论文阅读和分析:A Robust Motion Artifact Detection Algorithm for Accurate Detection of Heart Rates from Photoplethysmographic Signals using Time-Frequency Spectral Features
主要内容:
1、提取PPG信号时频谱特征,然后使用机器学习算法SVM来判断PPG信号运动伪影部分;
2、提取特征的方式比较新颖,值得学习;
流程图:
提取的特征:
1、使用VFCDM将PPG信号转换到时频谱;
2、计算特征:
P
n
o
i
s
e
P_{noise}
Pnoise:
P
n
o
i
s
e
=
P
T
F
S
−
∑
i
=
1
3
∑
t
A
M
i
,
t
P_{noise}=P_{TFS}-\sum_{i=1}^{3}\sum_{t}AM_{i,t}
Pnoise=PTFS−i=1∑3t∑AMi,t
d
f
F
M
df_{FM}
dfFM:
d
f
F
M
=
∑
i
=
2
3
∑
t
∣
F
M
i
,
t
−
i
×
F
M
1
,
t
∣
df_{FM}=\sum_{i=2}^{3}\sum_{t}\bigl|FM_{\mathrm{i},t}-i\times FM_{1,t}\bigr|
dfFM=i=2∑3t∑
FMi,t−i×FM1,t
d
f
H
R
df_{HR}
dfHR:
d
f
H
R
=
∑
t
∣
F
M
1
,
t
−
m
e
d
i
a
n
(
1
P
P
)
∣
df_{HR}=\sum_t\left|FM_{1,t}-\mathrm{median}\left(\frac{1}{PP}\right)\right|
dfHR=t∑
FM1,t−median(PP1)
3、使用运动造成运动伪影,作为标签label;
实验结果:
对比其他几种方法,可以准确的将运动伪影识别:
参考:
A Robust Motion Artifact Detection Algorithm for Accurate Detection of Heart Rates from Photoplethysmographic Signals using Time-Frequency Spectral Features