Details
Presenter(s)
![Jianzhong Chen Headshot](https://confcats-catavault.s3.amazonaws.com/CATAVault/ieeecass/master/files/styles/cc_user_photo/s3/user-pictures/70831.jpg?h=d0470b75&itok=kqsxnhYH)
Display Name
Jianzhong Chen
- Affiliation
-
AffiliationShanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences
- Country
Abstract
This paper presents an automatic delineator based on the deep learning method, which uses a one-dimensional U-Net network to realize the feature point detection of ABP signals. The network can divide ABP signals into three parts to accurately detect feature points. The method was validated on an ABP data set of 68 people, 500s per person. The performance is good and the average time difference is less than 4 ms. Finally, the method performed with a sensitivity of 99.87%, 99.82%, and 99.24%, a positive predictivity of 99.73%, 99.99%, and 99.26%, and an error rate of 0.4%, 0.2%, and 1.5% for the onsets, systolic peaks, and dicrotic notches with a tolerance threshold of 30 ms.