influence of non-physiological blood pressure artifacts on ... · influence of non-physiological...
TRANSCRIPT
Influence of Non-Physiological Blood Pressure Artifactson Cerebral Autoregulation
Erica. M. Rutter1 Adam Mahdi 2
1Center for Research in Scientific ComputationDepartment of Mathematics
North Carolina State University
2Institute of Biomedical EngineeringUniversity of Oxford
JMM, 2017
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 1 / 29
Outline
1 Cerebral AutoregulationBiological IntroductionMathematical Introduction
2 ArtifactsImplementationE↵ect on ARI
3 Conclusions and Further Directions
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 2 / 29
Outline
1 Cerebral AutoregulationBiological IntroductionMathematical Introduction
2 ArtifactsImplementationE↵ect on ARI
3 Conclusions and Further Directions
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 3 / 29
Cerebral Autoregulation
Cerebral Autoregulation: Combination of various physiological processesmodulating blood vessel properties in order to maintain constant cerebralblood flow despite changes in arterial blood pressure
Figure: Cerebral Autoregulation Zone. Source: Pires, Paulo W., et al. ”The e↵ects of
hypertension on the cerebral circulation.” American Journal of Physiology-Heart and Circulatory
Physiology 304.12 (2013): H1598-H1614.[4]
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 4 / 29
Cerebral AutoregulationWhy do we care?
Imparied Autoregulation seen in patients with:strokehead injurieschronic hypertension
Impaired Autoregulation is associated withbrain tissue injury by changes in blood pressuredisruption of blood-brain barrierdeath
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 5 / 29
Cerebral AutoregulationHow to measure?
ARI: Autoregulation Index (clinical tool) measures how well cerebralautoregulation is occurring
0 ! 9No Autoregulation Best Autoregulation
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 6 / 29
ARIHow to measure?
Raw ABP signals are pre-processed to obtain a beat-to-beat average
ARI is calculated using processed ABP and produces predicted CBFV
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 7 / 29
Data and Processing
36 one-minute baseline ABPand CBFV readings ofnormotensive subjects
low-pass filtered usingzero-phase 4th-orderButterworth filter in bothdirections with cuto↵ frequencyof 20 Hz
beat-to-beat detection
downsampled to 10 Hz foruniform signals
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 8 / 29
Outline
1 Cerebral AutoregulationBiological IntroductionMathematical Introduction
2 ArtifactsImplementationE↵ect on ARI
3 Conclusions and Further Directions
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 9 / 29
Mathematical ModelARI
Tiecks et. al [5] formulated the following phenomenological model:
dP =P � Pm
Pm � Pcr
x2 = x2 +x1 � 2Dx2
fT
x1 = x1 +dP � x2
fT
V̂ (T ,D,K ) = Vm (1 + dP � Kx2)
ARI = minARI2[0,9]
||V̂ (T ,D,K )� CBFV ||
whereP - ABP input, Pm - mean ABP
V̂ - CBFV output, Vm - mean CBFV
f - sampling frequency
Pcr - critical pressure (12 mmHG)Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 10 / 29
Mathematical ModelVariations
Tiecks et. al [5] model was not specific of time points for x1 and x2:
dP(t) =P(t)� Pm
Pm � Pcr
V̂ (t) = Vm (1 + dP � Kx2(t))
Panerai et. al 1999 [3]
x1(t) = x1(t � 1)+
dP(t)� x2(t � 1)
fTx2(t) = x2(t � 1)+
x
1
(t)� 2Dx2(t � 1)
fT
Panerai 2010 et. al [1]
x1(t) = x1(t � 1)+
dP(t� 1)� x2(t � 1)
fTx2(t) = x2(t � 1)+
x
1
(t� 1)� 2Dx2(t � 1)
fT
Current Formulation
x1(t) = x1(t � 1)+
dP(t)� x2(t � 1)
fTx2(t) = x2(t � 1)+
x
1
(t� 1)� 2Dx2(t � 1)
fT
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 11 / 29
Mathematical ModelComparisons
We observe little di↵erence between the various formulations for models,so we stick with the 2016 “current” model
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 12 / 29
Mathematical ModelFinal Formulation
dP(t) =P(t)� Pm
Pm � Pcr
x1(t) = x1(t � 1) +dP(t)� x2(t � 1)
fT
x2(t) = x2(t � 1) +x1(t � 1)� 2Dx2(t � 1)
fT
V̂ (t,T ,D,K ) = Vm (1 + dP � Kx2(t))
ARI = minARI2[0,9]
||V̂ (t,T ,D,K )� CBFV (t)||
whereP - ABP input, Pm - mean ABPV̂ - CBFV output, Vm - mean CBFVf - sampling frequencyPcr - critical pressure (12 mmHG)
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 13 / 29
Outline
1 Cerebral AutoregulationBiological IntroductionMathematical Introduction
2 ArtifactsImplementationE↵ect on ARI
3 Conclusions and Further Directions
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 14 / 29
ABP MeasurementSetup
Figure: ABP measurement set-up. Source: “Arterial Line and Intra-arterial Blood
Pressure Monitoring.” Life in the Fast Lane Medical Blog. 14 June 2015. Web. 03 Jan. 2017
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 15 / 29
ABP MeasurementNon-Physiological Artifacts
Figure: Where non-physiological artifacts can occur in ABP measurement.Source: “Arterial Line and Intra-arterial Blood Pressure Monitoring.” Life in the Fast Lane
Medical Blog. 14 June 2015. Web. 03 Jan. 2017Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 16 / 29
Non-Physiological Artifacts
Figure: Real examples of non-physiological artifacts in ABP measurements.Source: Li, Qiao et. al. ”Artificial arterial blood pressure artifact models and an evaluation of
a robust blood pressure and heart rate estimator.” Biomedical engineering online 8.1 (2009):.[2]
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 17 / 29
Non-Physiological ArtifactsImplementation
Only raw ABP data is altered to include non-physiological artifacts
CBFV data is not altered!
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 18 / 29
Non-Physiological ArtifactsImplementation
Each artifact is inserted in to the raw ABP measurements 5 secondsinto the time series
Severity of artifacts are measured in ‘levels’ where level 0 correspondsto no artifact and level 20 corresponds to maximal artifact.
For each artifact formulation, there may be multiple parametersgoverning the shape of the artifact. Upper and lower bounds will begiven for each parameter. Levels 0-20 correspond to linearlyincreasing all parameters from their lower to upper bounds
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 19 / 29
Artifact 1Saturation to maximal ABP
Amax(↵, L,Pmax) = tanh(↵⇡t)(Pmax � Pdias) + Pdias
Parameter Lower Bound Upper Bound Meaning↵ 0 0.1 saturation rate
Pmax (mmHG) 190 210 maximal ABPL (s) 0 5 total time of artifact
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 20 / 29
Artifact 2Square Wave Pulse
Asw (Pmax, L) =
⇢Pmax t 2 [0, L/2]0 t 2 [L/2, L].
Parameter Lower Bound Upper Bound MeaningPmax (mmHG) 190 210 maximal ABP
L (s) 0 10 total time of artifact
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 21 / 29
Artifact 3Pulse Pressure Reduction
Reduces systolic pressure linearly over the span of 45 seconds to the value
Pend = MPsys
Parameter Lower Bound Upper Bound MeaningM 1 0.1 ratio at end of artifact
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 22 / 29
Artifact 4Impulse
fL =
8<
:
L sin(2⇡t/L)
2⇡tt 2 [�L/2, L/2]
0 otherwise.
A
imp
(L) = P + (Psys � Pdia) fL, t 2 [�L/2, L/2]
Parameter Lower Bound Upper Bound MeaningL (s) 0 2 total time of artifact
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 23 / 29
Outline
1 Cerebral AutoregulationBiological IntroductionMathematical Introduction
2 ArtifactsImplementationE↵ect on ARI
3 Conclusions and Further Directions
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 24 / 29
Artifact E↵ects
50
100
150
200
AB
P [
mm
Hg
]A
max
Level 10Level 20
40
60
80
100
120
140
App
AB
P [
mm
Hg
]
0
50
100
150
200
Asw
AB
P [
mm
Hg
]
0 10 20 30 40 50
50
100
150
200
Aimp
AB
P [
mm
Hg
]
Time [s]
1
3
5
7
9
Amax
AR
I
Amax
mA
RI±
SD
1
3
5
7
9
App
AR
I
App
mA
RI±
SD
1
3
5
7
9
Asw
AR
I
Asw
mA
RI±
SD
0 4 8 12 16 20
1
3
5
7
9
Aimp
Art. Size
AR
I
0 4 8 12 16 20
Aimp
Art. Size
mA
RI±
SD
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 25 / 29
Artifact E↵ects10% ARI Change
Definition
The critical artifact size is the level of artifact which generates a change of10% change in the mean ARI of all patients.
Artifact Critical size ParametersA
max
5.21± 2.1 L=1.3 s Pmax = 195.2mmHg ↵ = 0.026A
sw
2.75± 2.1 L=1.4 s Pmax = 192.8mmHgA
pp
8.06± 5.1 M = 0.64A
imp
11.52± 3.9 L=1.05 s
Table: The mean critical artifact size ± standard deviation and the correspondingparameters that generate it.
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 26 / 29
Conclusions and Further Directions
ARI values can be sensitive to non-physiological artifacts present indata.
Saturation to maximal ABP and square wave pulses will alwayspredict an ARI value of 9 if present long enough.
Reduced pulse pressure and impulse artifacts are not as sensitive butcan still change ARI values up to 10% if present long enough.
Further DirectionsSensitivity of parameters of artifactsHow do multiple artifacts or repetition change ARI values?Implement detection system to filter out nonphysiological artifactsHypertensive vs. normotensive
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 27 / 29
Acknowledgements
Group 1 Participants:
P. Sang Chalacheva
Katrina Johnson
Greg Mader
Kevin O’Kee↵e
Grant Number: 1321794
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 28 / 29
Works Cited I
Dineen, Nicky E and Brodie, F G and Robinson, Thompson G and Panerai, Ronney B.Continuous estimates of dynamic cerebral autoregulation during transient hypocapnia andhypercapnia.Journal of applied physiology, 108(3):604–613, 2010.
Li, Qiao and Mark, Roger G and Cli↵ord, and Gari D.Artificial arterial blood pressure artifact models and an evaluation of a robust bloodpressure and heart rate estimator.Biomedical Engineering Online, 8(1):2009.
Panerai, Ronney B and Dawson, Suzanne L and Potter, and John F.Linear and nonlinear analysis of human dynamic cerebral autoregulation.American Journal of Physiology-Heart and Circulatory Physiology, 277(3): H1089–H1099,1999.
Pires, Paulo W and Ramos, Carla M Dams and Matin, Nusrat and Dorrance, and Anne M.The e↵ects of hypertension on the cerebral circulation.American Journal of Physiology-Heart and Circulatory Physiology, 204(12):H1598–H1614,2013.
Tiecks, Frank P and Lam, Arthur M and Aaslid, Rune and Newell, and David W.Comparison of static and dynamic cerebral autoregulation measurements.Stroke, 26(6):1014–1019, 1995.
Erica. M. Rutter, Adam Mahdi Artifact Influence on ARI JMM, 2017 29 / 29