applied to electrocardiogram data. Background technique

[0002] Atrial fibrillation, the most common irregular heartbeat, occurs with very rapid and irregular contractions of the atria and carries a high risk of death, stroke, heart failure or coronary artery disease. According to statistics, the incidence of atrial fibrillation in the population is generally 1% to 2%. At present, the general diagnosis scheme is that the patient collects the electrocardiogram through the corresponding equipment, and then submits it to the doctor for diagnosis. This method obviously consumes a lot of manpower and material resources, and the efficiency is low. Therefore, it is necessary to construct an algorithm that can perform efficient and accurate diagnosis based on ECG da...

Examples

Embodiment Construction

[0025] Such as figure 1 Shown, the specific content of the present invention is described below.

[0026] Step 1. Obtain the content of the training data. The data set used comes from the atrial fibrillation (AF) recognition competition held by PhysioNet in 2017, which contains 8528 single-lead ECG data, ranging in length from 9 seconds to 61 seconds. The sampling frequency is 300Hz. The data is labeled by experts and divided into four categories: Normal rhythm (N), AF rhythm (A), Other rhythm (O) and Noise (~). The number of samples in the four categories is uneven, of which Normal has 5154 , AF has 771, Other has 2557, and Noise has 46. Examples of different categories of data are as follows image 3 shown. Symptoms of arrhythmia that are not atrial fibrillation are grouped under other arrhythmias. Due to the different lengths of data, it is impossible to meet the requirements of batch training. Therefore, the data is first filled with data in batches. In addition, beca...