In vivo assessment of a digital angiographic method to measure absolute coronary artery diameters

CORONARY ARYERY DISEASE

American Journal of Cardiology1

JULY 15, 1989, VOL. 64, NO.3

In Vivo Assessment of a Digital Angiographic Method to Measure Absolute Coronary Artery Diameters Jeffrey J. Popma, MD, Eric J. Eichhorn, MD, and Gregory J. Dehmer, MD

Several techniques exist for the quantiition of absohde coronary artery diameters ustng radiotogic methods. An in vtvo assessment of a quantitative technique based on direct digitally acquired images was performed by imaging inflated angioplasty balbans (II = 26), battoon catheter shafts (n = 16) and coronary guidewires (n = 20) at the time of coronary angtoptasty. After this, the actual size of the objects was determined with a micrometer. Diameters measured by the quantitative digital method had an exceknt correlation with the actual diameters (digital diameter = 0.60 [actual diameter] + 0.32; n = 61; r = 0.97; standarderror of the estimate = 0.26 mm; p 1 mm were usuatty underestimated by the digital technique, although the actual magnitude of the error was smatt. Objects l mm were underesttmated by 0.23 f 0.19 mm. Based on an analysis of the error present, correction atgorithms were formutated and tested prospectively u&g an additional 29 object measurements. This resulted in an improvement in the quantification of the diameters with a smaller magnitude of error. This in vivo assessment suggests that the rapid on-

line assessment of absolute coronary artery diameters is possible, but also demonstrates important errors inherent in this method. (Am J Cardiol196S;64:131-136)

usualinterpretation is the technique usedmost frequently to evaluate the severity of stenosesseen by coronary arteriography. Although visual interpretation is convenient and rapid, it is limited by substantial intra- and interobserver variabilityie4 and the inability to quantify absolute coronary artery size and stenosis severity.5-9 Because coronary atherosclerosis can be diffuse, visual estimatesof the percent diameter stenosismay underestimate the true extent of luminal narrowing.lO-l2To provide a more preciseassessmentof coronary size and reduce observervariability, computerbasedtechniqueshave been developedto quantify absolute arterial diameter and stenosiscross-sectionalarea.s*~~-~sBoth cineangiographic film5*6,8J5-18 and recently, direct, digitally acquired images7J9-*lcan be used to derive such measurements. These quantitative techniques have been validated using radiographic phantoms,8J5J6surgically implanted plastic arterial stenoses in animals14and arterial segmentsfrom human cadavers.8.21In some of these validation studies, attempts have been made to model the imaging conditions found clinically;8Js~21 however, exact duplication of these conditions is difficult. During coronary angioplasty procedures, objectsof known size are advancedinto the coronary arteries and can be imaged. This provides a practical opportunity to assessquantitative methods under imaging conditions identical to those used during coronary angiography, eventhough the coronary arteries are From the Cardiac Catheterization Laboratory, the Dallas Veterans Administration Medical Center, and the Department of Internal Medi- not actually measured. Therefore, the purpose of this cine (Cardiology Division), the University of Texas SouthwesternMed- study was to perform an in vivo evaluation of the accuical Center, Dallas, Texas. This work was supportedin part by a grant racy and limitations of a quantitative analysis method from the Research Service of the Veterans Administration, Washington, DC. Manuscript received December 19, 1988;revised manuscript developed specifically for digitally acquired coronary arteriograms. receivedand acceptedApril 28, 1989. Dr. Dehmer’s presentaddress:C.V. RichardsonCardiac Catheterization Laboratory, North Carolina Memorial Hospital, Manning Drive, SecondFloor, Chapel Hill, North Carolina 27514. Address for reprints: Eric J. Eichhorn, MD, Cardiac Catheterization Laboratory (11lA2), Veterans Administration Medical Center, 4500 South Lancaster Road, Dallas, Texas 75216.

V

METHODS Patients: The study material consisted of digitally acquired images of inflated angioplasty balloons, balloon catheter shafts and coronary guidewires obtained THE AMERICAN JOURNAL OF CARDIOLOGY JULY 15, 1989

131

ASSESSMENT OF QUANTITATIVE ARTERIOGRAPHY

in 30 male patients undergoing coronary angioplasty. Their mean body weight was 82.3 f 14.0 kg (range 57.7 to 118.2) and mean body surface area was 1.98 f 0.21 m* (range 1.60 to 2.30). Digital images were acquired during angioplasty of the right coronary artery in 18 patients, left anterior descending artery in 12 patients and circumflex artery in 6 patients. In 5 patients, more than 1 balloon size was imaged during the angioplasty. A total of 90 objects were imaged: 34 inflated balloons, 21 balloon catheter shafts and 35 coronary guidewires. Balloon shafts could not be measuredin 16 patients becausethe lesion dilated was in the proximal artery and the catheter shaft remained within the guiding catheter. For this study, the object images were divided into 2 groups. Images of the first 61 objects (25 inflated balloons, 16 balloon catheter shafts and 20 coronary guidewires) were analyzed to define the accuracy of the technique and correlation between inter- and intraobserver measurements.Based on an analysis of the errors detected, correction algorithms were derived and tested prospectively using the images of the next 29 objects (9 inflated balloons, 5 balloon catheter shafts and 15 coronary guidewires). Image acquisition: Single plane images were acquired with a standard cineangiographic system (Philips Optimus M-200 generator and Poly-C) interfaced to a digital radiographic computer system (ADAC DPS4100C). A constant radiographic input signal was used (fixed kVp, mA and pulse width) and images were recorded with a 5-inch image intensifier field size. Digital

FIGURE angbpmybanooninnaedk;the;ipMeo;;onyartery rfferlgayrule--bhdci-lw ---I ---I-a -_-. -AL---d .- bv0d-a --_-II------.

132

THE AMERICAN JOURNAL OF CARDIOLOGY VOLUME 64

images were acquired at 30 frames/s into a 512 X 512 X 8 bit pixel matrix and were recorded so that the angioplasty balloon was centered in the field of view. The maximum effect of pincushion distortion in 5-inch mode on this system was <5%. The resolution of the digital and tine systemsused in this study was tested by imaging a line pair phantom at 40 kVp without grid removal. In the 5-inch field size, resolution of 2.5 line pairs/mm was obtained for digital images; thus, the ultimate resolving capacity of the digital imageswould be up to 0.4 mm. In contrast, the resolution achieved on cineangiographic film was 3.8 line pairs/mm or a resolving capacity of 0.26 mm. Image processing

and radiographi

diameter

mea-

surement: Images were analyzed without mask subtraction using an automated quantification program described in detail elsewhere.7J4Briefly, the best single frame that demonstrated the guiding catheter, inflated balloon, balloon catheter shaft and coronary guidewire was selected, Gray scale inversion was performed to produce a white-on-black image and window widths were adjusted to maximize the contrast betweenthe object and background (Figure 1). Gray-scale modification was performed to linearly expand the individual dynamic range to fill the full 8-bit range of the digital system.The resulting image was stored for further analysis. Calibration of the selectedframe was performed using the guiding catheter shaft as the reference.A circular region of interest was positioned over the catheter shaft and the centerline of the catheter segment was derived automatically. Linear density profiles perpendicular to the catheter centerline were calculated over the selectedsegmentof the catheter. A weighted average of the first and secondderivative of the perpendicular density profiles was used to determine the edge points. These initial gradient-determined edge points were examined for spatial continuity and outliers were discarded. Each perpendicular profile was then reanalyzed and the location of the final edge points derived. The diameter was computed as the distance along each perpendicular profile between edge points on opposite sidesof the catheter. Calibration was achievedby entering the known catheter diameter obtained by micrometer; the resulting calibration factor (mm/pixel) was displayed automatically by the computer. In a similar fashion, a circular region of interest was placed over the inflated angioplasty balloon, balloon catheter shaft or coronary guidewire. Edge detection was performed automatically as just described and the mean diameter of the object obtained by averaging 8 to 12 perpendicular diameters over the length of the object (Figure 2). Angioplasty balloon measurementswere obtained from the image recordedduring the highest inflation pressure used. All object measurementswere performed by observers blinded to the actual diameter measurements obtained by micrometer. The distance between the calibration object (i.e., guiding catheter) and the object measuredwas approxi-

mated by a direct measurementfrom the video display monitor. The number of pixels between these 2 objects was determined and multiplied by the correction factor (mm/pixel) previously determined. Becauseonly single plane images were acquired, the distance computed in this fashion would equal the actual distance only if both objects were in the same horizontal plane with respect to the image intensifier. If the 2 objects were in different planes, the measureddistance would underestimate the actual distance. Detenninatii of absolute diameters: After completion of the angioplasty, actual diameters of the various objects were measured with a micrometer accurate to 0.01 mm. Balloons were inflated to the samepressureas used for the imaged inflation and all objects were measured at body temperature. Data analysis: For the first group of 61 objects, all diameters were determined twice by 1 observer and independently by a secondobserver. For each determination, a unique image was created and calibrated. Two expressionsof error were used to evaluate the digital measurements.First, the actual error was defined as the difference (in mm) between the diameter measuredby the digital technique and the actual diameter deter-

mined by micrometer. In this format, a positive number for the actual error representsan overestimation of the actual diameter by the digital analysis and a negative number the converse.Second, to determine the magnitude of error irrespective of the direction (i.e., over- or underestimation of the actual diameter), the absolute value of the actual error was determined; this was called the magnitude of the actual error. Statistical analysis: All correlations were evaluated using standard linear, nonlinear and multivariate regressionanalyses.All data were tested for homogeneity by Bartlett’s test and nonparametric methods used when necessary.Differencesbetweengroups were determined by a Student t test for all parametric data or a Mann-Whitney analysis for nonparametric data. A p value <0.05 was consideredsignificant. RESULTS Correlation

of measured and actual diameters: In the first 61 objects measured, there was an excellent correlation betweenthe diameter measuredby the digital technique and the actual diameter (digital diameter = 0.8 actual diameter + 0.32; r = 0.97; standard error of the estimate = 0.26 mm, p
PIGURE 2. Precwmdand andyzedlmageefthelnllatedaan&pbQMbencauwttWShOWlBinPll1.A ETapNcQplarbrrperlmpeadewrthahnageand 8hewsareconrbuclSonoftlw regment--w PI-O(PI and (W C?lIdSildKkd.Arhgkbanetewvrlueof~mmisindicatedbythevert&alrme aavssthe-ilnage.lRadbnetervalueusad fortha8tUdyW~theever8geefmdtipkvahwssampkd-*bmMof the--mm lower

graph

shews

a pIad

ef

uledlemebm(~axis) MlhskngthoftJ==s--(horiwntel axk).

THE AMERICAN JOURNAL OF CARDIOLOGY JULY 15, 1989

133

ABBE!SSMENT OF QUANTITATIVE ABTEBlOGRAPHY

TABLE I Analysis of Error Obtained by Digital Angiography

Factor

Images (n)

image size Otolmm >l mm

21 40

Distance (calibration catheter to image) 33 Oto3Omm 28 >30 mm

Actual Error (mm) 0.41 fO.ll -0.23 f 0.20 p < 0.01 -0.14 f 0.27 0.15f0.36 p < 0.05

Coronary artery Riiht LAD LC

34 17 10

-0.01 f 0.36 -0.01 f 0.16 0.04 f 0.34 p = 0.96

Projection RAO LAO LAT

28 23 10

0.04 f 0.34 0.03 f 0.34 -0.12 f 0.38 p=o.48

LAD = laft anterior descending IA0 = left anterior oblique; MT = lateral projection: LC - left drcumfhzx artery: RAO - r@t anterior oblique.

TABLE II Correlation Interobserver Object

l

Achml Measured Actunl Measured Actual Measured Coronay Bollwn Cothekr Balloon Guidrwlrr Shaft 15Dots

ofactud-meamndby !ziz!zzF dlmKbrsmeaswedbytbeaeiwraliiobdwns,baMoasscatllSVPNC-.-m

eterstldband caronry--~-Tbememandstadarddavktkn aresbowntoBwddeoftbe hdihhd points. Asterfsk mpmsmts ackwterd15poi~ because~dtlle CO?OMrY~WWetlWB dimwter, 0.014 indlesh

134

THE AMERICAN JOURNAL OF CARDIOLOGY VOLUME 64

No.

lntraobserver correlation All images 61 Balloon 25 Shaft 16 Wire 20 lntraobsetver correlation All images 61 Balloon 25 Shaft 16 Wke 20 SEE - standard

r value

SEE (mm)

Slope

P value

0.99 0.98 0.93 0.81

0.09 0.11 0.06 0.07

0.99 0.96 0.81 0.56


0.99 0.87 0.80 0.81

0.16 0.24 0.09 0.06

0.94 0.97 0.86 0.65

<0.001
error of the estimate.

rate comparisonsof the actual and measureddiameters for the 3 different objects measureddemonstratedthat the digital measurementsconsistently overestimatedthe actual diameters of coronary guidewires and tended to underestimatethe diameters of the inflated balloons and balloon catheter shafts (Figure 3). Analysis

p
Between lntraobserver and Measurements by Digital Analysis

actual errw

and magnitude

of actual er-

ror: Becausethe diameters of someobjects were underestimated and others overestimated,the mean actual error for the first group of 61 objects approached0 (0.01 f 0.35 mm). However, the magnitude of the actual error also was small (0.31 f 0.16 mm). For objects <0.5 mm in diameter, the mean actual error was 0.41 f 0.11 mm, indicating that the digital measurementsoverestimated the actual diameter. Conversely,the mean actual error was -0.23 f 0.19 mm for objects >l mm, indicating an underestimation of the actual diameter by the digital technique. In addition to evaluating the relation betweenthe actual error and the actual diameter of the object, we assessedthe relation between actual error and 3 other variables (Table I). The mean actual error was greater for objects >30 mm from the calibration point. Mean actual error for objects 0 to 30 mm from the calibration object was -0.14 f 0.27 mm whereas mean actual error for objects >30 mm from the calibration object was 0.15 f 0.36 mm (p <0.05). Using a multivariate linear regression analysis both the object size and distance were independently correlated with the actual error (r = 0.72, p
balloon catheter shafts and coronary guidewires and this relation to correct the error was tested prospectively was excellent in each group (Table II). in a second group of 29 objects. Error determined by Caneded dlgltal measuremed s ef diameter: The this equation was applied to the measureddiameters to relation between the actual error and the actual diame- derive a corrected diameter assessment.An excellent ter was not linear. This relation was best fitted by the correlation existed between the corrected measureddithird order polynominal equation, actual error = 0.90 - ameter and the actual diameter (corrected measureddi1.60 (AD) + 0.69 (AD)* - 0.09 (AD)3, where AD = ameter = 0.86 [AD] + 0.21 [r = 0.98, standard error actual diameter (r = 0.87, standard error of the esti- of the estimate = 0.21 mm, p l mm seemedto be systematically underestimated. Although this relation could be characterized by a first order regression equation [actual error = 0.017 (AD) - 0.261,the slope approached0. To simplify this correction, the mean actual error of these objects (-0.23 f 0.19) was applied to the measureddiameters to derive a corrected diameter. The mean actual error OBSERVATION A for the uncorrected measurements>0.5 mm was -0.39 (mm) f 0.15 mm and decreasedto -0.18 f 0.15 mm (p
CORRELATION

t Pi.61 AE=O.~-l~OlADl+O.69~AD~-DO9,bDl’ R* 0.07

l.Oci 9

0.5 <

5 aq

.

T=&

0:

i 0z Q

.o

-0.5-

’

.

.

..

-1 .o

t

L.

..I

. 1 .o ACTUAL

1.0

B

2.0

oBsFnK~R A

3.0

.

. 2.0

.

.

. .

DlfmMm:TER

. 3.0

...’ 4.0

(AD)

4.0 FlCURES.Relathbetwesntheactualt&neterasmeawed bymkrsmehrandtheactualetvorhmm.llwactualbythedlgitalraarleWS3SddilldUthO-graphknlethedmlnustheastual~-bY -.lllebeStfftW~achkwdWRhthOthirdordar

[email protected]<0Smm -withcawldsr~ranabnl WC-vdatbnwhlkobjscts>lmmwensydmatkdyundsr-.

THE AMERICAN JOURNAL OF CARDIOLOGY JULY 15. 1989

135

ASEMMENT

GF QUANTITATIVE ARTERIGGRAPRY

sessmentsof lesion severity provide indirect evidencefor the validity of quantitative measurements.Direct validation studies of quantitative coronary arteriographic methods also have been performed using several different approaches.Arterial models or phantoms have been constructed of various materials such as brass or plexiglassand imaged in a manner that attempts to duplicate the conditions found clinically.8J3J5 Although validation studies using phantoms have uniformly demonstrated a high degreeof accuracy for the determination of the model vesseldiameters, somefactors such as vessel motion and the complex attenuation of the chest wall cannot be duplicated. Other validation studieshave been performed by harvesting coronary arteries from cadavers,distending them with contrast agentsat physiologic pressures and imaging them under conditions that attempt to recreate those present clinically. Such measurementsare then correlated with the actual diameters obtained by direct grossor microscopicanalysis of the artery lumen.8*21Despite careful preparation of these specimens,some distortion of the artery may occur and, similar to the other validation studies,the actual imaging conditions vary from those occurring cliniprevious validations of quantitative arteriography: The agreement between anatomic and physiologic as- cally. One in vivo validation study has been performed in a canine model.14Precision-drilled plastic cylinders were surgically placed into arteries, creating intralumin0.6 al stenosesof known diameters. When the animal fully t recovered,quantitative arteriographic imaging of these 0.5 arteries demonstrated an excellent correlation between 0 . t the measureddiameters and both the actual diameters and physiologic indexes of coronary flow reserve. In vivo assessment of quantitative angiography: We z 0.3 recorded images of angioplasty hardware and used quantitative arteriographic methodsto determine the di+5 0.2 ameters of these objects. There was a significant correI 8 01 lation between the measured and actual diameters E . t (standard error of the estimate = 0.26 mm) and an excellent agreement between intra- and interobserver measurements.However, there was a consistent error in -O.lthe measurementsthat was related to the size of the z object imaged. Object diameters <0.5 mm were always 2 -0.2 overestimatedand object diameters > 1 mm were nearly always underestimated by the quantitative angiographic -0.3 system. The actual error was 0.41 f 0.11 mm for objects l mm. There appeared to be 2 different types of error -0.5 present. In addition to being systematically overesti-0.4I mated, measurementsof objects l mm were subFlGURE&Acludewwsforthe lawmmctsd-moaject to somerandom variation, but they also appearedto systematically underestimate the actual diameters. Prospectively, 2 different methods were used to correct for --bYthe these observed errors. One included objects
2

136

THE AMERICAN JOURNAL OF CARDIOLOGY VOLUME 64

Because of the different mathematical relations used, both resulted in a similar reduction in the actual error. Advantages of digital acquisition: The quantitative angiographic method used in this study was developed by LeFree et al7 and validated using both arterial phantoms and an in vivo canine preparation.14Rather than deriving these measurements from cineangiographic film, this method uses direct digitally acquired images. This on-line digital method has several advantages. First, it allows rapid processingand immediate quantitative estimation of diameters becausedeveloping cineangiographic film is not necessary;the approximate processingtime for determining the diameter of an object was only 4 to 5 minutes in our study. Second,accurate diameter determination is dependenton preciselocalization of the vesseledge.Vesseledgesare often not sharp ly demarcated and thus must be defined by the varying density of the edge profile.5 Our quantitative technique used an automated edge detection program that more closely approximates the true arterial wall edge than visual estimation alone and was optimized for the noise frequency of the digital images.7J4 Although other automated edge detection methods exist, this method has beendemonstratedto most closely approximate true coronary artery diameters in human cadaver coronary arteries.*’ Finally, the correlation between intra- and interobserver observations was excellent and probably related to the fully automated nature of the analysis. Limitations of digital acquisition: Although the digital technique we used yielded excellent correlations with the actual diameters, it has several limitations. First, it provides no correction for pincushion distortion. The amount of error introduced at the periphery of the image field as a result of pincushion distortion has been estimated to be 5 to 8%8 and was <5% in our study. Becausethe images were oriented so that the inflated angioplasty balloon was in the center of the field of view, the guiding catheter, used as the calibration object, was always closer to the periphery. The lack of correction for pincushion distortion would result in a small error in the determination of the calibration factor and thereby causethe diameters of objects in the center of the field of view to be underestimated slightly. This probably contributed to a portion of the underestimation of the inflated angioplasty balloons and balloon catheter shafts. Second,the digital method used did not correct for out-of-plane magnification. Out-of-plane magnification is the differential magnification that occurs when objects lie in different parallel planes with respect to the image intensifier and has been estimated at approximately 1.5% for each centimeter that separates the objects.6The calibration factor determined for each image is strictly applicable only to objects located in the same plane as the catheter, parallel to the image intensifier input screen. Correction for out-of-plane magnification requires acquisition of 2 nearly orthogonal views and 3-dimensional reconstruction algorithms.6,s,22-24 Unfortunately, it is not always possibleto obtain orthogonal images of all lesions becauseof vessel overlap and other technical factors. Finally, there are

concernsregarding the resolution possiblewith the 5 12 X 512 pixel matrix used for the digital images.Lack of resolution due to the combined effects of radiation scatter and veiling glare plus the lack of correction for pincushion distortion and out-of-plane magnification could all explain the inability to accurately determine diameters of objects <0.5 mm. Overestimation of diameters
137

ASSESSMENT

OF QUANYITATIVE ARTERIOGRAPHY

correction algorithms, the overall accuracy of these measurementsis still limited by the absolute resolving power of the imaging system. The results of this study help define the limits of accuracy of such digital measurements and suggest that these measurementsmay have their greatest applicability in the acute assessment of relative changesin vesseldiameter during the course of a single angiographic study.

RA. Automated quantitative coronary angiography:morphologicand physiologic validation in viva of a rapid digital angiographic method. Circulation 1987; 7P452-460.

18. SpearsJR, SandorT, Als AV, Malagold M, Markis JE, GrossmanW, Serur JR, Paulin S. Computerized image analysis for quantitative measurementof vesseldiameter from cineangiograms.Circulation 1983,68:453-461. 16. Nichols AB, Gabrieli CFO, Fenoglio JJ, E&r PD. Quantification of relative coronary arterial stenosisby cinevideodensitometricanalysisof comnary arteriograms. Circulation 1984,69:512-522. 17. Vas R, Egler N, Miyazono C, Pfaff JM, ResserKJ, WeissM, Nivatpumin T, Whiting J, Forrester J. Digital quantification eliminates intraobserver and interobservervariability in the evaluation of coronary artery stenosis.Am J Cardiol Adcnowledgment: We expressour gratitude to Don- 1985;56:718-723. ald Haagen, RCPT, Janie McCoulsky, CCRN, Willie 18. Serruys PW, Reiber JHC, Wijns W, van den Brand M, Kooijman CJ, ten HJ, Hugenholtz PG. Assessmentof percutaneoustransluminal coronary Gary and Gracilia Fields for their expert help during Katen angioplastyby quantitative coronary angiography:diameter versusdensitometric the performance of these studies. area measurements.Am J Cardiol 1984;54:482-488. 19. Tobii J, Nalcioglu 0, Iseri L, JohnstonWD, Rceck W, CastlemanE, Batter B, Montdli S, Henry WL. Detection and quantitation of coronary artery stenoses from digital subtraction angiogramscomparedwith 35-millimeter film cineangio grams. Am J Cardiol 1984;54:489-496. REFERENCES 20. Sam ML, Man&i GBJ, LeFree MT, Mickelson JK, Starling MR, Vogel 1. Detre KM, Wright E, Murphy ML, T&am T. Observeragreementin evaluat- RA. Variability of quantitative digital subtraction coronary angiography before ing coronary angiograms.Circulation 1975;52:979-986. and after percutaneous transluminal cornnary angioplasty. Am J Cardiol 2. DeRouen TA, Murray JA, Owen W. Variability in the analysis of coronary 1987,60:55-60. 21. RosenbergMC, Klein LW, Agarwal JB, StetsG, Hermann GA, Helfant RH. arteriograms. Circulation 1977;55:324-328. 3. Fisher LD, Judkins MP, LesperanceJ, Cameron A, Swaye P, Ryan T, MayQuantification of absolute luminal diameter by computer-analyzeddigital sub nard C, BourassaM, Kennedy JW, GosselinA, Kemp H, Faxon D, Wexler L, traction angiography: an assessmentin human coronary arteries. Circulation Davis KB. Reproducibility of coronary arteriographic reading in the Coronary 1988:77:484-490. Artery Surgery Study (CASS). Cathet Cardiovasc Diagn 1982;8:565-575. 22. Kirkeeide RL, Gould KL, Parse1L. Assessmentof coronary stenosesby 4. Zir LM, Miller SW, Dinsmore RE, Gilbert JP, Harthome JW. Interobserver myocardial perfusion imaging during pharmacologiccoronary vascdilation. VII. Validation of coronary flow reserveas a single integrated functional measureof variability in coronary angiography. Circulation 1976;53:627-632. 5. Brown BG, Bolson EL, Dodge HT. Quantitative computer techniques for stenosisseverity reflecting all its geometric dimensions.JACC 1986;7:103-113. analyzing coronary arteriograms. Prog Cardiovasc Dis 1986:28:403-418. 23. Harrison DG, White CW, Hiratzka LF, Doty DB, BarnesDH, EastmanCL, 6. Reiber JHC, Kooijman CJ, Slager CJ, Gerbrands JJ, SchuurbiersJCH, Den Marcus ML. The value of lesion cross-sectionalarea determinedby quantitative Boer AD, Wijns W, Serruys PW. Computer assistedanalysis of the severity of coronary angiography in assessingthe physiologic significance of proximal left obstructions from coronary cineangiograms:a methodological review. Autoanterior descendingcoronary arterial stencxs. Circulation 1984x59:1 11I-l 119. medico 1984;5:219-231. 24. Harrison DG, FergusonDW, Collins SM, Skorton DJ, Ericksen EE, Kits7. LeFree MT, Simon SB, Mancini GBJ, Vogel RA. Digital radiographic assess- chcsJM, Marcus ML, White CL. Rethrombosii after reperfusionwith streptokiment of coronary arterial diameter and videodensitometriccross-sectionalarea. nase:importance of geometry of residual lesions.Circulation 198459:991-999. Proc Sot Photooptical Inrtrum Eng 1986,626:334-341. 25. McMahon MM, Brown BG, Cukingnan R, Rolett EL, BolsenE, Frimer M, 8. Brown BG, BolsonE, Frimer M, DodgeHT. Quantitative coronary arteriograDodge HT. Quantitative coronary angiography: measurementof the ‘%ritical” phy. Estimation of dimensions,hemcdynamicresistance,and athemma massof stenosisin patientswith unstableangina and single-vesseldiseasewithout collatercoronary artery lesionsusing the arteriogram and digital computation. Ctrculaals. Circulation 1979,60:106-l13. tion 1977;55:329-337. 26. Wijns W, Sermys PW, Reiber JHC, van den Brand M, Simoons MJ, 9. Gould KL. Quantification of coronary artery stenosis in viva. Circ Res Kooijman CJ, Balakumaran K, Hugenholtz PG. Quantitative angiographyof the 1985;57:341-352. left anterior descendingcoronary artery: correlations with pressuregradient and 10. Amett EN, lsner JM, Redwood DR. Kent KM, Baker WP, Ackerstein H, results of exercise thallium scintigraphy. Circulation 1985;71:273-279. Roberts WC. Coronary artery narrowing in coronary heart disease:comparisonof 27. Zulstra F, van OmmerenJ, Reiber JHC, Serruys PW. Doesthe quantitative cineangiographic and necropsy findings. Ann Intern Med 1979,91:350-356. assessmentof coronary artery dimensionspredict the physiologicsignificanceof a 11. Vkxlaver Z, Frech R, Van Tassel RA, Edwards JE. Correlation of the coronary stenosis?Circulation 1987;75:1154-1161. antemortum coronary arteriogram and the postmortem specimen.Circulation 28. Wilson RF, Marcus ML, White CW. Prediction of the physiologicalsignifi1973;47:162-169. canceof coronary arterial lesionsby quantitative lesiongeometryin patients with 12. Grondin CM, Dyrda I, PasternacA, CampeauL, BourassaMG, Lesperance limited coronary artery disease.Circulation 1987:75:723-732. J. Discrepanciesbetween cineangiographicand postmortemfindings in patients 29. Nichols AB, Brown C, Han J, Nickoloff EL, EsserPD. Effect of coronary with coronary artery diseaseand recent myocardial revascularization.Circulation stenotic lesions on regional myocardial blood flow at rest. Circulation 1986; 1974;49:703-711. 74:746-757. 13. Reiber JHC, Kooijman CJ, Slager CJ, GerbrandsJJ, SchuurbiersJCH, Den 38. Milne ENC. The role and performanceof minute focal spotsin roentgenology Boer AD, Wijns W, Serruys PW, Hugenholtz PG. Coronary artery dimensions with special reference to magnification. CRC Crit Rev Radio1 Sci 1971:2:269from cineangiograro-methodology andvalidation of a computer-assistedanalysis 310. procedure. IEEE Trans Med Imaging 1984&U-51:131-141. 31. Seibert JA, Nalcioglu 0, Rceck WW. Characterization of the veiling glare 14. Mancini GBJ, Simon SB, McGillem MJ, LeFree MT, Friedman HZ, Vogel PSF in x-ray image intensified fluorcacopy. Med Phys 1984;11:172-179.

138

THE AMERICAN JOURNAL OF CARDIOLOGY VOLUME 64