Our results point out that certain conditions may lead to incomplete or wrong angiographic or CT findings in patients with AD. These conditions consist, at first, of at least 1 entry tear large enough to provide similar blood flow in the true as well as false lumina, thereby preventing the identification of any temporal and densitometric differences in contrast enhancement, which are the most specific signs of AD by angiography or CT.2-4 Second, as soon as the false lumen is completely tilled with thrombotic material, a clear delineation of the intima1 flap may be obscured on the CT scan or angiogram; however, thrombosis of the aorta may occur in conditions other than dissection (e.g., thrombosis of true aneurysm). Finally, the image quality of CT scans may be impaired by artifacts due to surgical clips, prosthetic devices, pacemaker wires or, as in a patient presented here, to the cannula of a tracheostoma. TEE may have certain limitations in the diagnosis of isolated short-distance type II AD due to the trachea obscuring a small region of the ascending aorta. In a
Right Ventricular Myocardial Resonance Imaging
recently published multicenter trial, no false-negative or false-positive TEE results were found in the area of the aortic arch and the descending aorta.5 The present report indicates that diagnosis of proximal AD extent using CT or angiography may be difficult at least in some patients with large entry tears or complete thrombosis of a lumen. Artifacts may also affect CT image quality. TEE may overcome the diagnostic difficulties under these conditions. 1. Borst HG. Laas J, Frank G, Haverich A. Surgical decision making in acute aortic dissection type A. Thorac Cardiooasc Surg 1987:35:134-135. 2. Shuford WH, Sybers RG, Weens HS. Problems in the aortographic diagnosis of dissect‘ng aneurysm of the aorta. N Engi J Med 1969:280:225-231. 3. Godwin JD, Herfkens RL, Skioldebrand CC, Federle MP, Lipton MJ. Evaluation of dissections and aneurysms of the thoracic aorta by conventional and dynamic CT scanning. Radiology 1980;136:125-133. 4. H&berg E. Wolverson M, Sundaram M, Canners J, Susman N. CT tindings in thoracic aortic dissection. AJR 1981:/36:/317. 5. E&l R. Engberding R, Daniel W. Roelandt J, Visser C, Rennollet H. Echocardiography in diagnosis of aortic dissection. Iancef 1989:/:/S460. 6. Daniel WG, Schrtier E, Miigge A, Lichtlen PR. Transesophageal echocardiography in infective endccarditis. Am J Cardiac Imaging 19882-78.-85.
-
Mass Quantification
with Magnetic
Edward S. Mackey, MD, Martin P. Sandler, MD, Robert M. Campbell, MD, Thomas P. Graham, James B. Atkinson, MD, PhD, Ronald Price, PhD, and Gordon A. Moreau, MD lterations of right ventricular (RV) size and shape are common in both congenital and acquired cardioA pulmonary disease. Noninvasive quantification of RV mass could enhance the assessment of various hemodynamic overloads in such patients, as well as response to treatment. Left ventricular mass can be estimated using magnetic resonance imagery,lh3 echocardiography, xray transmission computed tomography,’ single photon emission computed tomography6 and contrast ventricFrom the Division of Pediatric Cardiology and the Vanderbilt University School of Medicine, Vanderbilt Medical Center, D-2217 MCN, Nashville, Tennessee 37232-2572. Manuscript received June 20, 1989; revised manuscript received and accepted October 19, 1989.
Jr., MD,
ulography.’ RV mass has been estimated invasively in a single reports and a single abstract has been published measuring RV mass by x-ray transmission computed tomography.9 Gated magnetic resonance imagery provides excellent detail of cardiac anatomy10-‘2 through the use of orthogonal planes, as well as angled views that simulate echocardiographic imaging planes. This study evaluates magnetic resonance estimation RV mass in vitro. Twelve formalin-fixed, structurally normal human hearts obtained at necropsyfrom patients aged I day to 77 years werestudied. Causesof death were trauma (5), chronic medical disorders(4) and congenital defects(3). Each heart wasremoved after great vesseltranssection.
THE AMERICAN
JOURNAL
OF CARDIOLOGY
FEBRUARY
15. 1990
529
Postmortem clots were removed and gauze placed to maintain chamber lumen. Hearts were perfied with 10% formalin at 40 mm Hg for 24 to 48 hours. This fixation procedure can changeheart weight by 2 to 3% without exceeding5%.‘-’ Imaging wasperformed usinga 0.5-Tesla superconducting magnetic resonancescanner (Technicare). Images were Tl weighted with the pulse repetition frequency 500 ms and the echo interval 38 ms. Hearts wereplaced in a 28-cm head coil (2%mm field of view) and imaged in the left ventricular short axis plane from apex to base (Figure IA). Slice thickness equaled 3 mm. Total slices varied with heart size from 10 to 20 slices. Pixel size was 1 X 2 mm; this set the minimum pixel area uncertainty to 20 mm2, which was consideredinsignificant relative to other factors. A gadolinium phantom was present in each image for real space correlation. Images were photographed using a 2X magnification factor to improve border identification.
The ventricular endocardial and epicardial borders were traced using a Compaq Deskpro computer with Summagraphics Summasketch digitizing board. For each image, the RVfree wall was separatedfrom left ventricular free wall and the interventricular septum during tracing by following the arc of the interventricular septumcontinuous to the left ventricular epicardium. The left ventricular epicardial and endocardial borders were defined as the outer edge of signal intensity. All papillary muscleswereconsideredpart of the myocardiurn. RV borders were defined as areasof uniform intensity. Any areasof intensity not continuouswith the endocardium were not ‘included in the tracing. Epicardial fat was excluded from analysis. Figure 1B showsa representative image with the digitized borders shown. Venti-icular masswascalculated asfollows: for each slice, endocardial area was subtracted from epicardial area. The resulting ventricular slice area wasmultiplied by the slice thickness (3 mm) to obtain slice volume. Slice volume was multiplied by the specific density of cardiac muscle (1.055 g/o?) to give slice mass. The
100 90
z
100 r
70
80
.!?’ 60 al $ 50 .a 5 40
70
= f
t
TN CP* 60 -a e 50 oEa 40 $2 Em zo 30
NMR=0,95(A)+0.72
30 20
NMR=l.Ol (AbtOsl4 SEE= 8.9gm
IO n
“0
A
IO 20 30’ 40 50 60 70 80 90 100 anatomic
250
-
225
-
200
-
“0
mass (gml
IO 20 30 40 50 60 70 80 90 100 NMR mass (gm) OBSEVER4b-1
A 250 /-
,,
200 -
II
0
30
1
60
90
II
120
I
150
anatomic
III
180 210 240
J
270
300
mass (gm)
FlGURE2.~campdumofa~tomicmass(axis)loNMR resenmm) haglng-esth&d (-nuenak n8te)ferlwltlfemMde(A)8ndlsR~(B)pllJsvantricubrseptm9,respedive4y.ThcdashtdRnempmsenblinc0f
B m8ss
THE AMERICAN
JOURNAL
OF CARDIOLOGY
50
loo NMR
150
200
250
mass (gm)
(erdlFIGURE 3. lnbmbsew er VW -~~b’Vndkft(8)ventddw wQ--nncafidantity.Abbmviatiomuh~2.
ldedty.3EE=standardemueftheesthate. 530
0
VOLUME
65
fer NMR imaghqgm8m. 08sfl8d
line
TABLE I Analysis
of Ventricular
Mass (g)
Right Ventricle
Left Ventricle NMR
NMR
Pt
Age Ws)
Anatomic
Observer
1 2 3 4 5 6 7 8 9 10 11 12
Newborn 0.67 3 37 66 45 17 77 22 34 67
4.7 6.5 14.1 41.1 42.0 56.9 57.5 58.2 63.6 65.0 66.4 92.7
4.8 6.4 14.8 38.5 44.2 61.5 53.0 60.1 56.0 56.2 50.0 92.6
* Average of 2 measurements. NMR = mass as estimated wng nuclear
magnetic
resonance
l*
Observer
2
4.8 6.6 13.8 48.6 33.3 51.9 63.3 63.4 46.3 81.7 51.1 91.1
Anatomic
Observer
14.0 19.8 44.9 125.0 112.8 195.2 227.0 152.4 157.0 165.0 195.1 258.1
16.4 18.0 41.8 133.8 106.3 185.0 221.5 148.6 155.5 170.2 155.0 248.5
l*
Observer
2
16.1 17.5 40.8 109.0 109.0 167.9 194.6 121.5 150.6 150.6 146.8 248.0
imaging.
estimated massfor each ventricle was calculated by summing consecutiveslicesfrom apex to base. Imagesfor eachheart were traced independently by 2 observersto determine interobserver variation. To determine intraobserver variation, observerno. 1 repeatedthe measurementsseveral days later. After magnetic resonanceimagery analysis, the measured anatomic ventricular masswas obtained. All epicardial fat and large epicardial coronary arteries were removed from the ventricular muscle. Atria were removed by dissecting along the atrioventricular groove; all valves and anular rings were then removed. The RV free wall was separatedfrom the left ventricular free wall and the interventricular septum as described.Each ventricle was dried and weighed separately. Standard linear regressionwas used to compareanatomic massto magnetic resonancemassas measuredby observer no. 1 and to express interobserver variability (observer no. I usobserverno. 2). Intraobserver variability (observer no. 1) was expressedaspercent meanabsolute variation divided by meanmagneticresonancemass. Patient data and ventricular massare summarized in Table I. RV massrangedfrom 5 to 93 g; left ventricular plus ventricular septal weights varied from 14 to 258 g. Anatomic and magnetic resonanceventricular masscorrelated wellfor both the right ventricle and left ventricle plus ventricular septum (Figure 2A and B). Interobserver variability, depicted in Figure 3A and B, showed a close correlation. Intraobsetver (observer no. I) variability was 2%for both right ventricle and left ventricle plus ventricular septum.
The present study demonstrates that RV mass can be accurately and reproducibly estimated in autopsied human hearts using magnetic resonance imagery. Arcilla et alI8 estimated RV mass using a combined echocardiographic/angiographic technique, but this method has not been duplicated or applied clinically. Technical problems have complicated RV mass evaluation using echocardiography alone. The substernal location of the RV does not permit consistent delineation of this chamber and the irregular RV endocardium is seldom clearly outlined. An in vitro study was deemed an essential intermedi-
ate study for assessing sources of errors in magnetic resonance RV mass estimation prior to in vivo experiments. In vitro estimates eliminate errors due to motion blurring, which includes errors resulting from irregular heart rates and images acquired during motion. In vitro determinations also minimize uncertainties due to edge definition. With in vivo imaging, surrounding tissues and blood within the chambers produce signal levels above the background noise level and thus necessitate the development of an intensity-level criterion that is “defined” to be the ventricular edge. In vitro imaging produced high-contrast edges in which the ventricular boundary is defined to be pixel values above background. Problems in differentiating epicardial fat, papillary muscles and interventricular septum are common to both in vitro and in vivo determinations. The results of this study should, therefore, be interpreted as the minimum volume errors which can be achieved using magnetic resonance imaging methods with the specified image resolution and slice thickness for ventricular volume estimates. The ability to obtain resolution satisfactory for crisp edge detection for accurate RV volumes and mass analysis in vivo remains unproven. In this study, a slice thickness of 3 mm was chosen to minimize partial volume effects created by oblique sectioning. Partial volume effects may cause errors in mass evaluation through difficulties in defining endocardial and epicardial borders. The imaging plane for this study was perpendicular to the long axis of the left ventricle. Coupled with the narrowed slice thickness, these angled views help to minimize left ventricular partial volume effects. Due to the complex geometric configuration of the right ventricle, there are no standard orthogonal views for RV imaging and the use of narrow slice thickness allows for more reliable and reproducible RV mass determinations. Recent studies indicate that left ventricular hypertrophy, together with abnormal ventricular function, may serve as a substrate for arrhythmia.14 Likewise, Garson et al15 have reported an increased risk for serious ventricular arrhythmias in postoperative tetralogy of Fallot patients with residual RV hypertension and dysfunction. Thus, the noninvasive estimate of RV mass would provide a
THE AMERICAN
JOURNAL
OF CARDIOLOGY
FEBRUARY
15, 1990
531
powerful new method to aid in overall assessment of a patient’s response to abnormal right heart hemodynamits, as well as providing further insight to possible correlations between mass and significant ventricular arrhythmias. The RV and left ventricular mass values resulting from this study will serve as a baseline for subsequent in vivo studies. Substantial improvements in magnetic resonance imaging will be needed before the type of accuracy reported herein can be reproduced with in vivo studies. 1. Florentine MD, Grosskreutz CL, Chang W, Harm&t JA, Dunn VD, Ehrhardt JC, Fleagle SR, Collins SM, Marcus ML, Skorton DJ. Measurement of left ventricular mass in viva using gated nuclear magnetic resonance imaging. JACC 1986:8:107-l 12. 2. Keller AM, Peshock RM, Malloy
CR, Buja LM, Nunnaly R, Parkey RW, Willerson JT. In viva measurement of myocardial mass using nuclear magnetic resonance imaging. JACC 1986;8:113-117. 3. Maddahi J, Crues J, Berman DS, Mericle J, Becerra A, Garcia EV, Henderson R, Bradley W. Quantitation of left ventricular myocardial mass by gatcd proton nuclear magnetic resonance imaging. JACC 1987;10:682-692. 4. Byrd BF, Wahr D, Wang YS, Bouchard A, Schiller N. Left ventricular mass and volume/mass determined by twodimensional echocardiography in normal adults JACC 1985,6:1021-1025. 5. Feiring AJ, Rumberger JA, Reiter SJ, Skorton DJ, Collins SM. Lipton MJ, Higgins CB, El1 S, Marcus ML. Determination of left ventricular mass in dogs
532
THE AMERICAN
JOURNAL
OF CARDIOLOGY
VOLUME
65
with rapid-acquisition 198%72:1355-l
cardiac computed tomographic
scanning. Circuhtion
364.
6. Wolfe CL, Corbett JR, Lewis SE, Buja LM, Willerson JT. Determination of left ventricular mass by single-photon emission computed tomography with thallium-20 1. Am J Cardiol 198354:1365-l 368. 7. Rackley CE, Dodge HT, Coble YD, Hay RE. A method for determining left ventricular mass in man. Circulafion 1964;24:666-671. 6. Arcilla RA, Mathew R. Sodt P, Lester L, Cahill N, Thilenius OG. Right ventricular mass estimation by angiocchocardiography. Cot/ret Cardiwasc Diagn 1976;2:125-136.
9. Hajducaok ZD, Weiss RM, Marcus ML. Right ventricular ma= can bc accurately assessed by ultrafast computed tomography (abstr). JACC 1989; 13:8A.
10. Higgins CB, Byrd BF III, McNamara MT, Lamer R, Lipton MI, Botrinick E, Schiller NB, Croks LE. Kaufman L. Magnetic resonance imaging of the heart: a review of the experience in 172 subjects. Radiology 1982;155:671-679. 11. Dinsmore R, Wismer GL, Levine RA, Okada RD, Brady TJ. Magnetic resonance imaging of the heart: positioning and gradient angle selection for optimal imaging planes. AJR 1984;143:1135-1142. 12. Dinsmore RE, Wismer GL, Miller SW, Thompson R, Johnson DL, Liu P, Okada RD, Saini S, Brady TJ. Magnetic resonance imaging of the heart using image planes oriented to cardiac axes: experience with 100 cases. AJR 1985;145:1177-1183.
13. Bove KE, Rowlands DT, Scott RC. Observations on the assessment of cardiac hypcrtrophy utilizing a chamber partition technique. Circulation 1966;33:558568. 14. Kowey PR, Eisenberg R, Engel TR. Sustained arrhythmias obstructive cardiomyopathy. N Engl J Med 1984:310:1566-1569.
in hypertrophic
15. Garson A Jr, Randall DC, Gillette PC, Smith RT, Moak JP, McVay P, McNamara DG. Prevention of sudden death after repair of tetralogy of fallot treatment of ventricular arrhythmias. JACC 1985,6:221-227.
Right Ventricular Myocardial Resonance Imaging
recently published multicenter trial, no false-negative or false-positive TEE results were found in the area of the aortic arch and the descending aorta.5 The present report indicates that diagnosis of proximal AD extent using CT or angiography may be difficult at least in some patients with large entry tears or complete thrombosis of a lumen. Artifacts may also affect CT image quality. TEE may overcome the diagnostic difficulties under these conditions. 1. Borst HG. Laas J, Frank G, Haverich A. Surgical decision making in acute aortic dissection type A. Thorac Cardiooasc Surg 1987:35:134-135. 2. Shuford WH, Sybers RG, Weens HS. Problems in the aortographic diagnosis of dissect‘ng aneurysm of the aorta. N Engi J Med 1969:280:225-231. 3. Godwin JD, Herfkens RL, Skioldebrand CC, Federle MP, Lipton MJ. Evaluation of dissections and aneurysms of the thoracic aorta by conventional and dynamic CT scanning. Radiology 1980;136:125-133. 4. H&berg E. Wolverson M, Sundaram M, Canners J, Susman N. CT tindings in thoracic aortic dissection. AJR 1981:/36:/317. 5. E&l R. Engberding R, Daniel W. Roelandt J, Visser C, Rennollet H. Echocardiography in diagnosis of aortic dissection. Iancef 1989:/:/S460. 6. Daniel WG, Schrtier E, Miigge A, Lichtlen PR. Transesophageal echocardiography in infective endccarditis. Am J Cardiac Imaging 19882-78.-85.
-
Mass Quantification
with Magnetic
Edward S. Mackey, MD, Martin P. Sandler, MD, Robert M. Campbell, MD, Thomas P. Graham, James B. Atkinson, MD, PhD, Ronald Price, PhD, and Gordon A. Moreau, MD lterations of right ventricular (RV) size and shape are common in both congenital and acquired cardioA pulmonary disease. Noninvasive quantification of RV mass could enhance the assessment of various hemodynamic overloads in such patients, as well as response to treatment. Left ventricular mass can be estimated using magnetic resonance imagery,lh3 echocardiography, xray transmission computed tomography,’ single photon emission computed tomography6 and contrast ventricFrom the Division of Pediatric Cardiology and the Vanderbilt University School of Medicine, Vanderbilt Medical Center, D-2217 MCN, Nashville, Tennessee 37232-2572. Manuscript received June 20, 1989; revised manuscript received and accepted October 19, 1989.
Jr., MD,
ulography.’ RV mass has been estimated invasively in a single reports and a single abstract has been published measuring RV mass by x-ray transmission computed tomography.9 Gated magnetic resonance imagery provides excellent detail of cardiac anatomy10-‘2 through the use of orthogonal planes, as well as angled views that simulate echocardiographic imaging planes. This study evaluates magnetic resonance estimation RV mass in vitro. Twelve formalin-fixed, structurally normal human hearts obtained at necropsyfrom patients aged I day to 77 years werestudied. Causesof death were trauma (5), chronic medical disorders(4) and congenital defects(3). Each heart wasremoved after great vesseltranssection.
THE AMERICAN
JOURNAL
OF CARDIOLOGY
FEBRUARY
15. 1990
529
Postmortem clots were removed and gauze placed to maintain chamber lumen. Hearts were perfied with 10% formalin at 40 mm Hg for 24 to 48 hours. This fixation procedure can changeheart weight by 2 to 3% without exceeding5%.‘-’ Imaging wasperformed usinga 0.5-Tesla superconducting magnetic resonancescanner (Technicare). Images were Tl weighted with the pulse repetition frequency 500 ms and the echo interval 38 ms. Hearts wereplaced in a 28-cm head coil (2%mm field of view) and imaged in the left ventricular short axis plane from apex to base (Figure IA). Slice thickness equaled 3 mm. Total slices varied with heart size from 10 to 20 slices. Pixel size was 1 X 2 mm; this set the minimum pixel area uncertainty to 20 mm2, which was consideredinsignificant relative to other factors. A gadolinium phantom was present in each image for real space correlation. Images were photographed using a 2X magnification factor to improve border identification.
The ventricular endocardial and epicardial borders were traced using a Compaq Deskpro computer with Summagraphics Summasketch digitizing board. For each image, the RVfree wall was separatedfrom left ventricular free wall and the interventricular septum during tracing by following the arc of the interventricular septumcontinuous to the left ventricular epicardium. The left ventricular epicardial and endocardial borders were defined as the outer edge of signal intensity. All papillary muscleswereconsideredpart of the myocardiurn. RV borders were defined as areasof uniform intensity. Any areasof intensity not continuouswith the endocardium were not ‘included in the tracing. Epicardial fat was excluded from analysis. Figure 1B showsa representative image with the digitized borders shown. Venti-icular masswascalculated asfollows: for each slice, endocardial area was subtracted from epicardial area. The resulting ventricular slice area wasmultiplied by the slice thickness (3 mm) to obtain slice volume. Slice volume was multiplied by the specific density of cardiac muscle (1.055 g/o?) to give slice mass. The
100 90
z
100 r
70
80
.!?’ 60 al $ 50 .a 5 40
70
= f
t
TN CP* 60 -a e 50 oEa 40 $2 Em zo 30
NMR=0,95(A)+0.72
30 20
NMR=l.Ol (AbtOsl4 SEE= 8.9gm
IO n
“0
A
IO 20 30’ 40 50 60 70 80 90 100 anatomic
250
-
225
-
200
-
“0
mass (gml
IO 20 30 40 50 60 70 80 90 100 NMR mass (gm) OBSEVER4b-1
A 250 /-
,,
200 -
II
0
30
1
60
90
II
120
I
150
anatomic
III
180 210 240
J
270
300
mass (gm)
FlGURE2.~campdumofa~tomicmass(axis)loNMR resenmm) haglng-esth&d (-nuenak n8te)ferlwltlfemMde(A)8ndlsR~(B)pllJsvantricubrseptm9,respedive4y.ThcdashtdRnempmsenblinc0f
B m8ss
THE AMERICAN
JOURNAL
OF CARDIOLOGY
50
loo NMR
150
200
250
mass (gm)
(erdlFIGURE 3. lnbmbsew er VW -~~b’Vndkft(8)ventddw wQ--nncafidantity.Abbmviatiomuh~2.
ldedty.3EE=standardemueftheesthate. 530
0
VOLUME
65
fer NMR imaghqgm8m. 08sfl8d
line
TABLE I Analysis
of Ventricular
Mass (g)
Right Ventricle
Left Ventricle NMR
NMR
Pt
Age Ws)
Anatomic
Observer
1 2 3 4 5 6 7 8 9 10 11 12
Newborn 0.67 3 37 66 45 17 77 22 34 67
4.7 6.5 14.1 41.1 42.0 56.9 57.5 58.2 63.6 65.0 66.4 92.7
4.8 6.4 14.8 38.5 44.2 61.5 53.0 60.1 56.0 56.2 50.0 92.6
* Average of 2 measurements. NMR = mass as estimated wng nuclear
magnetic
resonance
l*
Observer
2
4.8 6.6 13.8 48.6 33.3 51.9 63.3 63.4 46.3 81.7 51.1 91.1
Anatomic
Observer
14.0 19.8 44.9 125.0 112.8 195.2 227.0 152.4 157.0 165.0 195.1 258.1
16.4 18.0 41.8 133.8 106.3 185.0 221.5 148.6 155.5 170.2 155.0 248.5
l*
Observer
2
16.1 17.5 40.8 109.0 109.0 167.9 194.6 121.5 150.6 150.6 146.8 248.0
imaging.
estimated massfor each ventricle was calculated by summing consecutiveslicesfrom apex to base. Imagesfor eachheart were traced independently by 2 observersto determine interobserver variation. To determine intraobserver variation, observerno. 1 repeatedthe measurementsseveral days later. After magnetic resonanceimagery analysis, the measured anatomic ventricular masswas obtained. All epicardial fat and large epicardial coronary arteries were removed from the ventricular muscle. Atria were removed by dissecting along the atrioventricular groove; all valves and anular rings were then removed. The RV free wall was separatedfrom the left ventricular free wall and the interventricular septum as described.Each ventricle was dried and weighed separately. Standard linear regressionwas used to compareanatomic massto magnetic resonancemassas measuredby observer no. 1 and to express interobserver variability (observer no. I usobserverno. 2). Intraobserver variability (observer no. 1) was expressedaspercent meanabsolute variation divided by meanmagneticresonancemass. Patient data and ventricular massare summarized in Table I. RV massrangedfrom 5 to 93 g; left ventricular plus ventricular septal weights varied from 14 to 258 g. Anatomic and magnetic resonanceventricular masscorrelated wellfor both the right ventricle and left ventricle plus ventricular septum (Figure 2A and B). Interobserver variability, depicted in Figure 3A and B, showed a close correlation. Intraobsetver (observer no. I) variability was 2%for both right ventricle and left ventricle plus ventricular septum.
The present study demonstrates that RV mass can be accurately and reproducibly estimated in autopsied human hearts using magnetic resonance imagery. Arcilla et alI8 estimated RV mass using a combined echocardiographic/angiographic technique, but this method has not been duplicated or applied clinically. Technical problems have complicated RV mass evaluation using echocardiography alone. The substernal location of the RV does not permit consistent delineation of this chamber and the irregular RV endocardium is seldom clearly outlined. An in vitro study was deemed an essential intermedi-
ate study for assessing sources of errors in magnetic resonance RV mass estimation prior to in vivo experiments. In vitro estimates eliminate errors due to motion blurring, which includes errors resulting from irregular heart rates and images acquired during motion. In vitro determinations also minimize uncertainties due to edge definition. With in vivo imaging, surrounding tissues and blood within the chambers produce signal levels above the background noise level and thus necessitate the development of an intensity-level criterion that is “defined” to be the ventricular edge. In vitro imaging produced high-contrast edges in which the ventricular boundary is defined to be pixel values above background. Problems in differentiating epicardial fat, papillary muscles and interventricular septum are common to both in vitro and in vivo determinations. The results of this study should, therefore, be interpreted as the minimum volume errors which can be achieved using magnetic resonance imaging methods with the specified image resolution and slice thickness for ventricular volume estimates. The ability to obtain resolution satisfactory for crisp edge detection for accurate RV volumes and mass analysis in vivo remains unproven. In this study, a slice thickness of 3 mm was chosen to minimize partial volume effects created by oblique sectioning. Partial volume effects may cause errors in mass evaluation through difficulties in defining endocardial and epicardial borders. The imaging plane for this study was perpendicular to the long axis of the left ventricle. Coupled with the narrowed slice thickness, these angled views help to minimize left ventricular partial volume effects. Due to the complex geometric configuration of the right ventricle, there are no standard orthogonal views for RV imaging and the use of narrow slice thickness allows for more reliable and reproducible RV mass determinations. Recent studies indicate that left ventricular hypertrophy, together with abnormal ventricular function, may serve as a substrate for arrhythmia.14 Likewise, Garson et al15 have reported an increased risk for serious ventricular arrhythmias in postoperative tetralogy of Fallot patients with residual RV hypertension and dysfunction. Thus, the noninvasive estimate of RV mass would provide a
THE AMERICAN
JOURNAL
OF CARDIOLOGY
FEBRUARY
15, 1990
531
powerful new method to aid in overall assessment of a patient’s response to abnormal right heart hemodynamits, as well as providing further insight to possible correlations between mass and significant ventricular arrhythmias. The RV and left ventricular mass values resulting from this study will serve as a baseline for subsequent in vivo studies. Substantial improvements in magnetic resonance imaging will be needed before the type of accuracy reported herein can be reproduced with in vivo studies. 1. Florentine MD, Grosskreutz CL, Chang W, Harm&t JA, Dunn VD, Ehrhardt JC, Fleagle SR, Collins SM, Marcus ML, Skorton DJ. Measurement of left ventricular mass in viva using gated nuclear magnetic resonance imaging. JACC 1986:8:107-l 12. 2. Keller AM, Peshock RM, Malloy
CR, Buja LM, Nunnaly R, Parkey RW, Willerson JT. In viva measurement of myocardial mass using nuclear magnetic resonance imaging. JACC 1986;8:113-117. 3. Maddahi J, Crues J, Berman DS, Mericle J, Becerra A, Garcia EV, Henderson R, Bradley W. Quantitation of left ventricular myocardial mass by gatcd proton nuclear magnetic resonance imaging. JACC 1987;10:682-692. 4. Byrd BF, Wahr D, Wang YS, Bouchard A, Schiller N. Left ventricular mass and volume/mass determined by twodimensional echocardiography in normal adults JACC 1985,6:1021-1025. 5. Feiring AJ, Rumberger JA, Reiter SJ, Skorton DJ, Collins SM. Lipton MJ, Higgins CB, El1 S, Marcus ML. Determination of left ventricular mass in dogs
532
THE AMERICAN
JOURNAL
OF CARDIOLOGY
VOLUME
65
with rapid-acquisition 198%72:1355-l
cardiac computed tomographic
scanning. Circuhtion
364.
6. Wolfe CL, Corbett JR, Lewis SE, Buja LM, Willerson JT. Determination of left ventricular mass by single-photon emission computed tomography with thallium-20 1. Am J Cardiol 198354:1365-l 368. 7. Rackley CE, Dodge HT, Coble YD, Hay RE. A method for determining left ventricular mass in man. Circulafion 1964;24:666-671. 6. Arcilla RA, Mathew R. Sodt P, Lester L, Cahill N, Thilenius OG. Right ventricular mass estimation by angiocchocardiography. Cot/ret Cardiwasc Diagn 1976;2:125-136.
9. Hajducaok ZD, Weiss RM, Marcus ML. Right ventricular ma= can bc accurately assessed by ultrafast computed tomography (abstr). JACC 1989; 13:8A.
10. Higgins CB, Byrd BF III, McNamara MT, Lamer R, Lipton MI, Botrinick E, Schiller NB, Croks LE. Kaufman L. Magnetic resonance imaging of the heart: a review of the experience in 172 subjects. Radiology 1982;155:671-679. 11. Dinsmore R, Wismer GL, Levine RA, Okada RD, Brady TJ. Magnetic resonance imaging of the heart: positioning and gradient angle selection for optimal imaging planes. AJR 1984;143:1135-1142. 12. Dinsmore RE, Wismer GL, Miller SW, Thompson R, Johnson DL, Liu P, Okada RD, Saini S, Brady TJ. Magnetic resonance imaging of the heart using image planes oriented to cardiac axes: experience with 100 cases. AJR 1985;145:1177-1183.
13. Bove KE, Rowlands DT, Scott RC. Observations on the assessment of cardiac hypcrtrophy utilizing a chamber partition technique. Circulation 1966;33:558568. 14. Kowey PR, Eisenberg R, Engel TR. Sustained arrhythmias obstructive cardiomyopathy. N Engl J Med 1984:310:1566-1569.
in hypertrophic
15. Garson A Jr, Randall DC, Gillette PC, Smith RT, Moak JP, McVay P, McNamara DG. Prevention of sudden death after repair of tetralogy of fallot treatment of ventricular arrhythmias. JACC 1985,6:221-227.
