Digital Information Extraction And Analysis Of Paper ECG

ECG detection is the main method for medical institutions to diagnose heart disease.The existing paper ECG archiving method is difficult to save and takes up a lot of space,which is not conducive to the construction of electronic medical record database.In addition,doctors are less efficient in diagnosing ECGs by manual identification and are prone to misdiagnosis.In view of the above problems,this paper extracted and analyzed the digital information of paper ECG from three aspects: preprocessing of paper ECG,extraction of ECG waveform and waveform detection of ECG signals.This paper firstly preprocessed the paper ECG and compares the edge detection effect of Sobel operator,Canny operator and LOG operator on paper ECG.The skew correction was performed on the image after edge detection based on Hough line detection.The morphological operation of ECG was then introduced,which was the basis of ECG waveform extraction.For the extraction of ECG waveforms,this paper studied the curve extraction of color ECG and monochromatic ECG.Aiming at color ECG,a K-means++ ECG curve extraction algorithm based on Laplace-Erode enhancement was proposed.For monochromatic ECG,an ECG curve extraction algorithm based on connected region analysis is proposed.Waveform segmentation was then performed based on the horizontal projection operation,and the 12-lead electrocardiogram was segmented into a single lead to facilitate the extraction of ECG data.In order to ensur e that the extracted waveform meets the one-to-one correspondence between the horizontal pixel and the vertical pixel,an improved curve skeleton thinning algorithm was proposed to refine the waveform curve.And the corresponding relationship between the pixel with time and voltage was reconstructed according to the background mesh.The waveform data was then converted into ECG signal data to complete the extraction of ECG data.The accuracy of the algorithm was verified by experiments.For the waveform detection of ECG signal data,this paper firstly proposed and implemented a denoising algorithm based on 8-layer wavelet transform according to the noise characteristics of ECG signals,which effectively removed the noise such as baseline drift and power frequency interference generated by ECG signals.Next,for the identification of R wave peaks,a differential threshold amplification method was proposed.By comparing with the R waves marked by experts in the MIT-BIH database,the detection results of R waves were verified based on the evaluation indexes such as accuracy and recall rate.Next,based on the position of the R wave peak,the Q wave,S wave,Q wave boundary,S wave boundary,and the P wave,T wave,P wave boundary,and T wave boundary were respectively identified according to the local region search algorithm.The detection effect of each characteristic waveform was analyzed.According to the above research content,this paper designed and implemented the ECG digital system,which realized the main functions of digital processing and waveform detection of paper ECG,and verified the main research of this paper.