Downloaded from tech.snmjournals.org by on October 31, 2018. For personal use only.

IMAGING

A New Method for Determination of Left-to-Right

Cardiac Shunts

J. S. Preslar, M. T. Madsen, and E. E. Argenyi

University q(Iowa Hospitals and Clinics, Iowa City, Iowa

4
{" Na) (2). By the 1960s, the pulmonary T AC was observed

Left-to-right cardiac shunt quantitation can be determined to better delineate between pulmonary and recirculation
from first pass radionuclide angiography. The traditional curves (3).
technique fits a gamma variate to the lung time-activity curve In 1974, Folse and Braunwald used the count ratio method
(TAC) and a second gamma variate to the recirculation (C 2/C) to describe the degree of shunting. The two main
portion of the curve. Problems may arise, however, with the problems with this technique are that single counts are greatly
fitting of the gamma variate to the subtracted recirculation influenced by statistical variation and that all laboratories
curve. We have investigated a new technique in which the
need to develop their own set of normal values.
recirculation fitting is no longer required. The new method
fits a gamma variate to the first-pass portion of the lung TAC. By 1975. Alderson used the exponential area ratio method
This gamma variate is used to generate a curve which simulates (X/Y) to establish shunt values. Alderson stated that the
the expected shape of a normal lung TAC in response to downslope of the lung curve has a monoexponential shape.
systemic recirculation. The simulated data is then subtracted Its value from the peak activity to the beginning of the early
from the observed lung TAC. A correlation coefficient ofr = recirculation is exponentially extrapolated to I% of the max-
0.87 was obtained when these two methods were compared. imum (3). This method gives a quadratic equation for X/Y
We now prefer the new method since it overcomes the problem versus a percent of shunt (4). The main problem with this
offitting the subtracted recirculation position of the curve. method is that normal values need to be established for each
laboratory (3).
In 1973, Maltz and Treves used a gamma variate method
A left-to-right cardiac shunt is defined as the premature return (5,6) to describe a curve with a rapid upslope, a peak, and an
of part of the systemic blood flow to the pulmonary circula- exponential downslope, using the following equation.
tion. Regardless of its cause, the quantitative assessment of a
cardiac shunt by noninvasive technique plays a key role in
the clinical management of a cardiac patient with a ventricular
septal defect (VSD). atrial septal defect (ASD), or patent
C(t) equals the concentration of the tracer at time t, a and B
ductus arteriosus (PDA) ( 1 ).
equal arbitrary parameters obtained from the curve fit, and
Historically, first-pass radionuclide angiography has been
K equals the constant scale factor.
used to evaluate the degree of shunting. The theory behind
The first-pass portion of the pulmonary curve is fitted by
this technique is that if a tracer is delivered to the central
the least squares technique to a gamma variate function. The
circulation as a compact bolus with uniform mixing in the
points chosen to fit the gamma curve are those that are
blood, and there is constant detector counting efficiency, the
approximately 10% of the maximum on the upslope to those
time-activity curve (T AC) of the tracer's first transit through
just before the shunt recirculation peak, which usually occurs
the lung and early recirculation will be proportional to the
respective blood volumes (2). below 70% of the maximum on the downslope. The fitted
lung area is represented as A 1; this is shown in Figure I. The
gamma function curve is then subtracted from the original
DEVELOPMENT OF TIME-ACTIVITY CURVES T AC to give the recirculation curve. This recirculation curve
As early as 1948, Prinzmeter described a T AC over the is then fitted by least squares to another gamma variate
precordium after an intravenous injection of sodium-24 function. The points chosen on the recirculation curve are
10% of the maximum on the upslope and the point after the
maximum on the downslope. This fitted recirculation area is
For reprints contact: J. S. Preslar. CNMT. Division of Nuclear Medicine.
represented as A2 (Fig. I). It is proportional to the shunt,
University of Iowa Hospitals and Clinics. Iowa City. Iowa 52242. which is expressed in terms of quantitative pulmonary/quan-

216 JOURNAL OF NUCLEAR MEDICINE TECHNOLOGY

 Downloaded from tech.snmjournals.org by on October 31, 2018. For personal use only.

FIG. 1. Maltz-Treves method of cardiac shunt quantification. A

gamma variate is fitted to the first transit portion of the observed lung FIG. 2. Proposed method for cardiac shunt quantification.
time-activity curve and subtracted. A second gamma variate is fitted
to the difference curve.

titative systemic blood flow (QP/QS).

QP/QS = AI
A1- Ac

This area ratio method (QP/QS) is only correct for shunts

between I and 3: I (5).
There is a poor fit at the tail of the observed curve when
the curve declines faster than the exponential function. An-
other problem is the difficulty of fitting the gamma variate to
the subtracted recirculation curve (3, 7,8). This method may
be inadequate because it is often difficult to distinguish be-
tween a small shunt and systemic recirculation (9). Since
there are rarely any definitive landmarks, the results are
generally arbitrary and operator dependent (8). To overcome FIG. 3. Normal lung time-activity curve (TAC). The lung radioactiv-
this problem, we have investigated a new technique in which ity falls as the bolus passes through the lungs and then rises as the
bolus recirculates. Eleven lung TAGs were examined from patients
the recirculation fitting is no longer required.
with no evidence of cardiac shunts. The values found at points B and
Care shown in Table 1.
MATERIALS AND METHODS

Method
Our technique fits the first-pass portion of the bolus through
the lung to a gamma variate function as does the Maltz-Treves TABLE 1. Normal Lung Data
method (5). This gamma variate fit is used to generate a curve Recirculation Recirculation
that simulates the expected shape of a normal lung T AC in Patient Minimum Maximum
response to systemic recirculation. The expected normal lung (B)* (C)*
curve is constructed by adding the scaled time integral of the 1 11% 24%
gamma variate to the initial gamma variate curve (10). The 2 17% 30%
integral function is scaled to equal the recirculation portion 3 15% 24%
of the observed lung curve, constrained by a maximum value 4 13% 26%
5 11% 25%
of25% of the curve's maximum (Fig. 2). The 25% maximum 6 16% 25%
level and the time offset of the scaled integral were determined 7 15% 29%
empirically from the inspection of normal lung T ACs (Fig. 3 8 12% 20%
and Table I). 9 17% 29%
The abnormal recirculation is then found by subtracting 10 19% 30%
11 12% 25%
the simulated curve or estimated normal lung curve, point by
point, from the observed lung curve. QP /QS is calculated as Mean 14.4% ± 2.7 26% ± 3.1
A 1/(AI - A2 ) where A 1 is the integral of the gamma variate • Minimum (B) and maximum (C) recirculation points are shown on
and Ac is the integral of the difference curve (Fig. 4 ). the normal lung curve in Figure 3.

VOLUME 19, NUMBER 4, DECEMBER 1991 217

 Downloaded from tech.snmjournals.org by on October 31, 2018. For personal use only.

FIG. 5. Plot of OP/OS values from Maltz-Treves method versus

FIG. 4. Calculation of the QPjQS in the new method. new method.

Patient Studies between the two methods with a correlation coefficient of r =

Scintigraphic studies were performed on 20 patients who 0.87.
were divided into two groups. The first group consisted of 15 When different operators used the traditional method to
patients. referred to the nuclear medicine department to rule calculate the QP/QS values on simulated lung TACs with
out left-to-right shunt. The second group consisted of five simulated cardiac shunts. the interoperator correlation coef-
patients, referred to the nuclear medicine department for ficient of QP/QS values varied from r = 0.85 to r = 0.90.
other types of cardiac studies, and we considered it unlikely When the same operators used the new method to calculate
that these patients would have a shunt. QP/QS values on the same simulated lung TACs, the inter-
All of the patients were studied with a Siemens LEM operator correlation coefficient varied from r = 0.96 to r =
(Siemens Gammasonics, Shaumburg. IL) small field of view 0.97 (Table 2). The wider range of QP/QS values obtained
scintillation camera. An intravenous line was established in when using the traditional method is probably due to variation
the right medial basilic vein using an infusion set with a # 18- in the selection of boundary points for the fitting of the
20 gauge needle. Twenty-five mCi oftechnetium-99m DTPA subtracted recirculation portion of the curve.
(Squibb Diagnostics, Princeton, NJ) in a volume of 0.5 ml or
less was then rapidly injected, followed by a bolus flush of 10 DISCUSSION
ml of normal saline. We have determined that this technique
provides a satisfactory compact bolus in the majority of Each method is thought to be a valid analytical method for
patients (FWHM = 1.5 sec/FW 1/10 M = 3 sec). In 10%- determining QP/QS values for the clinical management of a
18% of patients, the technique produces an unsatisfactory left-to-right cardiac shunt. A few problems have arisen with
bolus score; in these cases we perform a deconvolution on the the Maltz-Treves method since it requires two fittings of the
lung curve (8, 11. 12). lung T AC. When fitting the second gamma variate function,
operators have found that there are no defined landmarks to
Data Acquisition guide them in the selection of points. In addition, the calcu-
Data were acquired at a rate of 2 frames/sec during the first lated result seems to be sensitive to the operator's selection of
transit of the tracer through the heart. The camera was these boundary points.
positioned for an anterior view in order to best visualize the The new method involves operator intervention in only
superior vena cava (SVC) and right lung (RL). The scinti- one part of the curve. and the accuracy of the QP/QS value
graphic data were then analyzed using two regions of interest depends only on how well the normal lung recirculation can
to produce T ACs of the SVC and RL. be estimated. Since there is less operator influence, the new
method tends to yield results that are more systematically
reproducible, and the method is easier to use. This is an
RESULTS important feature since left-to-right cardiac shunt studies are
infrequently performed in many clinics, and they are often
Calculations were done to determine the QP/QS values analyzed by individuals with limited experience.
with the Maltz-Treves method (MDS A~ Computer, Medical Under certain conditions, operators may have problems
Data Systems, Ann Arbor, MI and MIPS version 2.0 software. calculating accurate QP/QS values using either analytical
Medasys Inc., Ann Arbor, MI) and the new method. Both of method. If blood is continually shunted and contributes to
the analytic methods were also evaluated for intraoperator early recirculation, then there may be an overestimation of a
variation. A comparison of the results of the two methods is large shunt. This inability to discriminate systemic recircula-
shown in Figure 5. Relatively good agreement was found tion will produce false QP/QS values. The validity of the lung

218 JOURNAL OF NUCLEAR MEDICINE TECHNOLOGY

 Downloaded from tech.snmjournals.org by on October 31, 2018. For personal use only.

TABLE 2. Three Operators' Determinations of Simulated Cardiac Shunts

OP/QS Values
Simulated
Operator 1 Operator 2 Operator 3
Curves
New Old New Old New Old

1.2 1.25 1.2 1.25 1.4 1.25 1.0

1.5 1.53 1.6 1.53 2.1 1.4 1.4
1.75 1.91 2.1 1.74 2.0 1.55 2.5
2.0 2.07 2.0 1.92 2.3 2.02 2.0
1.2 1.24 1.5 1.23 1.3 1.24 1.1
1.5 1.48 1.5 1.45 1.7 1.48 1.5
1.75 1.70 1.8 1.75 2.1 1.58 1.8
2.0 2.10 2.1 2.07 3.2 2.03 2.2
1.2 1.25 1.4 1.28 1.4 1.18 1.4
1.5 1.57 1.7 1.46 1.7 1.57 1.7
1.75 1.8 1.8 1.66 3.2 1.8 2.2
2.0 2.04 2.2 2.0 3.3 2.0 2.6
1.2 1.25 1.2 1.22 1.3 1.22 1.1
1.5 1.45 1.7 1.47 1.7 1.45 1.6
1.75 1.68 2.3 1.87 2.3 1.82 2.7
2.0 1.76 2.4 1.88 3.3 1.97 2.3
1.2 1.22 1.3 1.24 1.4 1.28 1.1
1.5 1.51 1.8 1.44 1.8 1.51 1.8
1.75 1.77 2.2 1.73 2.1 1.77 2.4
2.0 2.04 2.1 1.99 2.2 2.04 2.1

T AC may also be altered when the bolus injection is too slow 3. Alderson PO. Jost RG. Strauss A W. et al. Radionuclide angiocardiography.
Improved diagnosis and quantitation of left to right shunts using area ratio
or when there are circulatory abnormalities.
techniques in children. Circulation 1975:51:1136-1143.
4. Anderson PAW. Jones RH, Sabiston DC. Quantitation of left to right
CONCLUSION cardiac shunts with radionuclide angiocardiography. Circulation
1974:49:512-516.
Based on the results of our study comparing the two meth- 5. Maltz DL Treves S. Quantitative radionuclide angiocardiography. Deter-
ods. we have chosen to use the new method. It yields more mination ofQp:Qs in children. Circulation 1973:47:1049-1056.
reproducible results and provides a more time efficient 6. Treves S. Detection and quantitation of cardiovascular shunts with com-
monly available radionuclides. Sl'/lun Nucl Med 1980: I 0: 16-26.
method of determining QP/QS values.
7. Alderson PO. Guadiani VA. Watson DC Mendenhall KG. Donovan RC.
Quantitative radionuclide angiocardiography in animals with experimental
ACKNOWLEDGMENTS atrial septal defects. J A'ucl .\fed 1978:19:364-369.
8. Kuruc A. Treves S. Smith W. et al. An automated algorithm for radio-
This manuscript would not have been possible without the nuclide angiocardiographic quantitation of circulatory shunting. Com-
efforts of the nuclear medicine staff in our hospital. Also, puters and Biomedical Rl'.lmrch 1984:17:481-493.
9. Houser TS. Macintyre WJ. Cook SA. et al. Recirculation subtraction for
thanks to Pat Van Dyke for her secretarial skills in putting
analysis of left-to-right cardiac shunts: concise communication. J Nucl Med
together this manuscript. 1981:22: I 033-1038.
10. Kveder M, Bajzer Z, :-.losil J. A mathematical model for the quantitative
REFERENCES study of left to right cardiac shunt. Ph.rs Med Biol1985:30:207-215.
11. Alderson PO. Douglass KH. Mendenhall KG. et al. Deconvolution analysis
I. Parker JA, Treves S. Radionuclide detection. localization, and quantitation in radionuclide quantitation of left-to-right cardiac shunts. J 1\'uc/ !vied
ofintracardiac shunts and shunts between great arteries. ProK in Cardiovasc 1979:20:502-506.
Dis 1977;20: 121-150. 12. Ham HR. Dobbeleir A. Yiart P. et al. Radionuclide quantitation of left-
' Alderson PO. Basic principles of shunt quantitation by radionuclide angio- to-right cardiac shunts using deconvolution analysis: concise communica-
cardiography. App. Radiologr/.1\'uc/ear .\/edicine 1980: Nov/Dec. tion. J .\'uc/ .\fed 1981:22:688-692.

VOLUME 19, NUMBER 4, DECEMBER 1991 219

 Downloaded from tech.snmjournals.org by on October 31, 2018. For personal use only.

A New Method for Determination of Left-to-Right Cardiac Shunts

J. S. Preslar, M. T. Madsen and E. E. Argenyi

J. Nucl. Med. Technol. 1991;19:216-219.

This article and updated information are available at:

http://tech.snmjournals.org/content/19/4/216

Information about reproducing figures, tables, or other portions of this article can be found online at:
http://tech.snmjournals.org/site/misc/permission.xhtml

Information about subscriptions to JNMT can be found at:

http://tech.snmjournals.org/site/subscriptions/online.xhtml

Journal of Nuclear Medicine Technology is published quarterly.

SNMMI | Society of Nuclear Medicine and Molecular Imaging
1850 Samuel Morse Drive, Reston, VA 20190.
(Print ISSN: 0091-4916, Online ISSN: 1535-5675)

© Copyright 1991 SNMMI; all rights reserved.