Download Objective and Subjective Analysis of Left Ventricular

Objective and Subjective Analysis of
Left Ventricular Angiograms
By BERNARD R. CHAITMAN, M. D., HENRY DEMOTS, M. D., J. DAVID BRISTOW, M.D.,
JOSEF R6SCH, M.D.,
AND
SHAHBUDIN H. RAHIMTOOLA, M. B., F. R. C. P.
Downloaded from http://circ.ahajournals.org/ by guest on June 14, 2017
SUMMARY
In order to determine the reproducibility of analyses of left ventriculograms, 35 cineangiograms were
evaluated by four observers, two using standard quantitative techniques to determine ventricular volumes
and a newly devised quantitative system to evaluate wall motion and two others using only visual inspection
of the angiograms.
Objective analysis repeated by the same observer correlated well for end-diastolic and end-systolic
volumes and ejection fraction (r = .98, .99, .99, respectively) and only one of 105 (1%) wall segments were
identified differently. Variability in assessments increased when comparisons were made with a second objective observer. Correlation coefficients for the three volumetric parameters were .93, .98, and .95 and there
was disagreement in the assessment of 8% of wall segments. Wide variability was present between an objective and two subjective observers in analyses of end-diastolic volumes (r = .63, .64). Regional wall motion
was assessed differently in 19% and 27% of segments, respectively. Though the correlation of objectively
and subjectively determined ejection fractions was much better than the correlation for volume (r = .92,
.84), it was not as good as the correlation between two objective observers. Occasional errors of clinical
significance occurred.
We conclude that subjective analysis has a significant error rate and that reproducibility and accuracy of
analysis of left ventriculograms require objective analysis.
made by two experienced observers; 4) the variability
between assessments made by subjective and objective observers.
Materials and Methods
LEFT VENTRICULAR CINEANGIOGRAPHY is
widely used to determine the nature and severity
of cardiac disease and the effects of surgical and pharmacologic interventions on left ventricular function.' Quantitative methods for determination of ventricular
volumes have been described and their accuracy
validated previously.7 However, the determination of
left ventricular function is frequently made by visual
inspection of the cineangiogram rather than by
application of objective quantitative techniques. This
is especially true of assessments of regional wall motion because well established quantitative techniques
are not available.
In this study we have examined: 1) the reproducibility of analyses by objective techniques when
employed by the same observer; 2) the variation in
objective analyses performed by two different
observers; 3) the variation of subjective evaluations
Thirty-five left ventricular cineangiograms with good
opacification, adequate centering of the image and without
frequent extrasystoles were chosen for evaluation. There
were 23 men and 12 women with a mean age of 47 years. Six
patients had no demonstrable cardiac disease, one had an
ostium secundum atrial septal defect, and the remaining 28
had significant coronary artery disease demonstrated by coronary arteriography. Uniplane left ventricular cineangiography was obtained in the right anterior oblique position
at 60 frames/sec using 30 to 48 ml of meglumine diatrizoate. The injection rate varied from 10 to 15 ml/sec.
Objective evaluation of left ventricular volumes, ejection
fractions, and segmental wall movements were made by two
observers who traced end-diastolic and end-systolic frames.
The largest and subsequent smallest appearing ventricular
silhouettes of a beat that did not represent or follow an extrasystole were used in each instance. Volumes for each
silhouette were calculated by the area-length method.7 The
extent of magnification which was determined by filming a
lead impregnated ruler at the height of the apex of the heart
varied from 1.38 to 1.66. Systolic wall motion was assessed
using external and internal reference systems as described
previously.8 A segment was called akinetic when the wall
did not move during systole and dyskinetic when outward
(paradoxical) movement occurred during systole. To assess
hypokinesis a line representing the long axis of the ventricle
was drawn from the midpoint of the aortic valve to the apex
of both the end-diastolic and the end-systolic silhouettes. A
perpendicular chord was drawn one-fourth of the way from
the apex to the base of both silhouettes, thus dividing the
From the Divison of Cardiology, Department of Medicine and
the Department of Radiology, University of Oregon Health
Sciences Center, Portland, Oregon.
Supported in part by Program Project Grant HL 06336, Graduate
Training Grant HL 05791 and Research Grant HL 16461 from the
National Heart and Lung Institute. Dr. Chaitman is supported in
part by the Canadian Heart Fund Fellowship Training Grant.
Address for reprints: Henry DeMots, M.D., Division of Cardiology, University of Oregon Health Sciences Center, 3181 S. W.
Sam Jackson Park Road, Portland, Oregon 97201.
Received December 30, 1974; revision accepted for publication
April 21, 1975.
420
Circulation, Volume 52, September 1975
ANALYZING LV ANGIOGRAMS
421
Table 1
Synopsis of Observations Compared
Anterior Segment
Long Axis\-
Intraobserver variation
Objective vs objective
Interobserver variation
Objective vs objective
Subjective vs subjective
Inter method variation
Objective vs subjective
- Apical Segment
Inferior Segment
Figure 1
The ventricle is divided along its long axis and a chord three-fourths
of the distance from the base to the apex. The two apical areas are
combined, thus defining three wall segments for evaluation.
Observer 1 vs Observer 1
Observer 1 vs Observer 2
Observer Y vs Observer Z
Observer 1 vs Observer Y
Observer 1 vs Observer Z
Downloaded from http://circ.ahajournals.org/ by guest on June 14, 2017
tensive experience in an environment which had allowed
them to compare their subjective impressions with objective
data. These two observers, working independently, estimated end-diastolic volumes and ejection fractions and
graded the wall motion in each of the segments as normal,
hypokinetic, akinetic, or dyskinetic. Magnification was
judged by visual inspection of the framed ruler. Time for the
evaluation was not restricted so that multiple viewings of
each angiogram could be made.
The reproducibility of objective techniques was evaluated
by comparing the two objective assessments of Observer 1
and by comparing the assessment of Observer 1 with the
assessment of Observer 2. The reproducibility of subjective
techniques was determined by comparing the assessments of
Observers Y and Z. Variability between the objective and
subjective methods was determined by comparing the
assessment of each of the subjective observers with the
assessments of Observer 1 (table 1).
ventricle into anterior, inferior and apical segments (fig. 1).
The percentage decrease in area for the anterior, inferior
and combined apical segments was then determined for
each patient. This method of assessing hypokinesis does not
require the actual superimposition of the ventricular
silhouettes, but has the same effect as superimposing them
along the long axis. The normal percentage decrease in area
for each segment was established by determining the
decrease in area for these segments in seven patients who
had neither coronary artery disease nor hemodynamic
evidence of left ventricular abnormality. In these patients,
the decrease in area of the anterior segment was 48 ± 16%
(mean ±1 SD), the apical segment 64 + 18%, and the inferior segment 40 ± 14%. Hypokinesis was diagnosed if
neither akinesis or dyskinesis was present and if the percentage of systolic decrease in area was less than two standard deviations from normal.
Observer 1 analyzed each angiogram twice at intervals
ranging from one week to one year without knowing the
frame or beat used previously. Observer 2 selected 21 of 35
angiograms at random and analyzed them without knowing
the frame or beat analyzed by Observer 1. Both observers
noted the frame numbers of each silhouette traced so that it
could be determined whether significant variability could be
introduced by tracing different beats.
Subjective evaluation of all 35 angiograms was performed
by two experienced observers (Observers Y and Z) with ex-
Results
Intraobserver Variation
End-diastolic volume (EDV), end-systolic volume
(ESV), and ejection fraction (EF) were highly
reproducible when measured by the same observer.
The mean variation (+SEM) for these three parameters
was 4.8 ± .9 ml/m2, 4.7 + 1.4 ml/m2, and .021 ± .003
respectively (table 2). Wall motion was assessed identically on both occasions except that one (1%) of 105
.80 - r-.95
X .60 --
'z
.40
2 .20
80
120
LVEDV (mi/M2)
OBS.
160
20
60
LVESV (m1/M2)
OBS. l
100
y=.97x+3.95
Syx=4.49 .
/
N-21
* -. --Line of
identity
-
.20 40 .60 80
EJECTION FRACTION
OBS. I
Figure 2
A good correlation is shown between two objective observers. There is an apparent systematic tendency for Observer 2 to
estimate volumes lower than Observer 1.
Circulation, Volume 52, September 1975
422
CHAITMAN ET AL.
Table 2
Variation in Volumetric Assessments
Difference
of means
Standard
deviation
Range
1.0
.9
.01
7.4
4.7
.02
-29 to +12
-20 to +7
-.07 to .03
4.8
4.7
.02
10.4
5.4
.02
9.4
4.4
.04
-39 to +3
-13 to +2
-.95 to +11
10.8
5.77
8.1
.03
27.1
.06
19.6
.01
27.1
.09
13
18
Downloaded from http://circ.ahajournals.org/ by guest on June 14, 2017
Obs 1 and Obs 1
EDV (ml/m2)
ESV (ml/m2)
EF
Obs 1 and Obs 2
EDV (ml/m2)
ESV (ml1m2)
EF
Obs 1 and Obs Y
EDV (mi/M2)
EF
Obs 1 and Z
EDV (ml /M2)
EF
Obs Y and Obs Z
EDV (mi/rM2)
EF
.02
.08
Mean variance
(4 SEM)
-
.9
1.4
.003
.039
-
1.9
.8
.005
-96 to +28
-.27 to +.08
21
.049
-
3.1
.008
-113 to +20
-.19 to +.27
24.2
.068
-
-
3.9
.01
-63 to +18
-27 to +.15
14.7
.056
-
2.1
.005
-
The difference of the means is the difference of the mean value for the compared groups. The standard
deviation quantifies the scatter about these means. Mean variance is the average absolute difference between
repeated determinations.
segments judged normal on one determination
judged hypokinetic on another (table 3).
was
Interobserver Variation
Objective versus objective. The variation between
the average of two determinations by Observer 1 and
the single observation of Observer 2 was greater than
intraobserver variation. The variation (mean ± SEM)
for end-diastolic volume, end-systolic volume and
ejection fraction was 10.8 + 1.9 ml/m2, 5.7 ± .8
ml/m2, and .039 ± .005, respectively (table 2). Figure
2 demonstrates that there was good correlation
between the two sets of measurements (r = .93, .98,
.95) and also suggests that most of the variation occurred because volumes determined by Observer 2
were systematically smaller than those determined by
Observer 1. The variation when the same beat was
chosen for analysis is 9.8 ± 1.4 ml/m2 and when
different beats were chosen 11.4 ± 2.1 ml/m2. There
was disagreement in the classification of wall motion
for 5 (8%) of 63 segments examined (table 3).
Subjective versus subjective. Interobserver variation for the two subjective observers for EDV and EF
was 14.7 + 2.1 ml/m2 and .056 ± .005, respectively.
The correlation was better for EF than EDV but
varied more than data collected by the two objective
observers (table 2). There was disagreement in the
assessment of 25 (24%) of 105 wall segments (table 4).
Variation of Subjective and Objective Methods
Subjective assessment of left ventricular enddiastolic volume by Observers Y and Z varied considerably from the objective assessments made by
Observer 1 (table 2). Correlation of volume deterable 3
Variation Between Objective Assessments of Wall Motion
Ohserver 1
Normal
Hypokinetic
Akinetic
Dyskinetic
Normal
Ohserver 2
Observer 1
Normal
Hypokinetic
70
1
0
12
0
0
0
0
Observer 1
48
2
Hypokinetic
0
Akinetic
1
5
0
Dyskinetic
0
1
Akinetic
0
0
8
0
Dyskinetic
1
0
1
0
0
0
0
1
13
0
0
4
Italicized numbers denote instances of agreement and numbers not italicized instances of disagreement.
For quantitation of error rate, akinesis and dyskinesis were considered the same.
Circulation, Volume 52, September 1975
ANALYZING LV ANGIOGRAMS
423
Table 4
Variation Between Subjective Assessments of Wall Motion
Observer Z
Normal
53
2
0
1
Normal
Hypokinetic
Akinetic
Dyskinetie
Observer Y
Hypokinetic
12
15
1
0
Akinetic
0
7
2
1
Dyskinetic
0
2
2
7
Dyskinetic
0
2
Akinetic
0
0
2
2
14
7
4
1
0
4
4
3
1
0
1
8
Table 5
Variation Between Objective and Subjective Assessments of Wall Motion
Normal
62
Normal
Observer Y
Hypokinetic
2
Downloaded from http://circ.ahajournals.org/ by guest on June 14, 2017
Akinetic
Dyskinetic
Observer 1
Hypokinetic
9
10
2
0
a
0
0
9
Observer 1
Normal
Observer Z
54
1
0
1
Hypokinetic
Akinetic
Dyskinetic
minations was poor (r = .63, .64) but the correlation
of objectively and subjectively determined ejection
fractions was much better (r = .92, .84) (fig. 3). There
was discrepancy from the objective evaluations in the
assessments of regional wall motion in 19% and 27%
of instances (table 5). Although assessments regarding
wall motion and ejection fraction and wall motion abnormalities were correct in many instances (fig. 4), discrepancies of great clinical significance were occasionally made by one or both of the subjective
observers. Figure 5 demonstrates an example where
both subjective observers thought that the ejection
fraction was markedly reduced and that there was abnormal wall motion. Objective assessments
demonstrated that wall motion and ejection fraction
were normal.
Discussion
Standard techniques for the determination of left
ventricular volumes and ejection fraction have been
160
OBS. I = Measured
OBS.Y= Visuol
236
`
k120
Cr
.80)l _r =.92
y =.90x+2.51
. 60 - Syx=6.17
)~ N = 35
.40
)h
X 80
*
0
)A
iz .20
)
80
120
-
160
line of identity
.20
.40
.60
.80
EJECTION FRACTION
OBS.
LVEDV (mI/M2)
OBS.
Figure 3
The correlation between objective Observer 1 and subjective Observer Y was poor for volume determination but better
for estimation of ejection fraction. All volumes greater than 105 ml/m2 were underestimated by subjective Observer Y.
The results of correlating Observers 1 and Z are similar to those of Observers 1 and Y.
Circulation, Volume 52, September 1975
CHAITMAN ET AL.
424
o-Externol Marker
MW
Coronory
Arteriogram:
L. Main 70% Stenosis
RCA Occl
EGG: Non-specific ST-T Abnormolities
OBS. 2
OBS. I
Wall Motion:
I
OBS. 2
Anterior Wall
Apex
Inferior Wall
Downloaded from http://circ.ahajournals.org/ by guest on June 14, 2017
Ejection Fraction:
0.62
0.60
OBS. Y
OBS. Z
HK
IHK
HK
0.34
HK
HK
HK
0.42
Figure 4
Assessments of wall motion and ejection fraction were similar for all four observers in this example though one subjective
observer underestimated the ejection fraction.
titated the variability in the measurement of left ventricular function by subjective and objective methods
by comparing the assessments of different observers
examining the same angiogram.
Repeated assessments by the same observer at
widely variable intervals were highly reproducible.
When the second measurement was performed by a
different observer the variability increased slightly.
This was especially true of absolute volumes and less
developed and validated previously.7 Variation from
day to day in these parameters has been shown and
quantitated.9 The variation demonstrated may be attributed to biologic variation or to errors in measurement. This variability would be greater if
angiographic studies were performed in close temporal proximity because angiographic contrast
material is known to affect ventricular performance
and intravascular volume.'0-13 This study has quan-
0
/
TG
Coronary
Arteriogram:
/
LAD 80% Stenosis
L-Cx 80% Stenosis
RCA Occl
K-----
EGG: IMI
OBS. I
OBS. 2
Anterior Wall
Apex
Inferior Wall
HK
DK
HK
Ejection Fraction:
0.37
Wall Motion:
I
OBS. Y
OBSR. 7
HK
DK
HK
HK
DK
HK
HK
DK
HK
0.37
0.28
0.40
-J
Figure 5
A disparity of clinical significance between objective and subjective observers in the estimate of both ejection fraction and
wall motion is demonstrated in this example.
Circulation, Volume 52, September 1975
ANALYZING LV ANGIOGRAMS
Downloaded from http://circ.ahajournals.org/ by guest on June 14, 2017
true of ejection fraction where systematic errors in
end-diastolic and end-systolic volume measurements
tend to cancel one another. No significant variability
resulted from selection of different beats so the major
source of variability arose from tracing of silhouettes.
The extent of variation is much greater if subjective
assessments are used instead of objective
measurements. Subjective observers did well in
judgments of ejection fraction although each made
errors of judgment with definite clinical implications
(fig. 5). The subjective observers did less well in the
assessment of absolute ventricular volumes probably
because this requires the viewer to integrate the visual
image and the degree of magnification.
The disagreement between the two subjective
observers contrasts with the close agreement of the
two objective observers and suggests that the
difference between subjective and objective
measurements is not systematic but rather due to random variability in the assessments of subjective
observers.
Akinesis, hypokinesis, and dyskinesis are easily
defined in qualitative terms but several problems
become apparent in clinical application. Various
reference systems have been proposed for the crucial
step of superimposition of end-diastolic and endsystolic silhouettes.8' 14, 15 It is obvious that the choice
of technique for superimposition will influence the
determination of segmental wall motion. It is also apparent that the designation of hypokinesis is frequently arbitrary when it is defined only in qualitative
terms. Although designations of segment borderlines
in our study are arbitrary and more satisfactory
systems may be devised, our results demonstrate that
subjective evaluation of wall motion is highly variable
and that reproducibility in the assessment of wall motion is facilitated by using objective techniques.
Precise determinations of the effects of surgical or
pharmacologic interventions on wall motion require
quantitative analysis and a clear statement regarding
the method of superimpositon of silhouettes and
quantitation of segmental wall motion.
Each observer must be aware of his or her own
limitations in the assessment of left ventricular cineangiograms whether objective or subjective techniques are used. Error rates in the assessment of left
ventricular function vary from one individual to
Circulation, Volume 52, September 1975
425
another but our study demonstrates that error will be
minimized if objective techniques are used.
References
1. CHATTERJEE K, SWAN HJC, PARMLEY WW, SUSTAITA H,
MARCUS HS, MATLOFF J: Influence of direct myocardial
revascularization on left ventricular asynergy and function in
patients with coronary heart disease: with and without
previous myocardial infarction. Circulation 47: 276, 1973
2. REES G, BRISTOW JD, KREMKAU EL, GREEN GS, HERR RH,
GRISWOLD HE, STARR A: Influence of aortocoronary bypass
surgery on left ventricular performance. N Engl J Med 284:
1116, 1971
3. GRIFFITH LSC, ACHUFF SC, CONTI CR, HUMPHRIES JO,
BRAWLEY RK, GOTT VL, Ross RS: Changes in intrinsic coronary circulation and segmental ventricular motion after
saphenous vein coronary bypass graft surgery. N Engl J Med
288: 589, 1973
4. MCANULTY JH, HATTENHAUER MT, ROSCH J, KLOSTER FE,
RAHIMTOOLA SH: Improvement in left ventricular wall motion following nitroglycerin. Circulation 51: 140, 1975
5. BONCHEK LI, RAHIMTOOLA SH, CHAITMAN BR, ROSCH J,
ANDERSON RP, STARR A: Vein graft occlusion: Immediate
and late consequences and therapeutic implications. Circulation 50 (suppl II): 11-84, 1974
6. RAHIMTOOLA SH, EHSANI A, LOEB HS, ROSEN KM, GUNNAR
RM: Left ventricular function in the convalescent phase of
acute myocardial infarction. Chest 66: 333, 1974
7. SANDLER H, DODGE HT: The use of single plane
angiocardiograms for the calculation of left ventricular
volume in man. Am Heart J 75: 325, 1968
8. CHAITMAN BR, BRISTOW JD, RAHIMTOOLA SH: Left ventricular
wall motion assessed by using fixed external reference
systems. Circulation 48: 1043, 1973
9. MCANULTY JH, KREMKAU EL, ROSCH J, HATTENHAUER MT,
RAHIMTOOLA SH: Spontaneous changes in left ventricular
function between sequential studies. Am J Cardiol 34: 23,
1974
10. BROWN R, RAHIMTOOLA SH, DAvIs GD, SWAN HJC: The effects
of angiocardiographic contrast medium on circulatory
dynamics in man: Cardiac output during angiocardiography. Circulation 31: 234, 1964
11. RAHIMTOOLA SH, DUFFY JP, SWAN HJC: Ventricular performance after angiocardiography. Circulation 35: 70, 1967
12. RAHIMTOOLA SH, GAU GT, RAPHAEL MJ: Cardiac performance
after diagnostic coronary arteriography. Circulation 41: 537,
1970
13. KLOSTER FE, BRISTOW JD, PORTER GA, JUDKINS MP, GRISWOLD
HE: Comparative hemodynamic effects of equiosmolar injections of angiographic contrast materials. Invest Radiol 2:
353, 1967
14. HERMAN MV, HEINLE RA, KLEIN MD, GORLIN R: Localized
disorders in myocardial contraction. N Engl J Med 277: 222,
1967
15. McDONALD IG: The shape and movements of the human left
ventricle during systole. Am J Cardiol 26: 221, 1970
Objective and subjective analysis of left ventricular angiograms.
B R Chaitman, H DeMots, J D Bristow, J Rösch and S H Rahimtoola
Circulation. 1975;52:420-425
doi: 10.1161/01.CIR.52.3.420
Downloaded from http://circ.ahajournals.org/ by guest on June 14, 2017
Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright © 1975 American Heart Association, Inc. All rights reserved.
Print ISSN: 0009-7322. Online ISSN: 1524-4539
The online version of this article, along with updated information and services, is located on
the World Wide Web at:
http://circ.ahajournals.org/content/52/3/420
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally
published in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not
the Editorial Office. Once the online version of the published article for which permission is being
requested is located, click Request Permissions in the middle column of the Web page under Services.
Further information about this process is available in the Permissions and Rights Question and Answer
document.
Reprints: Information about reprints can be found online at:
http://www.lww.com/reprints
Subscriptions: Information about subscribing to Circulation is online at:
http://circ.ahajournals.org//subscriptions/