Ⅰ. 서 론
(blood pressure)
. (invasive)
(non-invasive)
[1].
(auscultatory method), (oscillometric method), (electronic palpation method) . ,
[2]
[3-4]
.
(systolic blood pressure, SBP),
(diastolic blood pressure, DBP)
(mean arterial pressure, MAP) (cuff) SBP DBP
[5]. SBP DBP
(characteristic ratio) (height-based) (slope-based) .
[6].
(artefact)
. Moraes[7]
신 준
Development of Blood Pressure Estimation Algorithm Using Variable Characteristic Ratios on Oscillometric Method
Joon Shin
Department of Mechanical and Automotive Engineering, Gangneung-Wonju National University (Received August 3, 2009. Accepted November 11, 2009)
In this paper, variable characteristic ratio algorithm based on oscillometric method is proposed to enhance the accuracy of blood pressure measurement. We combined the slope-based approach and fuzzy inference technique to change the characteristic ratios of height-based method. The proposed algorithm was assessed on 255 measurements from 85 subjects and compared with the conventional height-based algorithm. The testing results showed that the developed algorithm achieved an overall grade A for both systolic and diastolic blood pressures according to the BHS protocol. And, mean standard deviation between the observers and the developed algorithm were 5.71mmHg and 6.29mmHg for systolic and diastolic pressures respectively, which also fulfilled the AAMI criteria. In conclusion, this algorithm was successfully developed and recommended for further clinical trials with the wider adult population.
Blood pressure(혈압), Oscillometric method(진동법), Fuzzy inference (퍼지 추론), Variable characteristic ratio(가변 특성비)
Corresponding Author :
신 준(220-711)
강원도 원주시 흥업면 흥업리 강릉원주대학교 기계자동차공학부Tel : +82-33-760-8744 / Fax : +82-33-760-8741
E-mail : [email protected]
본 연구는
2008
년도 강릉대학교 장기해외파견연구지원에 의하여 수행되었음, Colak[8] - . Al-Jaafreh[9] PPG 2
, Huo[10]
, Lin[11]
. ,
, .
.
Ⅱ. 혈압추정 알고리즘 개발
A. 특징 특징 추출 특징 특징 추출 추출 추출
.
.
Fig. 1 (increasing
section) (decreasing section)
(SBP bas )
(DBP bas ) .
. Fig. 1
0.5~30Hz DC
2Hz (trough)
.
(heart rate) .
0.5~1.5
.
, (rule)
[12,13].
(SBP slp )
SBP bas
.
(DBP slp ) DBP bas
. Fig. 1
(MCP max ) MAP
,
3
A m p litu d e (V )
2.5 2 1.5 1 0.5 0 -0.5 -1 -1.5
Increasing section Decreasing section
Time (sec) 0
-2
5 10 15 20
MCP
max그림 1. 가압대 압력센서의 진동신호와 특징들의 정의 .
Fig. 1. Oscillation signal of cuff pressure sensor and definition of several features.
(MCP inc ) MCP max
.
(MBP dec ) MAP
. MCP inc
(p-p) ,
MBP dec
. ,
MCP dec /MCP max
MBP dec /MAP
MCP dec /MCP max (correlation value) 0.39
MBP dec /MAP 0.58 .
B. 퍼지 퍼지 퍼지 퍼지 추론 추론 추론 추론
.
[14].
0 1
(membership function) .
Fig. 2
(a) (b) 2
. Fig. 2 (a) x
3 “Negative
(NE)”, “Approximately Zero(AZ)”, “Positive(PO)”
, Fig. 2 (b) x
“Low(LO)”, “Medium(ME)”, “High
(HI)” .
, Fig. 3
”Negative Big(NB)”, “Negative Small(NS)”, “Approximately Zero(AZ)”, “Positive Small(PS)”, “Positive Big (NB)” 5
.
9
3 27
. SBP bas -SBP slp
±5mmHg ,
MCP inc /MCP max 27
0.669, 0.054 Fig. 2 (b)
.
0.687, 0.062
Fig. 3 .
2 1
M e m b e rs h ip v a lu e 1.0
0.0 NB
0.75 Systolic characteristic ratio
0.72 0.69 0.66 0.63
NS AZ PS PB
그림 3. 수축기 퍼지 출력변수에 대한 관계 함수 . Fig. 3. Membership function of fuzzy sets for systolic output.
1.0
M e m b e rs h ip v a lu e
0.0
NE AZ PO
10 0
-10
SBP
bas-SBP
slp(mmHg)
M e m b e rs h ip v a lu e 1.0
0.0
LO ME HI
0.77 0.67
0.57
MCP
inc/MCP
max(a) First input to systolic fuzzy inference engine (b) Second input to systolic fuzzy inference engine 그림 2. 수축기 퍼지 입력변수에 대한 관계 함수 .
Fig. 2. Membership functions of fuzzy sets for systolic input.
DBP bas -DBP slp MBP dec /MAP .
0.748, 0.034
0.690, 0.102
.
, 9 Table 1
Table 2 .
Mandani -
[15]. ,
.
Ⅲ. 실험 및 결과 고찰
A. 실험장치 실험장치 및 실험장치 실험장치 및 및 방법 및 방법 방법 방법
85 3
255
.
.
5 3
2 2 (3M Littmann,
USA)
±5mmHg
.
, DAQCard-AI-16XE
-50(National Instrument, TX, USA) 250
LabView .
NIBP (Welch Allyn, USA) 3~4mmHg .
, Fig. 4
. B. 실험결과 실험결과 실험결과 실험결과 고찰 고찰 고찰 고찰
2
. /
(American Association for the Advancement of Medical Instrumentation, AAMI) , A~
D (British Hypertension
Society, BHS) . AAMI
±5mmHg
±8mmHg . BHS 60%
input 1
input 2 NE AZ PO
LO NB NS AZ
ME NS AZ PS
HI AZ PS PB
a. Input 1은 SBP
bas-SBP
slp이다.
b. Input 2는 MCP
inc/MCP
max이다.
표 1. 수축기 특성비 추출을 위한 9 개의 규칙과 판단 테이블 .
Table 1. Decision table for systolic characteristic ratio and 9 fuzzy rules.
input 1
input 2 NE AZ PO
LO AZ NS NB
ME PS AZ NS
HI PB PS AZ
a. Input 1은 DBP
bas-DBP
slp이다.
b. Input 2는 MBP
dec/MAP이다.
표 2. 이완기 특성비 추출을 위한 9 개의 규칙과 판단 테이블 .
Table 2. Decision table for diastolic characteristic ratio and 9 fuzzy rules.
압력구간(상승, 하강) 설정 혈압 신호 입력
특징 추출 (MCP
inc, MCP
max, MBR
dec, MAP)
경사혈압(SBP
slp, DBP
slp) 계산 기본혈압(SBP
bas, DBP
bas) 계산
퍼지 추론
특성 비 추출
최종 혈압 계산
그림 4. 개발된 알고리즘에 대한 흐름도 .
Fig. 4. Flowchart of the developed algorithm.
±5mmHg 85% ±10mmHg
95% ±15mmHg A
. Table 3 BHS .
Fig. 5
Bland-Altman .
Fig. 5 (a) (b)
. , AAMI
0.15mmHg, 5.71mmHg
-0.21mmHg, 6.29mmHg . BHS
Table 4
Conventional
Curve-fitting
. ,
AAMI . ,
100mmHg 180mmHg
10% 60mmHg
100mmHg 10%
. 14.7% 100mmHg
17.8% 60mmHg
.
25cm 35cm
10% 25cm
12.8% 35cm .
AAMI
BHS A
.
.
Grade
Absolute difference between reference and test device (%)
< 5mmHg < 10mmHg < 15mmHg
A 60 85 95
B 50 75 90
C 40 65 85
D Worse than C
표 3. BHS 기준의 등급판정 기준 . Table 3. Grade criteria used by BHS.
20
15
10
5
0
-15
-20
60 70 80 90 100
Average BP of auscultatory & developed algorithm (mmHg) 110 120 130 140 150 160 Mean - 1.96SD
Mean + 1.96SD
Mean SB P D iff e re n c e (m m H g ) -5
-10
20
15
10
5
0
-15
-20
30 40 50 60 70
Average BP of auscultatory & developed algorithm (mmHg)
80 90 100 110
Mean - 1.96SD Mean + 1.96SD
Mean
D B P D iff e re n c e (m m H g )
-10 -5
(a) Systolic blood pressure (b) Diastolic blood pressure 그림 5. 개발된 알고리즘과 청음법 결과에 대한 Bland-Altman 도표 .
Fig. 5. Bland-Altman plot of the developed algorithm and auscultatory result.
Ⅳ. 결 론
. ,
1
. ,
. ,
BHS A
AAMI .
3
.
참고문헌