def AOPSinusoidModeling(filename, WindowSize=2):
path = 'C:\\Users\\shijingliu\\workspace\\PowerFactor\\'
file = filename
finalname = path + file + '.txt'
LabelVector, DataMatrix = DataFileLoader.load(finalname)
AOPcolumn = VectorOperations.FindColumn('AOP', LabelVector)
if AOPcolumn > 10:
print "finding new AOP"
AOPcolumn = VectorOperations.FindColumn('AOP 1', LabelVector)
AOP = DataMatrix[:, AOPcolumn]
'declare a file'
saveEVFile = open('SinusoidModeling CPR Flow 347.csv', 'wb')
fileWriter = csv.writer(saveEVFile)
fileWriter.writerows(
[["index", "estimated A:", "estimated B", "estimated D"]])
for i in range((len(AOP) / 100) - WindowSize):
'define the start time and end time of each interval'
StartTime = i
EndTime = StartTime + WindowSize
AopInterval = AOP[StartTime * 100:EndTime * 100]
t = np.linspace(StartTime, EndTime - 0.01, WindowSize * 100)
'figure out the appropriate guess using FFT'
amplitude = np.abs(np.fft.fft(AopInterval))[1:(len(t) / 2 + 1)]
frequency = np.fft.fftfreq(len(t), 0.01)[1:(len(t) / 2 + 1)]
'find out the maximum amplitude and its corresponding frequency'
maxFreq, maxAmp = MaximumRecorder(amplitude)
'now we guess the a, b, c, d for the sinusoid modeling'
guess_a = VectorOperations.RMS(
scipy.signal.detrend(AopInterval)) * np.sqrt(2)
guess_b = (3.1415926 * 2.0) * frequency[maxFreq]
guess_c = 0.0
guess_d = np.mean(AopInterval)
'optimize the guess values using least square algorithm'
optimize_func = lambda x: x[0] * np.sin(x[1] * t + x[2]) + x[
3] - AopInterval
result = leastsq(optimize_func, [guess_a, guess_b, guess_c, guess_d])
'we only record the sets when the least square can converge'
if (result[1] == 1) or (result[1] == 2) or (result[1]
== 3) or (result[1] == 4):
est_a, est_b, est_c, est_d = result[0]
fileWriter.writerows([[i, est_a, est_b, est_d]])
saveEVFile.close()
def AMSAcalculation(filename, windowSize=4, timeStep=1):
path = 'C:\\Users\\shijingliu\\workspace\\PowerFactor\\'
file = filename
finalname = path + file + '.txt'
LabelVector, DataMatrix = DataFileLoader.LoadADIData(finalname)
EKGcolumn = VectorOperations.FindColumn('EKG', LabelVector)
time = DataMatrix[:, 0]
EKG = DataMatrix[:, EKGcolumn]
AMSA = []
for i in xrange(0, len(time) / 100, timeStep):
intervalStart = i
intervalEnd = i + windowSize
currentTS = time[intervalStart * 100:intervalEnd * 100]
currentES = EKG[intervalStart * 100:intervalEnd * 100]
'first of all get rid of nan value in EKG'
recordNan = []
for j in range(len(currentES)):
if np.isnan(currentES[j]):
recordNan.append(j)
currentESRefine = currentES[~np.isnan(currentES)]
currentTSRefine = np.delete(currentTS, recordNan)
'now we can apply the FFT'
time_step = 0.01
freqs = np.fft.fftfreq(len(currentTSRefine), time_step)
halfFreqs = freqs[0:len(freqs) / 2]
'then we calculate the amplitude using FFT'
amplitude = np.abs(np.fft.fft(currentESRefine))**2
amplitudeNorm = amplitude / np.sum(amplitude)
'only analyze half of the amplitude due to symmetrical problems'
halfAmplitude = amplitudeNorm[0:len(amplitude) / 2]
'now we need to get rid of the frequency beyond the range of 4 and 48 '
amplitudefilter = []
frequencyfilter = []
for k in range(len(halfFreqs)):
if halfFreqs[k] >= 4.0 and halfFreqs[k] <= 48.0:
frequencyfilter.append(halfFreqs[k])
amplitudefilter.append(halfAmplitude[k])
'now calculate the amsa score here'
AMSA.append(np.sum(np.multiply(amplitudefilter, frequencyfilter)))
return AMSA
def FFTAnalysis(filename, windowSize=4, timeStep=1):
path = 'C:\\Users\\shijingliu\\workspace\\PowerFactor\\'
file = filename
finalname = path + file + '.txt'
LabelVector, DataMatrix = DataFileLoader.LoadADIData(finalname)
EKGcolumn = VectorOperations.FindColumn('EKG', LabelVector)
time = DataMatrix[:, 0]
EKG = DataMatrix[:, EKGcolumn]
Fmean = []
Fmax = []
Amean = []
Amax = []
for i in xrange(0, len(time) / 100, timeStep):
intervalStart = i
intervalEnd = i + windowSize
currentTS = time[intervalStart * 100:intervalEnd * 100]
currentES = EKG[intervalStart * 100:intervalEnd * 100]
'first of all get rid of nan value in EKG'
recordNan = []
for j in range(len(currentES)):
if np.isnan(currentES[j]):
recordNan.append(j)
currentESRefine = currentES[~np.isnan(currentES)]
currentTSRefine = np.delete(currentTS, recordNan)
'now we can apply the FFT'
time_step = 0.01
freqs = np.fft.fftfreq(len(currentTSRefine), time_step)
halfFreqs = freqs[0:len(freqs) / 2]
'then we calculate the amplitude using FFT'
amplitude = np.abs(np.fft.fft(currentESRefine))**2
amplitudeNorm = amplitude / np.sum(amplitude)
'only analyze half of the amplitude due to symmetrical problems'
halfAmplitude = amplitudeNorm[0:len(amplitude) / 2]
'now we calculate the Fmean, Fmax, Amean, and Amax'
Fmean.append(halfFreqs.mean())
Fmax.append(halfFreqs.max())
Amean.append(halfAmplitude.mean())
Amax.append(halfAmplitude.max())
return Fmax, Fmean, Amax, Amean
def EigenValueAnalysis(filename, windowSize=1):
path = 'C:\\Users\\shijingliu\\workspace\\PowerFactor\\'
file = filename
finalname = path + file + '.txt'
LabelVector, DataMatrix = DataFileLoader.LoadADIData(finalname)
EKGcolumn = VectorOperations.FindColumn('EKG', LabelVector)
time = DataMatrix[:, 0]
EKG = DataMatrix[:, EKGcolumn]
'first of all get rid of nan value in EKG'
recordNan = []
for j in range(len(EKG)):
if np.isnan(EKG[j]):
recordNan.append(j)
EKGrefined = EKG[~np.isnan(EKG)]
Timerefined = np.delete(time, recordNan)
'declare a file'
saveEVFile = open('EigenVector and EigenValue CPR Flow 347.csv', 'wb')
fileWriter = csv.writer(saveEVFile)
'now we go through the entire txt data file'
for i in xrange(0, (len(Timerefined) / 100)):
intervalStart = i
intervalEnd = i + windowSize * windowSize
currentTS = Timerefined[intervalStart * 100:intervalEnd * 100]
currentES = EKGrefined[intervalStart * 100:intervalEnd * 100]
'matrix construction'
MConstruction = np.zeros(shape=(10 * windowSize, 10 * windowSize))
for j in range(10 * windowSize):
center = (
currentES[j * 10 * windowSize + 5 * windowSize] +
currentES[j * 10 * windowSize + 5 * windowSize - 1]) / 2.0
for k in range(10 * windowSize):
MConstruction[j][k] = currentTS[j * 10 * windowSize +
k] - center
'then calculate the eigenvectors'
eigenvalues, eigenvectors = np.linalg.eig(MConstruction)
fileWriter.writerows([[i]])
fileWriter.writerows([["eigenvalues:"]])
fileWriter.writerows([eigenvalues])
fileWriter.writerows([["eigenvectors:"]])
fileWriter.writerows(eigenvectors)
fileWriter.writerows([""])
saveEVFile.close()
def PowerFactor(filename, time, type):
path = 'C:\\Users\\shijingliu\\workspace\\PowerFactor\\'
file = filename
filename = path + file + '.txt'
LabelVector, DataMatrix = DataFileLoader.LoadADIData(filename)
AORTScolumn = VectorOperations.FindColumn('Aorta', LabelVector)
AOPcolumn = VectorOperations.FindColumn('AOP', LabelVector)
RAPcolumn = VectorOperations.FindColumn('RAP', LabelVector)
IVCTScolumn = VectorOperations.FindColumn('IVC', LabelVector)
if AOPcolumn > 10:
print "finding new AOP"
RAPcolumn = VectorOperations.FindColumn('RAP 1', LabelVector)
AOPcolumn = VectorOperations.FindColumn('AOP 1', LabelVector)
if type == 'aorta':
'obtain data from the first t seconds'
Aorts = DataMatrix[:, AORTScolumn][0:time * 100]
'process the flow one more time get all absolute values'
AbsAorts = np.abs(Aorts)
Aop = DataMatrix[:, AOPcolumn][0:time * 100]
S = np.multiply(AbsAorts, Aop)
elif type == 'vena':
Ivcts = DataMatrix[:, IVCTScolumn][0:time * 100]
AbsIvcts = np.abs(Ivcts)
Rap = DataMatrix[:, RAPcolumn][0:time * 100]
S = np.multiply(AbsIvcts, Rap)
else:
print "wrong input!"
return
'smooth the value with 11 in window and hanning in type'
sasmooth = VectorOperations.smooth(S, 11, 'hanning')
'obtain the RMS value of t minutes baseline'
SValue = VectorOperations.RMS(sasmooth[0:12000], 0)
'obtain the min, max value of each phase'
min, max = VectorOperations.FindExtrema(sasmooth, 0.6, 100)
'now we obtain the RMS value of each phase during t minutes'
SEachPhase = []
for i in range(2, len(max)):
SEachPhase.append(
VectorOperations.RMS(sasmooth[max[i - 1][0]:max[i][0]], 0))
'obtain the power factor vector'
PF = [x / (SValue * 1.0) for x in SEachPhase]
return PF
