|
|
|
Why Do We Perform Quantitative Left-Ventricular
Angiography?
LV angiography is relatively easy to do and tells us quite
a bit about the condition of the heart. By just viewing the
angiogram, we can qualitatively assess the size of the
left ventricle and its pumping efficiency, the condition of
the mitral and aortic valves and something about the adequacy
of the coronary artery blood supply. If we quantitatively
measure left-ventricular volumes, we can calculate cardiac output and ejection fraction
- important indicators of the heart's health and even the
probability of survival of the patient. Although coronary
angiography can tell us whether blood is flowing to a region
of the LV, only the analysis of the ventriculogram can tell us
whether this part of the heart muscle is contracting effectively.
By performing repeated QLVA, we also can assess the
adequacy and effectiveness of interventions such as CABG and
PTCA. This is important because, although coronary angiography
can tell us about an improvement in blood supply to the heart
muscle, it cannot tell us whether this improvement was
accompanied by improvement in cardiac function.
|
|
Left-Ventricular Measurements
We begin by tracing the outline of the LV shadow at
end-diastole (when it is the biggest) and end-systole (when it
is the smallest).
We then can use these outlines to calculate end-diastolic
volume (EDV) and end-systolic volume (ESV). (In truth,
we can't calculate the volumes - there is not enough
information in one or two X-ray projections. We actually estimate
the volumes using mathematical models and assumptions about
the symmetries of the LV cavity.)
These numbers are then plugged into a bunch of formulas to
calculate other, more interesting, parameters, such as:
Stroke Volume (SV) = EDV - ESV
Ejection Fraction (EF) = SV / EDV
Cardiac Output (CO) = SV x Heart Rate |
|
Measuring Left-Ventricular Volume
There are two fundamentally different methods of
estimating LV volume: Simpson's Rule and the Dodge-Sandler
Area-Length Method. However, both methods also have much in
common. Both depend on the accurate determination of the LV
border. This is usually done manually by an operator because
automated LV edge-detection does not work very well. Both
methods also require the determination of the long axis of the
LV. The long axis is typically defined as a line between the
mid-point of the aortic valve and the LV apex . Because the
apex can be defined as the point farthest from the mid-point
of the AO valve, the long axis can be determined
automatically.
|
|
|
Simpson's Rule Method
Simpson's Rule is a fundamental mathematical principle. It
is based on the idea that the volume of an object can be
determined by "cutting" the object into thin
"slices", measuring the volume of each slice and
summing the volumes of all slices. Simpson's Rule is applied
to the LV by slicing the LV into "discs" along the
long axis, as shown in figure A. If a single-plane angiogram
is used, each disc is assumed to be circular, since only one
diameter is known. If biplane angiography is used, each disc
is assumed to be an ellipse, with a major axis determined from
one plane and the minor axis determined from the other plane.
The area of each disc is calculated and multiplied by the
disc's thickness to determine its volume. Simpson's Rule works
pretty well for determining LV volume, but it turns out that
there is a simpler and better way. |
|
|
Dodge-Sandler
Area-Length Method
Many years ago, Drs. Harold Dodge and Harold Sandler
observed that the left ventricle looks elliptical when viewed
in the 30-degree RAO projection. To make things easy, they
made the long axis of the ellipse coincide with the long axis
of the LV. Now, if an ellipse is spun around its long
axis, it forms an ellipsoid of revolution. The
calculation of the volume of an ellipsoid of revolution only
requires knowledge of the length of the axis of
revolution and the area of the ellipse. The way Dodge and Sandler
defined things, the length of the axis of revolution was just
the length of the long axis of the LV. Easy! Determining the
area of the ellipse was a bit stickier. This is where the real
intellectual breakthrough happened. They asked "Why not
measure the area of the LV shadow and use that area for the
area of the ellipse?" The area of the LV can be measured
mechanically by tracing its border with a planimeter.
Now our two Harolds had a method, but how would they know
if it was accurate? Measuring LV volume in-vivo was not
practical. So they collected some cadaver hearts and injected
the LV chambers with a mixture of silicone rubber molding
compound and barium powder. The result was a set of
anatomically and dimensionally accurate radiopaque casts of human left
ventricles. It's pretty easy to measure the true volume of a
cast - you just plunge it into a beaker of water and see how
much water is displaced. They then X-rayed the casts and did
the area-length calculation. The results were very good! Thus,
the Area-Length method was born and soon there was a
planimeter in every cath lab. These days, planimetry is
performed with a computer, but the principle is exactly the
same.
|
|
But wait... there's more!
It turns out that both volume methods can be
improved a bit more through the use of statistics. This is
done by performing a linear regression which compares true and
measured volumes of the set of models. The regression
coefficients can then be used to correct any new volume that we calculate. Several sets of regression coefficients exist today,
each with its own set of adherents. Actually, they are all
quite good.
|
|
(Kids! Don't try this at home!)
|
Two Personal Asides
- I know about much of this because many years ago Drs.
Dodge and Sandler were nice enough to help me develop what
probably was the first computerized version of their method.
Dr. Sandler actually loaned me their original set of LV casts to use
in my validation! The casts sat in a cardboard box in my
office at Stanford for many years. Unfortunately, I didn't
appreciate their historical importance until it was too
late - at some point they disappeared.
- The LV volume system that I developed was based on the
use of a video disc recorder and light-pen. It was
sufficiently novel that it was actually featured in TV
Guide! My 15 minutes of fame came (and went) pretty early
in my career. The guy with the sideburns pushing the
catheter is Dr. Ed Alderman, my friend and co-conspirator
for most of my career. Some of you old-timers
may remember X-ray systems like this one - the gantry was stationary
and we rotated the patient!
|
|
Left-Ventricular
Wall Motion
Volume measurements give us a good idea about how well the
entire left ventricle is functioning. However, the left
ventricle is supplied by two major coronary arteries, each
with many branches. Each of the branches supplies blood to a
distinct region of left-ventricular muscle. It is very useful
to look at the performance of the LV in each of these regions
separately, so that we can estimate adequacy of the blood
supply from each of the branches. We do this by measuring how
"well" each point on the LV contour contracts during systole.
This idea raises two interesting questions:
- Given a point on the end-diastolic contour, how do we
identify the location of the same point on the
end-systolic contour?
- How do we know what is the normal amount of contraction
for any given point?
Looking at some random point on the end-diastolic LV
shadow, it is impossible to know where the point will
be on the end-systolic shadow. However, we can predict
where the point will be through the use of a model.
Many contraction models have been proposed through the
years. |
|
|
The Orthogonal Models
The earliest and simplest types of models are illustrated in
figure A. In these models, each point on the LV wall moves on
a path orthogonal (perpendicular) to the long axis of the LV. However,
although these models are appealing in their simplicity, it is easy to see that they do not work at all well near the
apex of the LV. |
|
|
The Radial Models
The next models to be developed were the so-called radial
models, illustrated in figure B. In the radial model, all
points move toward a single point somewhere in the center of
the LV. Although the exact location of the point is debatable,
radial models were a vast improvement over the orthogonal
models. The most well-known radial model is probably
the one developed at Stanford by Ed Alderman and myself. It was
based on a body of really amazing work done with implanted
myocardial markers at Stanford by Craig Miller, Neil Ingels
and George Daughters. |
|
The Centerline Model
Some might argue that all points on the LV probably do not
move toward a single point - and they would be right! To
eliminate the need for such a "magic point",
Florence Sheehan, Ed Bolson and colleagues at the University
of Washington developed the centerline model. In the
centerline model, an artificial line is constructed midway
between the end-diastolic and end-systolic contours. The model
assumes that all points on the LV wall move perpendicular to
this centerline, as illustrated in figure C. Both the Stanford
method and the Centerline method are quite accurate. |
|
The Truth About LV Wall Motion
In the early days, there was controversy over the question
of whether the LV also "rocked and rotated" as it contracted and
whether there should be compensation for this rocking by
realigning the ED and ES contours. The myocardial marker
studies have put this controversy to rest, as well
as a lot of other erroneous conceptions about LV wall motion.
In fact, we now know that LV inward contraction is only one
factor. Of more importance in normal hearts is thickening and
twisting of the muscle, making the cavity smaller. But contraction is what we still
call it, and as long as we all know what we mean, that's ok.
|
|
Copyright © 2006 (Sanders Data Systems). All rights reserved.
|
