A NEW DATA COMPRESSION TECHNIQUE

FOR ELECTROCARDIOGRAMS

Murat ZVIN and Mehmet KORUREK

Technical U n i v e r s i t y of I s t a n b u l
E l e c t r i c a l and E l e c t r o n i c s Engineering Faculty
80626 Ayazaga, I s t a n b u l , TURKEY

A b s t r a c t : During t h e l a s t 20 years, a broad One of the scan-along-polygonal approximation

spectrum of d a t a reduction algorithms have been proposed techniques (SAPA-2) has got the t h e o r e t i c a l bases t h a t
f o r Electrocardiaogram (ECG) d a t a compression. Such a l - the d e v i a t i o n between t h e s t r a i g h t l i n e s and t h e o r i g i n a l
gorithms, t h a t o p e r a t e i n r e a l time have been v i t a l i n s i g n a l i s never more than t h e p r e s e t e r r o r t o l e r a n c e 131.
reducing t h e d i g i t a l ECG d a t a volume f o r s t o r a g e , t r a n s -
mission and a n a y s i s without l o s i n g c l i n i c a l information I n t h e p r e s e n t paper, an hybrid appritaeh based on
c n n t e n t . I n t h i s paper,a new approach f o r real-time ECG two o t h e r d i r e c t ECG d a t a compression algorithms, ADAP-
d a t a compression i s presented. This technique i s a h y b r i d TIVE and SAPA-2, i s a p p l i e d t o d i f f e r e n t single-lead ECG
of w e l l known techniques c a l l e d SAPA-2 and Adaptive signals.
compression algorithm. A s p e c i a l decompression algorithm
and smooth-curve approximation a r e used t o r e c o n s t r u c t
t h e compressed d a t a i n t o o r i g i n a l form. I n o r d e r t o
compare t h i s new approach on t h e same b a s e s with o t h e r Data A c q u i s i t i o n
well known techniques which a r e widely used i n prac-
t i c e , a l l t h e techniques a r e a p p l i e d t o t h e same s i n g l e The d i f f e r e n t s i n g l e l e a d analog ECG s i g n a l s ,
l e a d ECG s i g n a l . The comparison i s given i n a t a b l e . which a r e obtained from an ECG s i m u l a t o r , a r e converted
by a PCL-812 A/D c o n v e r t o r , i n o r d e r t o process with a
Introduction PC computer.

Compression of d i g i t a l E C G ' s in a more r e l i a b l e , Preprocessing

a c c u r a t e and e f f i c i e n t way has received g r e a t a t t e n t i o n
because of continuing p r o l i f e r a t i o n of computerized ECG The o r i g i n a l sampled d a t a t o b e approximated a r e
processing systems along w i t h t h e increased f e a t u r e per- r e p r e s e n t e d by the d i s c r e t e sequence X(i) where t h e in-
formance requirements and demand f o r lower c o s t medical t e g e r i is t h e sample number. Denote hx and X+ as
care. maximum and minimum value of samples, of which i n i t i a l
Data compression i s t h e process of d e t e c t i n g and v a l u e s a r e set t o be e q u a l t o t h e f i r s t sample X(1).
e l i m i n a t i n g redundancies i n a given d a t a s e t and has the L e t +E(%) r e p r e s e n t allowable, adaptive e r r o r i n t h e
main goal t o achieve maximum d a t a volume r e d u c t i o n while approximation and T( i ) a d a p t i v e t h r e s h o l d . When begining
preserving t h e s i g n i f i c a n t s i g n a l morpology f e a t u r e s t h e approximation algorithm, t h e f i r s t d a t a p o i n t i s
upon r e c o n s t r u c t i o n . The p r a c t i c a l importance of d a t a s t o r e d . Upon r e c e i p t of l a t e r salqple a t i + n , x(i+n) w i l l
compression i n computerized e l e c t r o c a r d i o g r a p h y can be be compared w i t h hX and &inand the normalized s l o p e s
shown i n following a s p e c t s : 1) increased s t o r a g e c a p a c i t y of t h e t h r e e l i n e s j o i n i n g X(i) t o X(i+n)+E and t o
of ECG's a s databases f o r subsequent comparison o r X(i+n)-E and f i n a l l y t o X(i+n) a r e c a l c u l a t e d a s ;
e v a l u a t i o n 2) f e a s i b i l i t y of t r a n s m i t t i n g r e a l time
E C G ' s over t h e p u b l i c phone network 3) implementation of
c o s t e f f e c t i v e real-time rhythm algorithms 4 ) economical
r a p i d transmission of o f f - l i n e ECG's over p u b l i c phone
l i n e s t o a remote i n t r e p r e t a t i o n c e n t e r and 5) improved and
f u n c t i o n a l i t y of ambulatory ECG-monitors and r e c o r d e r s
Ill.
M3 = X(i.n)-x(i)
n
where (n = 1,2, ...)
The e x i s t i n g ECG d a t a compression methods can be I f t h e new sample X(i+n) i s g r e a t e r then bx, then
c l a s s i f i e d i n t o t h r e e c a t a g o r i e s : 1 ) d i r e c t d a t a handling hX
i s made equal t o t h e value of t h e sample. I f t h e
methods, 2) transformation methods and 3) parameter e x t - sample i s less then b i n , t h e n b i n i s made equal t o t h e
r a c t i o n methods 111, 12 I. I n t h i s paper, some of d i r e c t v a l u e of t h e sample. These p r o c e s s e s a r e repeated while
d a t a handling methods such a s AZTEC, TP, SAPA-2,ADAPTfVE -
t h e c o n d i t i o n s Ml>M3>M2 and Lx G i n < T ( i ) a r e v a l i d .
a r e compared with t h e proposed new approach. Whenever one of t h e s e c o n d i t i o n s a r e v i o l a t e d , t h e
inmediately previous d a t a p o i n t is s t o r e d a s ( b a n , t )
The amplitude zone time epoch coding (AZTEC) a l - where ,X, i s t h e amplitude of t h e l i n e c a l c u l a t e d a s
gorithm, developed f o r preprocessing real-time E C G ' s f o r t h e average value of samples X(i) i n t h e c o r r e n t region,
rhythm a n a l y s i s , became a popular d a t a r e d u c t i o n algo- and t is t h e l e n g t h of t h e l i n e r e p r e s e n t i n g t h e number
rithm f o r ECG monitors and d a t a b e s e s with an achieved of samples i n t h e region.
compression r a t i o of 1 O : l (500 Hz samped ECG with 12 b i t
1,
r e s o l u t i o n ) 11 121. However, t h e r e c o n s t r u c t e d s i g n a l The adaptive approximation e r r o r E ( i ) and t h r e s -
demonstrates d i s c o n t i n u i t i e s and d i s t o r s i o n , which occurs hold T ( i ) depend on t h e n a t u r e of t h e s i g n a l . The v a l u e s
i n t h e r e c o n s t r u c t i o n of t h e P and T waves due t h e i r of E ( i ) and T ( i ) f o r t h e low information regions ( t h e
slow varying s l o p e s . base l i n e ) w i l l be g r e a t e r then t h e v a l u e s f o r high in-
formation r e g i o n s (Pand T waves, ST segments, QRS complex
Adaptive compression a l g o r i t h m i s a modified o r tachycardia s i g n a l s ) .
AZTEC algorithm, whereby the e r r o r t h r e s h o l d i s made
a d a p t i v e t o the ECG s i g n a l v a r i a t i o n s 12 I.
This technique I n o r d e r t o r e c a l c u l a t e the a d a p t i v e E ( i ) and
has r e s u l t e d i n a s l i g h t improvement ( b e t t e r compromise T ( i ) values, t h e algorithm c a l c u l a t e s permeanently t h e
between compression r a t i o of 1 O : l and r e c o n s t r u c t e d s i g - following s t a t i s t i c a l parameters:
nal fidelity.)

InternationalBiomedical Engineering Days, 1992 / 92 $3.00 0 1992 IEEE

0-7803-0743-7 61
 I n o r d e r t o obcain a good compressibn w i t h o u t
l o o s i n g t h e o r i g i n a l form of i n p u t d a t a h i g h CR and
low PRD a r e d e s i r e d . In o r d e r t o compare t h i s new hyb-
r i d Technique w i t h o t h e r compression a l g o r i t h m s such
a z Turning P o i n t (TP), AZTEC, SAPA and ADAPTIVE t e c h -
n i q u e s , we a l s o a p p l i e d t h e a l g o r i t h m s t o the same ECG
s i g n a l a n d gave t h e CR and PRD r e s u l t s of t h e a l g o r i t h m s
f o r d i f f e r e n t h e a r t r a t e s , i n T a b l e s 1 and 2 -

T a b l e 1- CR and PRD r e s u l t s of t h e compression

t e c h n i q u e s f o r 60 bpm ECG d a t a ( p , i T l i n g
frequency =500 Hz, r e s o l u t i o n 12 b i t s )

Technique ! CR ' PRD

TP 2.0 / 4.8

AZTEC 10.0 ! 8.7

SAPA-1 9.8 ' 8.2

SAPA-2 9.3 7.6

1 :,,L;)LivL, ap;>rottinition e r r o r and t h r e s h o l d ' d e i i n e d 2s
ADAPTIVE 10 .o 8.8
:( i) :( i-l)-a- (Cf( i ) -Cf
T( i ) --.C(i-l)-a:
( i-lk ( i-1)

(C, ( i ) -Cf( i-IYT( i-1)

(5)
(6)
HYBRID i 10.5 1 7.0

wllerc t h e c o n s t a n t s a , and a, can be determined empri-

C d l ly. Table 2 - CR and PRD r e s u l t s of the compression
techniques f o r 120 bpm ECG d a t a .
-
Decompress ion Algorithm

Compressed d a t a i s a sequence o i p a i r s o i d a t a
Technique 1 CR PRD
(.b,, t) . k c ~ u s c of s t e p l i n e q r r a n t i z a t i o n , t h e I
2 . 1 ~t o~ t e d i s p l a y e d won't be a c c e p t a b l e t a r p h y s i c i a n s . TP I 2.0 4.8
i11~~r,~to3 r c ,curve-smootiting a l s o r i t h r n t o p r o c e s s t h e
AZTEC 5.2 7.5
t - o r ; > r e s s z dd a t a is a p p l i e d , which uses t h e i o l l o w i n p
,.ormu l a : SAPA-1 5.0 7.5
SAPA-2 5.8 7.6
:( i) &(-?S( i - j ) - j x ( i - 2 ) - 5 ~ (~ - L ) - I ~ s ( i )
30
ADAPTIVE 4 .9 8.0
-6S( i - l ) - j S ( i - ? ) - ? S ( i - j ) ) (7)

./U :.c : i i ) is nev d a t a p o i n t , while X ( i - n )

'1 31-c the
HYURID 1 7.2 I 7.6

: i:,i.ial data points.

fi
R-__
esults I' ORIGINAL. MTA

: ; ! c d c s c r i b e d mc'tiicd w i t h v a r i a b l e -ipproximat ion

.rror and e r r o r and T h r e s h o l d v a l u e s has been a p p l i e d
t o t h e compression of d i f f e r e n t s i n g l e l e a d ECG s i g n a l s
sampled a t 500 Hz w i t h a 12-bit A/D c o n v e r t e r . The app-
roximation and t h r e s h o l d v a l u e s a r e l i m i t e d t o h e b e t -
ween 1.5-30 and 2-13 and t h e c o n s t a n t s a , and a 2 have
been determined e m p i c a l l y between 0,7-1,l and 0,05-0,2.
Other a l g o r i t h m s have been a l s o a p p l i e d t o t h e same ECG
s i g n a l s i n o r d e r t o compare on t h e same b a s e s w i t h t h e
proposed a l g o r ithm.

I n a d d i t i o n to v i s u a l i n s p e c t , t h e performance
index P R D ( t h e p e r c c n t root-mean square d i f f e r e n c e ) i s
computed a s f o l l o w s ;

I
ADAppNE CR: 10.0
Pw: 8.8
1
4
wticrt. X ( i ) and XreC(i) a r e samples of t h e o r i g i n a l and
r e c o n s t r u c t e d d a t a sequences. The compression r a t i o
(CR) i s given by;

CR Number of samples of o r i g i n a l d a t a
Number of samples of compressed d a t a
(9)
)L '-

F i g u r e 1. O r l g i n a l and r e c o n s t r u c t e d ECG s i g n a l s

\
62 I
 To v i s u a l i n s p e c t , the o r i g i n a l and r e c o n s t r u c t e d
ECG s i g n a l s a r e given i n Figure 1. As shown i n Figure 1
t h e new hybrid technique g i v e s the h i g h e s t CR and t h e
lowest PRD compared with o t h e r technique r e s u l t s . By
changing t h e t h r e s h o l d parameters of t h e p r e s e n t e d hyb-
r i d compression algorithm, d i f f e r e n t CR and PRD r e s u l t s
of t h e a l g o r i t h m a p p l i e d t o same ECG d a t a a r e obtained
and the r e s u l t s a r e shown i n Figure 2.

ORIGINAL wm

Figure 2. A p p l i c a t i o n r e s u l t s of p r e s e n t e d hybrid
compression a l g o r i t h m f o r d i f f e r e n t t h r e s h o l d
parameters.

bnc 1 7 s ion

An a d a p t i v e real-time algorithm f o r on-line com-

p r e s s i o n and transmission of ECG d a t a i s presented. The
algorithm combined the b e s t f e a t u r e s of two o t h e r tech-
niques c a l l e d SAPA-2 and Adaptive compression technique.
On a database of 650 electrocardiograms sampled a t 500
Hz, 1 2 b i t p r e c i s i o n , we obtained compression r a t i o s
between 8-13 and PRD's below 10, and without l o s i n g c l i -
n i c a l important content.

References

I11 S.M.S. J a l a l e d d i n e , C.G.Hutchens e t . a l . , "ECG data

compression techniques -A u n i f i e d approach", IEEE
T r . on BME, Vo1.37, pp. 329-340, April 1990.

121 B . Furht and A.Perez, "An Adaptive Real-time ECG

cornpre s s ion a l g o r i t h m with v a r i a b l e threshold" ,
IEEE T r . on BME, Vo1.35, pp.489-493, June 1988.

I31 M.Ishijima, S.B. Shin e t . a l . , "Scan -Along poly-

gonal approximation f o r d a t a compression on
electrocardiograms", IEEE T r . on BME, V01.30, pp.
723-728, nov. 1983.

Acknowledzement

This work i s supported by Turkish S t a t e Plan-

nig Organization.

63