LV WALL MOTION ANALYSIS/Gelberg et al.

15. Zissermann D, Strand EM, Rogers W: On-line digital analysisof the left ventricular pressure waveform. Proc 27th ACEMB:364, 1974

16. Dodge HT, Sandler H, Ballew DW, Lord JD: The use ofbiplane angiocardiography for the measurement of left ven-tricular volume in man. Am Heart J 60: 762, 1960

17. Feild BJ, Russell RO Jr, Dowling JT, Rackley CE: Regionalleft ventricular per.ormance in the year following myocardialinfarction. Circulation 46: 679, 1972

18. Cohn PF, Liedtke AJ, Serur J, Sonnenblick EH, Urschel CW:Maximal rate of pressure fall (peak negative dP/dt) during ven-

tricular relaxation. Cardiovasc Res 6: 263, 197219. Weisfeldt ML, Scully HE, Frederiksen J, Rubenstein JJ,

Pohost GM, Beierholm E, Bello AG, Daggett WM:Hemodynamic determinants of maximum negative dP/dt andperiods of diastole. Am J Physiol 227: 613, 1974

20. Langer GA: Ion fluxes in cardiac excitation and contractionand their relation to myocardial contractility. Physiol Rev 48:708, 1968

21. Bodenheimer MM, Banka VS, Trout RG, Hermann GA,Pasdar H, Helfant RH: Local characteristics of the normal andasynergic left ventricle in man. Am J Med 61: 650, 1976

Quantitative Left Ventricular Wall Motion Analysis:A Comparison of Area, Chord and Radial Methods

HARRIS J. GELBERG, M.D., BRUCE H. BRUNDAGE, M.D.,

STANTON GLANTZ, PH.D. AND WILLIAM W. PARMLEY, M.D.

SUMMARY We were interested in divising a relatively simple quantitative technique that could be used on a

routine clinical basis for wall motion analysis. Three quantitative methods of left ventricular (LV) regionalanalysis were compared in the 300 right anterior oblique and 60° left anterior oblique projections. The controlgroup consisted of 17 patients with qualitatively normal LV wall motion; the abnormal group comprised 17patients with at least one region of severe, qualitative wall motion abnormality. Normal regional values were

determined for area, chord and radial methods by applying the techniques to the ventriculograms of the controlgroup. Each technique was then applied to the abnormal group's ventriculograms to determine the percentageof qualitatively abnormal regions not detected by each method. The area method had the lowest failure rate(p < 0.001) and the best separation of measured normal and abnormal regions' ejection changes (p < 0.001),and best reflected symmetric uniform motion of the ventricular silhouette. We conclude that the area method,of the techniques examined, was best for the quantitative analysis of LV wall motion abnormalities.

LEFT VENTRICULAR contractile abnormalitiescan be an important manifestation of coronary arterydisease. These wall motion changes may representischemia or infarction of myocardium."3 Quantifyingthe extent of regional wall motion abnormality mayaid in determining the myocardial effects of coronaryartery disease. It would also simplify analysis of wallmotion changes after diagnostic and therapeutic in-terventions and would permit comparison of differentimaging techniques to assess their diagnostic ac-curacy. Often, wall motion evaluation is done on asubjective basis; however, subjective, qualitativeanalysis has a substantial error rate. An objective,quantitative system is required for accurate regionalevaluation.4 No quantitative technique is universally

From the Adult Cardiac Catheterization Laboratory, Car-diovascular Research Institute, and the Cardiovascular Division ofthe Department of Medicine, University of California, San Fran-cisco, California.

Supported in part by NHLBI Program Project grant HL 06285.Dr. Glantz is the recipient of an NIH Research Career Develop-

ment Award.Address for reprints: Bruce H. Brundage, M.D., Room 1186

Moffitt Hospital, University of California, San Francisco, Califor-nia 94143.

Received March 28, 1978; revision accepted December 4, 1978.Circulation 59, No. 5, 1979.

accepted, many of the more recent methods requiresophisticated computer facilities,5 and normalsegmental wall motion values have not been es-tablished for simultaneous 30° right anterior oblique(RAO) and 60° left anterior oblique (LAO) views. Weevaluated three commonly used quantitative methods(chord, radial and area) of wall motion analysis todetermine which technique best detects abnormalfunction. The methods chosen are relatively simpleand therefore suitable for routine clinical use. Thestudy was performed by analyzing simultaneous 300RAO and 600 LAO projections of contrast left ven-triculograms by each method.

Methods

Technically adequate cineventriculograms ofpatients evaluated for possible coronary artery diseasewere included in this investigation. Ventriculogramswere performed using a General Electric 16 mmcinegraphic biplane system at 60 frames/sec withalternate firing of the cameras. The patient waspositioned in the 30° RAO to the anterior-posteriortube and 60° LAO to the lateral tube. We performedleft ventriculography using 0.8 ml/kg of Renografin-76 to a maximum dose of 60 ml injected in 4 seconds.After ventriculography, each patient had selective cor-onary angiography by the Judkins technique.

991

by guest on January 11, 2016http://circ.ahajournals.org/Downloaded from

VOL 59, No 5, MAY 1979

Before drawing the ventricular outlines, two car-diologists simultaneously evaluated the films subjec-tively according to the American Heart Associationsystem for segmental left ventricular wall motionassessment. Each of seven anatomical regions wasscored as normal, mildly, moderately or severelyhypokinetic, akinetic, aneurysmal, or dyskinetic (fig.1).6 Cineventriculograms of poor quality leading tounknown function of a region were excluded from thisstudy. Ventriculograms of sinus beats were traced atend-systole and end-diastole. Premature and post-premature beats were not used in this study. The ven-tricular silhouette with the largest total area im-mediately before mitral valve closure was traced forend-diastole. The end-systolic silhouette was definedas that which had the smallest area before mitral valveopening. These maximum and minimum areas weredetermined by visual inspection. The traced silhouettewas then compared with several preceding and follow-ing cine frames to insure proper selection. The samecontraction was used to trace the RAO and LAOviews. Several beats were evaluated in both views toensure that the selected silhouettes were representativeof the end-systole and end-diastole in each patient. Allventriculograms were performed with the patientsholding their breath at end-inspiration to minimizecardiac rotation about the long axis. No patient in thestudy demonstrated diaphragmatic motion betweendiastole and systole.Two groups were studied. The control group com-

prised 17 patients with qualitatively normal LV wallmotion. These patients had normal coronary arteriesor only minor luminal irregularities on selectiveangiography, as well as normal hemodynamics, end-diastolic and end-systolic LV volumes and normalejection fractions. All control patients had normal

chest x-rays and ECGs. None of these patients hadstenotic or incompetent valvular lesions. This controlgroup was considered to represent a normal popula-tion. The abnormal group consisted of 17 patientswith at least one region of severe, qualitative wall mo-tion abnormality. To minimize the problems inherentin this subjective selection process, only segmentsclassified as severely hypokinetic, akinetic,aneurysmal or dyskinetic were analyzed by the quan-titative methods. The patients chosen for the abnor-mal group also had one or more of the following: 1) adocumented myocardial infarction; 2) Q-wave infarctpattern on the ECG; or 3) >90% occlusion of the cor-onary artery supplying the region of severe, qualita-tive wall motion abnormality. Thirteen patients hadall three associated findings. One had an infarcthistory and a critical coronary lesion, but no elec-trocardiographic pattern. One patient had infarctionby ECG and a coronary occlusive lesion, but nohistory of infarction. In two patients with severequalitative regional abnormalities there was no historyof infarction or abnormal Q waves on the ECG, butthere was >90% occlusive coronary lesions.We considered three methods for evaluating

regional wall motion. Each system defined the majoraxis in the RAO view as the line from the apex to themidpoint of the aortic valve plane; in the LAO viewthe major axis for each technique was a line from theaortic-mitral valve junction to the apex. The apicalregion furthest away from the aortic valve plane mid-point (RAO) and aortic-mitral valve junction (LAO)was taken as the apex point. All reference systemswere applied to the end-systole and end-diastolesilhouettes independently, and were thus corrected forany motion of the heart in space except rotation.Figure 2 shows the format for each system. In the

AMERICAN HEART ASSOCIATION QUALITATIVEREPORTING SYSTEM FOR THE LEFT VENTRICULOGRAM

ANTEROSASAL

DYS El EdT_ ANTEROLATERAL

ED I

APICAL

EE1EWPOSTEROBASAL N HYP AK

EDI ED DIAPHRAGMATICDYS Eli E

FIGURE 1. American Heart Association qualitative reporting system for wall motion assessment:N = normal, HYP = hypokinetic; AK = akinetic; DYS = dyskinetic; A = aneurysmal; U = unknown.Hypokinetic regions were further characterized in this study as mild, moderate or severe. By permission ofthe American Heart Association, Inc, and the authors.

992 CIRCULATION

by guest on January 11, 2016http://circ.ahajournals.org/Downloaded from

LV WALL MOTION ANALYSIS/Gelberg et al.

RADIAL SYSTEM

A

CHORD SYSTEMFIGURE 2. Formats for techniques ex-amined. A) radial system with 14 radial linemeasurements; B) chord system with eightrectilinear line measurements; C) areasystem with eight area measurements. Foreach technique, measurements were taken atend-diastole and end-systole, and the per-cent change was then calculated.

B

C

radial system (fig. 2A) lines are drawn from the mid-point of the major axis in the RAO and LAO views tothe ventricular outline at 450 intervals. For the rec-tilinear chord system (fig. 2B) in the RAO view, per-pendiculars are drawn from the points that divide themajor axis into thirds to the points of intersection ofthe ventricular outline. In the LAO view, perpendic-ulars are drawn from the midpoint of the major axis.

In the area method (fig. 2C) we constructedplanimetered regions by drawing perpendiculars frompoints that divide the major axis into thirds in theRAO view. The two apical areas were combined andtreated as one region. In the LAO view, a perpen-dicular was drawn from the midpoint of the long axisto the posterolateral wall, dividing the ventricularsilhouette into three areas. The chord and radial tech-

993

by guest on January 11, 2016http://circ.ahajournals.org/Downloaded from

VOL 59, No 5, MAY 1979

niques measured systolic regional motion as the per-cent axis shortening from end-diastole to end-systole.The area system measured systolic change as thesegmental ejection fraction (i.e., the differencebetween diastolic and systolic regional areas dividedby the diastolic regional area).

Each system was applied to ventriculograms fromabnormal patients to determine the percentage ofmarkedly abnormal regions not detected by each tech-nique. This percentage is defined as the failure rate.Only severely hypokinetic, akinetic, dyskinetic andaneurysmal segments were considered for this deter-mination, since subjective, qualitative analysisappears to characterize severely abnormal defectsmost accurately.4 Each method was defined as detect-ing abnormality if any of the percent ejection changesfor the region were below the 97% confidence intervaldefined by the normal group. Since there were 17patients in our normal group, this cutoff point is twostandard deviations below the mean of the normalgroup.* This strict definition of lower limits of normal(i.e., 97% confidence level) was used, since we testedonly for detection of severe abnormalities.The quantitative techniques (figs. 2A, 2B and 2C)

sometimes subdivided areas defined by the qualitativeanalysis (fig. 1). Therefore, if any quantitativemeasurement within a region defined by figure 1 wasabnormal, the quantitative technique was credited fordetecting abnormality. For example, abnormalshortening of R12, R13 or R14 of the radial methodmeant that this technique detected abnormal septalwall motion.

Statistical Methods

In normal patients, different chords (C), areas (A)or radial (R) measurements changed differently withejection. We were interested in separating normalsegments from abnormal segments by comparingdifferent methods and different segments within amethod. This comparison requires that values be com-parable. Therefore, we normalized all measurementsaccording to the following scheme:

1) Compute the mean percent ejection change (M)for the 17 normal patients for each region (1, 2, 3 . . . )by each method (R, C, A): MRl, MR2, MR3 . . .MR14;MC1, MC2, MC3 MC8; MA1, MA2, MA3 . .- MA8.

2) Compute the standard deviation, (s) for the ejec-tion changes for the 17 normal patients for each regionby each method: SR1, SR2, SR3 . . SR14; SC1, SC2, SC3SC8; SA1, SA2, SA3 SA8.

3) Normalize the regional ejection change, ob-tained with m in region r, Xmr, according to

Z _ Xmr-Mmrmr

5mr(1)

In other words, Zmr represents the number of normal

standard deviations by which the measurement Xvaries above or below the mean of the 17 normalpatients; Z is dimensionless.

For example, a) normal: if the anterolateral regionby the area technique has a mean ejection change of57.2 ± 13.5% (SD) and one illustrative normal patienthas an area ejection change in this region of 69%, then

z 69% - 57.2% = +0.87 (2)

b) abnormal: in a patient with qualitatively abnormalanterolateral wall motion, an area ejection change was11%, SO

11% - 57.2%13.5%

In other words, the more negative Z, the worse theregional function. The lower limit of normal was de-fined as two standard deviations below the normalmean (i.e., Z < -2).

Each method was applied to 46 severely abnormalregions defined on qualitative analysis. We questionedif any method produced significantly more negative Znumbers than the other methods. This can be tested bythe null hypothesis: "The values of Z in qualitativelyabnormal segments are the same in all three methodsof quantifying regional wall motions." To test thishypothesis, we used the two-way analysis of variance(method fixed, patient random). After rejecting thisnull hypothesis, we used the Student-Newman-Keulsmultiple range test to define which method produced,on the average, more negative values for Z in the ab-normal regions.The next question addressed was, "Does any

method detect a significantly higher fraction of themarkedly abnormal regions?" The null hypothesis is:"All methods are equally likely to detect abnormalregions." To test this hypothesis, we constructed a2 x 3 contingency table. After rejecting this nullhypothesis, we subdivided x2 by first testing the nullhypothesis: "The chord and radial methods areequally likely to detect an abnormal region." We didnot reject this hypothesis. We then pooled the chordand radial failure rates and tested the null hypothesis:"The chord and radial methods, taken together, are aslikely to detect abnormal regions as the area method."We rejected this hypothesis.To quantify the amount of scatter in values of the

different methods used to measure regional ejectionchanges, we examined the sizes of the standard devia-tion for the different methods as a fraction of the meanchange. This coefficient of variation, defined for eachregion, is

Vmr Smr

Mmr(4)

where smr is the standard deviation of a method in a

region and Mmr is the mean of the same method inthat region. A large number means the spread in

*The cutoff point for the 95% confidence interval is -1.75 [p <0.05 (17)].

994 CIRCULATION

(3)

by guest on January 11, 2016http://circ.ahajournals.org/Downloaded from

LV WALL MOTION ANALYSIS/Gelberg et al.

values of the measure of segmental wall motion islarge by comparison with the mean value for thatregion. The smaller the coefficient, the more unifor-mity among the mean regional ejection changes of atechnique. This procedure permits us to compare thedifferent methods even though the intrinsic units aredifferent.We decided that if a region was quantitatively below

97% of the normal population, we would consider itabnormal. We used this strict confidence level becausewe tested only severely abnormal regions. Accordingto the one-tail distribution with 16 degrees of freedom,Zmr for a normal population will be below -2 in <3%of cases. Therefore, we called a segment abnormalwhen Zmr was below -2 (i.e., two or more normalgroup standard deviations below the normal groupmean).

Results

Normal mean regional ejection changes for radial,chord and area techniques are presented in table 1.The lowest mean ejection changes of all systems oc-curred in regions that incorporated aortic and mitralvalve structures. Area 5 had the lowest normal meanarea ejection fraction, and much of its perimeter wasformed by the posterior mitral valve and aortic valveplane. Chords 3 and 7 terminate at the aortic valveand have the lowest axis shortening by the rectilineartechnique. For the radial system, numbers 4, 7 and 14had the lowest values. Radial 4 incorporated the aorticvalve plane, 14 fell on or near the aortic valve plane,and 7 tended to incorporate the posterior mitral valve.The coefficients of variation listed in table 1 repre-

sent the standard deviations relative to the ejection

TABLE 1. Comparative Regional Ejection Changes in 17 Normal Patients

LowerEjection change limit of Coefficient

Method (M) Region (r) Mean (SD) normal* of variation

1.

2.

3.4.

5.6.

7.8.

9.10.11.12.

13.

14.

1.2.3.4.5.6.

7.8.

1.2.3.4.5.6.

7.8.

AnterobasalHigh anterolateralLow anterolateralApicalAnterior diaphragmaticMid-diaphragmaticPosterobasalHigh posterolateralMid-posterolateralLow posterolateralApical/inferiorLow septalMid-septalHigh septal

AnterobasalAnterolateralApicalDiaphragmaticPosterobasalPosterolateralApical/inferiorSeptal

AnterobasalAnterolateralApicalDiaphragmaticPosterobasalHigh posterolateralLow posterolateralSeptal

43.544.0

38.424.139.735.223.4

33.530.730.727.138.229.214.8

46.143.224.147.737.230.727.129.2

58.857.260.1

54.139.555.151.149.2

13.517.1

20.07.8

14.312.511.910.711.910.6

5.717.122.220.1

15.420.97.8

21.111.6

11.95.7

22.2

12.213.512.37.7

11.1

10.911.811.1

16.5

9.8- 1.6

8.511.1

10.2- 0.4

12.1

6.99.5

15.74.0

-15.2-25.4

15.31.48.55.5

14.0

6.915.7

-15.2

34.430.2

35.538.717.333.3

27.527.0

0.31

0.390.520.32

0.36

0.360.510.320.390.350.210.450.761.36

0.33

0.480.320.440.310.390.210.76

0.210.240.200.140.280.200.230.23

*At the 97% confidence level by the one-tail t distribution.

Radial (R)

Chord (C)

Area (A)

995

by guest on January 11, 2016http://circ.ahajournals.org/Downloaded from

VOL 59, No 5, MAY 1979

change means (Vmr = smr/Mmr) for each region by allmethods. The area technique, when compared with theother methods, had the smallest coefficients of varia-tion for every region of the ventricle and thereforedemonstrated the lowest relative variation about themeans in the 17 normal patients.The area technique had the greatest uniformity

among the mean regional ejection changes. The meanregional ejection changes for the different regions were(mean ± SD): radial, 32.32 ± 8.29; chord, 35.66 ±

9.15; area, 53.13 ± 6.62. A one-way analysis of vari-ance, followed by a Student-Newman-Keuls multiplerange test, shows that the mean ejection change isgreater for the area method than for the radial orchord methods (p < 0.001). The coefficients of varia-tion are: radial, 0.26; chord, 0.26; and area, 0.12.Thus, the absolute ejection changes were greatest andthe relative variation among the regions of the ventri-cle were smallest in the area technique.The average Z for the abnormal regions was most

negative for the area method (-3.37), compared withthe radial (-2.27) and the chord (-1.95) techniques.The two-way analysis of variance rejected thehypothesis that there was no difference among the Zvalues computed for the abnormal segments by thevarious methods (p < 0.001). The Student-Newman-Keuls multiple range test showed greatest separationbetween the normal controls and the qualitativelydefined abnormal group by the area technique. Thenull hypothesis that there was no difference betweennormal and abnormal Z values by the area vs radialand area vs chord methods was rejected asp < 0.001.The multiple range test suggested that the separationbetween the normal and abnormal was probablybetter by the radial than by the chord technique(p = 0.06).

Each system's failure rate was defined as thepercentage of severely abnormal segments notdetected. These data are presented in table 2. The areamethod had the lowest failure rate (11%), comparedwith the chord (52%), and the radial (46%) techniques.The x2 test rejected the hypothesis that there was nodifference between these failure rates at p < 0.001:however, the x2 test showed no significant differencebetween the failure rate of the radial and chordmethods. The radial and chord values were thereforecombined in a contingency table and tested against thearea failure rate by x2. The failure rate for the areamethod was significantly smaller (p < 0.001). In the

anterobasilar region there was only one severely ab-normal example tested and each system detected thatabnormality. In six of six other anatomical regions,the area method had a lower failure rate than eitherthe chord or radial methods. In every case where thearea method failed to detect an abnormality, the chordand radial methods also failed.

DiscussionThe relationship between ischemic heart disease and

the potential for left ventricular regional wall abnor-malities has been firmly established since the classicexperiments of Tennant and Wiggers.7 A measure-ment of the myocardial effect of coronary artery dis-ease can be obtained by studying regional function.Other etiologies of segmental wall motion abnor-malities have been discussed, such as abnormal ven-tricular activation due to conduction defects and dis-proportionate hypertrophy.8 Wall motion analysis inthese settings may also prove clinically relevant. Un-fortunately, no quantitative method for regionalanalysis has been universally accepted.4 5

Recently, clinical interventions have been ad-vocated to determine the potential for regional wallmotion change. With these interventions, quantitationof segmental function has taken on new importance.Potentially ischemic regions of apparently normallycontracting myocardium may demonstrate functionaldeterioration with stress, such as pacing and exer-cise.", 13 Conversely, it has been reported that abnor-mal regions may improve with nitroglycerin, postex-trasystolic potentiation, inotropes and coronary arterybypass grafting.'4-l8 A quantitative system able todefine normal segmental function may assist in deter-mining clinically relevant regional change occurringwith interventions. Techniques such as two-dimen-sional echocardiography and nuclear angiography arebeing used to assess left ventricular regional function.Analyzing the reliability of these methods is mademore difficult by the lack of an accepted quantitativesystem for cineventriculogram regional evaluation.Any acceptable wall motion analysis must account

for several factors. In the 300 RAO view the analyticalsystem should compensate for lifting of the apexcaused by systolic rotation.9 Artifacts of thoracic,diaphragmatic or camera motion must also be con-sidered. Since ischemic heart disease is a major causeof regional abnormalities, an analytical system mightdo best to measure segmental ejection change of

TABLE 2. Qualitatively Abnormal Regions not Detected by Each Quantitative System

Antero- Diaphrag- Postero- Postero- Abnormal OverallAnterobasal lateral Apical matic basal lateral Septal regions percentAbn. R = 1 Abn. R = 7 Abn. R = 10 Abn. R = 8 Abn. R = 5 Abn. R = 7 Abn. R = 8 not notND %O ND % ND % ND % ND % ND % ND %0 detected detected

Radial 0 0 3 43 3 30 3 38 5 100 2 29 5 63 21 46Chord 0 0 4 57 3 30 3 38 3 60 5 71 6 75 24 52Area 0 0 1 14 1 10 1 13 1 20 0 0 1 13 5 11

Total Abn. R = 46.Abbreviations: Abn. R = number of qualitatively abnormal regions tested; ND = not detected.

996 CIRCULATION

by guest on January 11, 2016http://circ.ahajournals.org/Downloaded from

LV WALL MOTION ANALYSIS/Gelberg et al.

regions supplied by major coronary branches.Reported techniques which bisect the apical region donot consider the usual left anterior descending cor-

onary artery distribution.'2 14 Finally, any methodconstructed should examine the entire left ventricle bytwo simultaneous orthogonal views.'4We examined the rectilinear chord, radial and area

ejection methods to determine which technique bestquantitates left ventricular regional wall motion. Eachcompensated for systolic apical rotation and aorticmotion by in effect overlapping the long axis ofdiastole and systole. Diaphragm motion was excludedby comparing the diaphragm level in diastole andsystole. Although the internal reference techniqueused makes diaphragm motion less important, rota-tion about the long axis with cardiac descent may in-troduce error by changing the portion of the ventricleseen in profile. Each method examines regions of theventricle that approximate the distribution of majorcoronary branches. All techniques used biplanesimultaneous orthogonal views.

The systems differ in several important ways. Therectilinear chord method measures wall motion as a

change occurring perpendicular to the long axis of theventricle. However, shortening may occur from thebase toward the apex in all regions outlined in the 30°RAO projection.'9 The chord method detects all base-to-apex change essentially as one aortic valve plane-to-apex measurement. All area measurements andsome radial line measurements take base-to-apexchange into account. The area method includes a longperimeter of the ventricular silhouette for each seg-ment, while the chord and radial methods depend on

change of one ventricular perimeter point toward one

reference point on the long axis. The assumption is

made in the chord and radial methods that the pointon the perimeter located in diastole is approximatelythe same point measured in systole. This may often bean invalid assumption and sometimes is clearly not thecase. For example, a radial line in the high septum indiastole may be seen in systole to intersect the aorticvalve plane. Since a large perimeter, rather than a

single silhouette point, determines change in the area

method, this technique is more likely to haveanatomically coinciding systolic and diastolic regions.

Choosing a standard for comparing the threemethods posed a problem because only subjectiveassessment of wall motion was available. The problemof observer variability with subjective analysis is wellrecognized.4 Therefore, we attempted to minimize thisvariability by analyzing only the regions with markeddyssynergy. The subjective selection process appearsreasonable, because the areas chosen were always inthe distribution of a coronary artery with at least 90%diameter narrowing, often correlated anatomicallywith a Q-wave infarct pattern on the ECG (14 of 17patients), and usually were associated with a history ofclinical infarction (14 of 17 patients).Our criteria for selecting one of the three methods

for clinical assessment of regional wall motion are: 1)the method should best illustrate the normally uni-

form symmetric nature of ventricular function on ex-amination of a normal control population, and 2) themethod should best detect the selected, markedly ab-normal regions. The area method meets these criteria,and demonstrated the best uniformity of the meanregional changes with a coefficient of variation 0.12 vs0.26 for the other methods. Only the posterobasilararea 5 was low, probably because much of this regionconsists of noncontractile posterior mitral valve andaortic valve structures. The data in table 2 substan-tiate that the area method detected abnormality best,independent of which ventricular segment was in-volved. It also had a significantly lower overall failurerate than the other methods.The ventriculogram shown in figure 3 was subjec-

tively judged to have severe wall motion abnormalitiesof four regions in the RAO view. These regions andthe Z values for each method are listed in the accom-panying chart. Three of four abnormal regions werenot detected (i.e., Z > -2) by the chord and radialmethods. All four abnormal segments in the areamethod had Z < -2.

Figure 4 shows Z values computed for each quali-tatively, severely abnormal region examined. The areamethod separates the normal and abnormal groupssignificantly better than the other two methods. Thedashed line in this diagram represents the lower limitof normal defined at the 97% confidence level by theone-tail distribution. For our population thisrepresents 2 standard deviations from the mean.Ninety-seven percent of the normal population wouldfall above this dotted line. The overlap between abnor-mal and normal is apparent for the radial and chordmethods, as demonstrated by the distribution of the Zvalues for the qualitatively abnormal regions. Three offive abnormal regions not detected by the area methodwere near the lower limits of normal and exceed the95% confidence limits. The mean changes and thelower limits of normal for each region by each systemare illustrated in figure 5. It is important to note howlarge the normal limits are in many regions of theradial and chord techniques. The lower limits of nor-mal in some regions fall near or below the zero changeline. Lower normal limits are well above the zero linein the area method. Area 5, the region of smallest per-cent regional change, encompasses mitral and aorticvalve structures in the RAO view. It might thereforebe expected to have less change than purely myocar-dial regions of the ventricular silhouette.

In each case where the area method failed to detectabnormality, the chord and radial methods also failed.Five qualitatively abnormal segments were notdetected by any method. Four of five of these regionswere supplied by critically occluded coronary arteriesin patients with infarction by history and currentECG. One region of the five was supplied by a com-pletely occluded coronary artery in a patient with ahistory of infarct, but the ECG was essentially nor-mal. There were three apparent reasons whyqualitatively abnormal segments were not detected bythe methods studied: 1) there were two instances of

997

by guest on January 11, 2016http://circ.ahajournals.org/Downloaded from

VOL 59, No 5, MAY 1979

VALUES OF ZMETHODS

REGIONSRADIAL CHORD AREA

ANTEROLATERAL - 2.59 -2.05 -2.64APICAL -1.13 -1.13 -2.25DIAPHRAGMATIC -1.15 -0.95 -2.25POSTEROBASAL -0.08 -0.92 -2.36

RADIAL

ANTEROBASAL

ANTEROLATERALX

V -

I

V APICAL

POSTEROBASADDIAPERAGMATIC

ANTEROBASAL

..ANTEBROLATERAL

V ."AZ

I/

II'II .0 APICAL

POSTEROBASA DIAPHRAGMATIC

AREA

AN TEROBASAL

ANTEROLATERAL

II r

I I I-I / 1

B / APICAL

POSTEROSASADIAPHRAGMA TIC

FIGURE 3. An example of wall motion analysis by the three techniques. This RAO silhouette was con-sidered by subjective analysis to have four regions of severe abnormality. The solid line indicates end-diastole, the dashed line end-systole. The accompanying chart lists the Z valuesfor each subjectively defined,severely abnormal region by the three methods. A Z value < -2 meets the criteria for detection ofabnor-mality.

QUANTITATIVE EVALUATION OF 46QUALITATIVELY ABNORMAL REGIONS

I FIGURE 4. The distribution ofZ values ofqualitatively defined severe abnormalities ineach technique. Dashed line represents lowerlimits of normal at the 97% confidence levelby the one-tail t distribution. This is 2 SDbelow the normal mean in our population(i.e., Z = -2). If the 95% confidence limit isused, all but two of the qualitatively abnor-mal areas are detected by the area method.

RADIAL CHORD AREA

I0

-1

-2

-3-4

-5

-6

-7

N

(4czLU

-mm

0zLUI:

0-mLU

LU

C)

z

-8

-9

CIRCULATION998

by guest on January 11, 2016http://circ.ahajournals.org/Downloaded from

LV WALL MOTION ANALYSIS/Gelberg et al.9

REGIONAL EJECTION CHANGES* Meanv Lower limits of normal

RADIAL

1 2 3 4 5 6 7 8 91011121314

CHORD AREA

a

I

FIGURE 5. Normal mean ejection changesand lower limits ofnormalfor each region ofthe ventricle in the three methods examined.Numerals for ventricular regions in thisfigure coincide with those offigure 2.

1 2 34 56 78 12345678

R E G I O N S

small discrete wall motion abnormalities where theremainder of the region appeared qualitatively nor-mal; 2) in one case the region of abnormality notdetected was small and contiguous with an abnormalsegment detected by all methods, and the abnormalportion of the region not detected was small in relationto the normal-appearing remainder of the segment;and 3) in retrospect, we may have qualitativelyoverestimated the degree of abnormality in twosegments. Moreover, if the confidence limits for nor-mality were reduced to 95%, all but these two of the 46qualitatively abnormal regions would have beendetected by the area method (see fig. 4). Any quan-titative wall motion analysis is likely to miss somesubtle subregional abnormalities, and since thistechnique only measures shortening at end-systole, ab-normalities during early ejection (tardokinesis) maynot be detected. Even so, objective methods arenecessary to assess the effects of therapy and permitcomparison of nuclear, echo and catheterizationstudies.Our goal was to evaluate methods that could be

easily automated with relatively inexpensive equip-ment. Clearly, a large number of measured radial orchord lines may adequately characterize regional wallmotion,20' 21 and averaging techniques for multipleradial and chord lines may add to the sensitivity of

these methods. However, complex methods are

difficult to implement for general clinical use. Thearea method appears to be the best of the simple quan-

titative wall motion techniques we analyzed.

References1. Baxley WA, Reeves TJ: Abnormal regional myocardial perfor-

mance in coronary artery disease. Progr Cardiovasc Dis 13:405, 1971

2. Gorlin R, Klein MD, Sullivan JM: Prospective correlativestudy of ventricular aneurysm. Mechanistic concept and clinicalrecognition. Am J Med 42: 512, 1967

3. Helfant RH, Kemp HG, Gorlin R: Coronary atherosclerosis,coronary collaterals and their relation to cardiac function. AnnIntern Med 73: 189, 1970

4. Chaitman BR, DeMots H, Bristow JD, R6sch J, RahimtoolaSH: Objective and subjective analysis of left ventricularangiograms. Circulation 52: 420, 1975

5. Brower RW, Meester GT: Computer based methods for quan-tifying regional left ventricular wall motion from cineven-triculograms. In Computers in Cardiology, IEEE ComputerSociety, Long Beach, California, 1976, pp 55-62

6. Austen WG, Edwards JE, Frye RL, Gensini GG, Gott VL,Griffith LSC, McGoon DC, Murphy ML, Roe BB: AHA Com-mittee Report: a reporting system on patients evaluated for cor-

onary artery disease. Circulation 51: 5, 19757. Tennant R, Wiggers CJ: The effect of coronary occlusion on

myocardial contraction. Am J Physiol 112: 351, 19358. Harrison TR: Some unanswered questions concerning enlarge-

ment and failure of the heart. Am Heart J 69: 100, 1965

60f

-m'

z0LU

z

LULU

50

40

30

20

10

0

-10

-20

-30'

999

_ Ab

by guest on January 11, 2016http://circ.ahajournals.org/Downloaded from

1000 CIRCULATION

9. Leighton RF, Wilt SM, Leurs RP: Detection of hypokinesis byquantitative analysis of left ventricular cineangiograms. Cir-culation 50: 121, 1974

10. Herman MV, Heinle RA, Klein MD, Gorlin R: Localized dis-orders in myocardial contraction: asynergy and its role in con-gestive heart failure. N Engl J Med 277: 222, 1967

11. Baltaxe HA, Alonso DR, Lee JG, Prat J, Husted JW, StakesJR: Impaired left ventricular contractility in ischemic heart dis-ease: angiographic and histopathologic correlations. DiagRadiol 113: 581, 1974

12. Pasternac A, Gorlin R, Sonnenblick EH, Haft JI, Kemp HG:Abnormalities of ventricular motion induced by atrial pacing incoronary artery disease. Circulation 45: 1195, 1972

13. Borer JS, Bacharach SL, Green MV, Kent KM, Epstein SE,Johnston GS: Real-time radionuclide cineangiography in thenoninvasive evaluation of global and regional left ventricularfunction at rest and during exercise in patients with coronaryartery disease. N Engl J Med 296: 839, 1977

14. Klausner SC, Ratshin RA, Tyberg JV, Lappin HA, ChatterjeeK, Parmley WW: The similarity of changes in segmental con-traction patterns induced by postextrasystolic potentiation andnitroglycerin. Circulation 54: 615, 1976

15. Helfant RH, Pine R, Meister SG, Feldman MS, Trout RG,Banka VS: Nitroglycerin to unmask reversible asynergy.

VOL 59, No 5, MAY 1979

Correlation with postcoronary bypass ventriculography. Cir-culation 50: 108, 1974

16. Horn HR, Teicholz LE, Cohn PF, Herman MV, Gorlin R:Augmentation of left ventricular contraction pattern in cor-onary artery disease by inotropic catecholamine. Theepinephrine ventriculogram. Circulation 49: 1063, 1974

17. Chatterjee K, Swan HJC, Parmley WW, Sustaita H, MarcusHS, Matloff J: Influence of direct myocardial revascularizationon left ventricular asynergy and function in patients with cor-onary heart disease. Circulation 47: 276, 1973

18. Dyke SH, Cohn PF, Gorlin R, Sonnenblick EH: Detection ofresidual myocardial function in coronary artery disease usingpost-extrasystolic potentiation. Circulation 50: 694, 1974

19. McDonald IG: The shape and movement of the human left ven-tricle during systole. A study by cineangiography and bycineradiology of epicardial markers. Am J Cardiol 26: 221,1970

20. Clayton PD, Bulawa WF, Urie PM, Marshall HW, KlausnerSC, Vincent GM, Warner HR: The characteristic pattern forthe onset of contraction in the normal human left ventricle.(abstr) Circulation 56 (suppl III): 111-126, 1977

21. Hegg TD, Stewart DK, Dodge HT, Frimer M, Tosch TJ:Quantitative left ventricular wall motion analysis. (abstr) Cir-culation 56 (suppl III): III-238, 1977

by guest on January 11, 2016http://circ.ahajournals.org/Downloaded from

H J Gelberg, B H Brundage, S Glantz and W W Parmleymethods.

Quantitative left ventricular wall motion analysis: a comparison of area, chord and radial

Print ISSN: 0009-7322. Online ISSN: 1524-4539 Copyright © 1979 American Heart Association, Inc. All rights reserved.

is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Circulation doi: 10.1161/01.CIR.59.5.991

1979;59:991-1000Circulation. 

http://circ.ahajournals.org/content/59/5/991the World Wide Web at:

The online version of this article, along with updated information and services, is located on

  http://circ.ahajournals.org//subscriptions/

is online at: Circulation Information about subscribing to Subscriptions: 

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

  document. Permissions and Rights Question and Answer information about this process is available in the

located, click Request Permissions in the middle column of the Web page under Services. FurtherEditorial Office. Once the online version of the published article for which permission is being requested is

can be obtained via RightsLink, a service of the Copyright Clearance Center, not theCirculationpublished in Requests for permissions to reproduce figures, tables, or portions of articles originallyPermissions:

by guest on January 11, 2016http://circ.ahajournals.org/Downloaded from