Spike Detection and Filtering
In the context of Peak Detection, a peak or local maximum is defined as any sample whose two direct neighbours have a smaller amplitude and a trough or local minimum is defined as any sample whose two direct neighbours have a larger amplitude. Peak detection is often used for identifying certain events which cause undulations in the signal which are recognised as peaks.
In this library, we are able to identify spikes and filter them based on these properties:
- Left Height of Spike: The height of the peak from the neighbouring trough on the left.
- Right Height of Spike: The height of the peak from the neighbouring trough on the right.
- Mean Height of Spike: The height of the peak computed as the mean of the 2 neighbouring trough on the left and right.
- Min Height of Spike: The height of the peak computed as the minimum of the 2 neighbouring trough on the left and right.
- Max Height of Spike: The prominence of a peak computed as the maximum of the 2 neighbouring trough on the left and right.
This diagram provides a explanation regarding the properties mentioned above.
double[] data = UtilMethods.splitByIndex(UtilMethods.electrocardiogram(), 3200, 4200);
Smooth sObj = new Smooth(data, 15, "rectangular");
double[] ecgSignal = sObj.smoothSignal("same");
The first step of spike detection is to identify all the peaks in the signal and along with that map out all the neighbouring troughs for each peak.
FindPeak fp = new FindPeak(this.highResSignal);
Spike out = fp.getSpikes();
double[] outLeftSpike = out.getLeftSpike();
double[] outRightSpike = out.getRightSpike();
double[] outMeanSpike = out.getMeanSpike();
double[] outMaxSpike = out.getMaxSpike();
double[] outMinSpike = out.getMinSpike();
outLeftSpike: [Double.NaN, 0.107, 0.598, 0.007, 0.236, 0.08 , 0.629, 0.016, 0.215,
0.056, 0.651, 0.323, 0.004, 0.074, 0.003, 0.558, 0.012, 0.021,
0.311, 0.09 , 0.006, 0.004, 0.604, 0.005, 0.007, 0.36 , 0.086,
0.846, 0.003, 0.001, 0.007]
outRightSpike: [0.278, 0.299, 0.66 , 0.003, 0.165, 0.289, 0.691, 0.062, 0.263,
0.323, 0.703, 0.007, 0.189, 0.001, 0.243, 0.558, 0.007, 0.004,
0.154, 0.003, 0.162, 0.029, 0.713, 0.004, 0.006, 0.2 , 0.22 ,
0.826, 0.011, 0.005, 0.002]
outMeanSpike: [0.278, 0.203 , 0.629 , 0.005 , 0.2005, 0.1845, 0.66 , 0.039 ,
0.239 , 0.1895, 0.677 , 0.165 , 0.0965, 0.0375, 0.123 , 0.558 ,
0.0095, 0.0125, 0.2325, 0.0465, 0.084 , 0.0165, 0.6585, 0.0045,
0.0065, 0.28 , 0.153 , 0.836 , 0.007 , 0.003 , 0.0045]
outMaxSpike: [0.278, 0.299, 0.66 , 0.007, 0.236, 0.289, 0.691, 0.062, 0.263,
0.323, 0.703, 0.323, 0.189, 0.074, 0.243, 0.558, 0.012, 0.021,
0.311, 0.09 , 0.162, 0.029, 0.713, 0.005, 0.007, 0.36 , 0.22 ,
0.846, 0.011, 0.005, 0.007]
outMinSpike: [0.278, 0.107, 0.598, 0.003, 0.165, 0.08 , 0.629, 0.016, 0.215,
0.056, 0.651, 0.007, 0.004, 0.001, 0.003, 0.558, 0.007, 0.004,
0.154, 0.003, 0.006, 0.004, 0.604, 0.004, 0.006, 0.2 , 0.086,
0.826, 0.003, 0.001, 0.002]
Spike Filtering can be done using thresholds on the existing properties.
int[] outLeftFilter = out.filterByProperty(0.5, 5.0, "left");
int[] outRightFilter = out.filterByProperty(0.5, 5.0, "right");
outLeftFilter = [93, 257, 415, 575, 747, 928]
outRightFilter = [93, 257, 415, 575, 747, 928]
int[] outMeanFilter = out.filterByProperty(0.5, 5.0, "mean");
outMeanFilter = [93, 257, 415, 575, 747, 928]
int[] outMaxFilter = out.filterByProperty(0.5, 5.0, "max");
int[] outMinFilter = out.filterByProperty(0.5, 5.0, "min");
outMaxFilter = [93, 257, 415, 575, 747, 928]
outMinFilter = [93, 257, 415, 575, 747, 928]