Heartbeats Classification Using Hybrid Time-Frequency Analysis and Transfer Learning Based on ResNet

被引:46
作者
Zhang, Yatao [1 ]
Li, Junyan [2 ]
Wei, Shoushui [3 ]
Zhou, Fengyu [3 ]
Li, Dong [1 ]
机构
[1] Shandong Univ, Sch Mech Elect & Informat Engn, Weihai 264209, Peoples R China
[2] Weihaiwei Peoples Hosp, Dept Neurol, Weihai 264209, Peoples R China
[3] Shandong Univ, Sch Control Sci & Engn, Jinan 17923, Peoples R China
基金
中国国家自然科学基金; 中国博士后科学基金;
关键词
Heartbeat classification; arrhythmias; time-frequency analysis; Transfer learning; ResNet-101; ATRIAL-FIBRILLATION; NEURAL-NETWORK; ECG;
D O I
10.1109/JBHI.2021.3085318
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
The classification of heartbeats is an important method for cardiac arrhythmia analysis. This study proposes a novel heartbeat classification method using hybrid time-frequency analysis and transfer learning based on ResNet-101. The proposed method has the following major advantages over the afore-mentioned methods: it avoids the need for manual features extraction in the traditional machine learning method, and it utilizes 2-D time-frequency diagrams which provide not only frequency and energy information but also preserve the morphological characteristic within the ECG recordings, and it owns enough deep to make better use of performance of CNN. The method deploys a hybrid time-frequency analysis of the Hilbert transform (HT) and the Wigner-Ville distribution (WVD) to transform 1-D ECG recordings into 2-D time-frequency diagrams which were then fed into a transfer learning classifier based on ResNet-101 for two classification tasks (i.e., 5 heartbeat categories assigned by the ANSI/AAMI standard (i.e., N, V, S, Q and F) and 14 original beat kinds of the MIT/BIH arrhythmia database). For 5 heartbeat categories classification, the results show the F1-score of N, V, S, Q and F categories are F-N 0.9899, F-V 0.9845, F-S 0.9376, F-Q 0.9968, F-F 0.8889, respectively, and the overall F1-score is 0.9595 using the combination data balancing. The results show the average values for accuracy, sensitivity, specificity, predictive value and F1-score on test set for 14 beat kinds the MIT-BIH arrhythmia database are 99.75%, 91.36%, 99.85%, 90.81% and 0.9016, respectively. Compared with other methods, the proposed method can yield more accurate results.
引用
收藏
页码:4175 / 4184
页数:10
相关论文
共 33 条
[1]   A deep convolutional neural network model to classify heartbeats [J].
Acharya, U. Rajendra ;
Oh, Shu Lih ;
Hagiwara, Yuki ;
Tan, Jen Hong ;
Adam, Muhammad ;
Gertych, Arkadiusz ;
Tan, Ru San .
COMPUTERS IN BIOLOGY AND MEDICINE, 2017, 89 :389-396
[2]   Automated detection of arrhythmias using different intervals of tachycardia ECG segments with convolutional neural network [J].
Acharya, U. Rajendra ;
Fujita, Hamido ;
Lih, Oh Shu ;
Hagiwara, Yuki ;
Tan, Jen Hong ;
Adam, Muhammad .
INFORMATION SCIENCES, 2017, 405 :81-90
[3]   A Fast Machine Learning Model for ECG-Based Heartbeat Classification and Arrhythmia Detection [J].
Alfaras, Miquel ;
Soriano, Miguel C. ;
Ortin, Silvia .
FRONTIERS IN PHYSICS, 2019, 7
[4]  
[Anonymous], 1998, EC572012 AAMI
[5]   A FAST METHOD FOR REAL-TIME MEDIAN FILTERING [J].
ATAMAN, E ;
AATRE, VK ;
WONG, KM .
IEEE TRANSACTIONS ON ACOUSTICS SPEECH AND SIGNAL PROCESSING, 1980, 28 (04) :415-421
[6]   Multi-information fusion neural networks for arrhythmia automatic detection [J].
Chen, Aiyun ;
Wang, Fei ;
Liu, Wenhan ;
Chang, Sheng ;
Wang, Hao ;
He, Jin ;
Huang, Qijun .
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE, 2020, 193
[7]   Automated arrhythmia classification based on a combination network of CNN and LSTM [J].
Chen, Chen ;
Hua, Zhengchun ;
Zhang, Ruiqi ;
Liu, Guangyuan ;
Wen, Wanhui .
BIOMEDICAL SIGNAL PROCESSING AND CONTROL, 2020, 57
[8]   Comparative study of morphological and time-frequency ECG descriptors for heartbeat classification [J].
Christov, Ivaylo ;
Gomez-Herrero, German ;
Krasteva, Vessela ;
Jekova, Irena ;
Gotchev, Atanas ;
Egiazarian, Karen .
MEDICAL ENGINEERING & PHYSICS, 2006, 28 (09) :876-887
[9]   Classification of normal sinus rhythm, abnormal arrhythmia and congestive heart failure ECG signals using LSTM and hybrid CNN-SVM deep neural networks [J].
Cinar, Ahmet ;
Tuncer, Seda Arslan .
COMPUTER METHODS IN BIOMECHANICS AND BIOMEDICAL ENGINEERING, 2021, 24 (02) :203-214
[10]   Neural network and wavelet average framing percentage energy for atrial fibrillation classification [J].
Daqrouq, K. ;
Alkhateeb, A. ;
Ajour, M. N. ;
Morfeq, A. .
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE, 2014, 113 (03) :919-926