Paradkar Neeraj, Chowdhury Shubhajit Roy
Annu Int Conf IEEE Eng Med Biol Soc. 2015;2015:1959-62. doi: 10.1109/EMBC.2015.7318768.
The paper presents a technique to detect significant systolic peaks, the percussion (P) and tidal peak (T) and diastolic peak (D) from the arterial blood pressure (ABP) waveform. The technique is aimed at robust detection even in presence of significant noise. Singular Value Decomposition (SVD) based dominant period extraction of the ABP waveform followed by wavelet transform and local peak detection is applied to detect the points of interest. MIMIC-II ABP databse serves as a training dataset to select SVD and wavelet transform parameters and CSL Benchmark database is used to analyze the technique. Salient systolic peak detection for the CSL dataset was performed with positive predictive value and sensitivity figures of 98.48% and 99.24% respectively.
本文提出了一种从动脉血压（ABP）波形中检测显著收缩峰、叩击峰（P）、潮波峰（T）和舒张峰（D）的技术。该技术旨在即使在存在大量噪声的情况下也能进行稳健检测。采用基于奇异值分解（SVD）的ABP波形主导周期提取，然后进行小波变换和局部峰值检测，以检测感兴趣的点。MIMIC-II ABP数据库用作训练数据集来选择SVD和小波变换参数，CSL基准数据库用于分析该技术。对CSL数据集进行显著收缩峰检测时，阳性预测值和灵敏度分别为98.48%和99.24%。
