Ultrafast CT was designed from the beginning as a "cardiac
scanner" using 50 msec,
polytomographic imaging for cardiac chamber and myocardial anatomy, function and
flow. There are probably over 100
peer reviewed papers supporting these "non-coronary" applications.
Several years following its first commercial introduction, the EBCT scanner was retrofitted with a collimator and set up to perform thin section,
higher spatial resolution imaging at 100 msec.
By improving the spatial resolution, but maintaining the ECG triggering
ability, excellent "stop action" images could also be acquired of the
proximal epicardial coronary arteries. Current
EBCT scanners have high spatial resolution [roughly 10 line pairs per cm.] to
facilitate the full range of cardiac functional assessment, coronary artery
imaging, as well as general body CT. EBCT functions as a true "body scanner" with motion
free imaging of the heart as well as "spiral" type applications for
scanning of the chest, abdomen, and head. The
common statement that EBCT can be used only for "coronary calcium
scanning" is an unfortunate misnomer spread largely by individuals
peripherally involved with the technology and reflects ignorance rather than
true experience. For instance at
the Mayo Clinic, Rochester, there are 3 EBCT scanners, working 12 hours per day,
performing coronary and vascular work [aorta, pulmonary arteries] as well as
conventional high resolution body CT [abdomen, pelvis, chest].
Mechanical scanners have been in use for decades and the introduction of
helical [also called "spiral" CT] by Dr. Willy Calendar [Siemens,
Erlangen, Germany] represented a major improvement in imaging by allowing the
slip ring technology to facilitate continuous scanning.
By moving the table at a particular "rate" during continuous
motion of the mechanical source/detector pair, excellent images can be obtained
of the body with variable slice thicknesses.
Major recent improvements have reduced but not eliminated the problems of
"heat loading" which originally significantly limited the number of
slices acquired in a single session. Helical
CT [HCT] imaging can then be done of large sections of the body in seconds
and/or during a single breathhold.
HCT scanners, however, still require the physical movement of an x-ray
source [about a stationary detector system] and this takes about 1 second.
So called "subsecond" imaging using HCT is actually imaging
during a partial scan. Physically, you need image only about 180 degrees [or a bit
more] of the full circle and invoke an assumption of symmetry to reduce
"effective" imaging times to as low as 500-750 msec for some scanners.
Recently Siemens and soon, General Electric, will introduce multiple
parallel detectors, such that multiple images can be obtained with a single
sweep of the x-ray source or sections of adjacent detectors can be processed to
reduce the "effective" scan time to possibly as low as 250 msec.
The ballistics of cardiac motion during the heart cycle are complicated
and three dimensional. I use the
pneumonic TARTTS for characterization
of cardiac motion. The heart Translates [physically moves across the tomographic plane]; Accordions
[with a piston like motion involving descent of the base of the heart and ascent
of the apex]; Rotates [about an
ill-defined vertical axis - which is likely curved and not linear]; Tilts [about another axis which wobbles again about the ill defined
vertical axis, sort of like the Earth tilting with the changes of the seasons]; Thickens
[with increase in left ventricular wall thicknesses, but with no net change in
muscle mass - or total intrapericardial volume]; and finally, Squeezes
with a wringing motion. Many of
these motions are not independent. The
base of the heart descends about 1 cm with each cardiac cycle and the apex
ascends about 1 cm with each cardiac cycle.
The coronary artery is anywhere from 3-4 mm diameter at the left main to
about 1 mm at the distal LAD. Studies
have shown that the coronary arteries move at a rate on average of 46 mm/sec;
furthermore, the total excursion of the artery during the cardiac cycle is a
distance at least equivalent to its own diameter..
Cardiac motion accelerates during early systole, rapidly decelerates at
the end of systole, accelerates a bit slower [and in the opposite direction]
during early diastole, and then accelerates [to return to its original position
and a bit slower] during late diastole. Generally,
the slowest rates of acceleration/deceleration of cardiac motion are found at or
about the mid to later phases of diastole.
Scanning using EBCT at 100 msec at a fixed tomographic plane during each
acquisition and triggered at the mid to late phase of diastole not uncommonly
produces some "motion artifacts", especially anatomically where the
ballistics of cardiac motion are most apparent [base of the heart, right
coronary artery]. Imaging at 750
msec takes roughly three-quarters of the cardiac cycle at an average heart rate
of 60/min. During this time there
is considerable cardiac motion with acceleration and deceleration during the
imaging. Spiral CT imaging, since
it is done concomitant with table incrementation actually produces an
"average" reconstructed tomographic image based upon multiple rays
traversing the scanning area during table movement. Imaging at speeds > 100 msec, especially with table
incrementation during imaging, introduces errors in coronary artery imaging such
as "blooming" and other artifacts of the arterial width due to
compounded motion occurring in and out of the tomographic plane. It also limits the necessary three-dimensional registration
of all of the images throughout the scanning volume. The validity of the HCT approach to quantification of
coronary artery calcium, or for coronary artery imaging in general, is thus
drawn into question, based solely upon the physics of imaging and the physiology
of cardiac motion.
Schemesh
and colleagues in Israel have published several papers on coronary artery
calcium scanning using the Elscint dual helix scanner.
Using a calcium threshold of 90 HU [rather than the one which has been
proved to correlated with ashed coronary artery calcium mass - 130 HU], they
determined in a group of patients who also had coronary arteriography a negative
predictive value [i.e. the absence of calcium] of 59% for advanced coronary
artery disease. The published
negative predictive value for EBCT is around 95% or greater.
By reducing the threshold to 90 HU, they attempted to make up for the
motion and reduced sensitivity of the slower mechanical scanner - however, they
likely increased the noise and artifact potential, thus the lowered negative
value. The studies by Schemesh were
well done, did show data which at least is consistent with the EBCT data, but has
not been reproduced to date by any other laboratory.
There are now over 250 publications on EBCT and coronary calcium with
independent confirmation of data pathologically and angiographically from
multiple laboratories in several countries.
There are data which have been presented by colleagues using a Siemens
HCT scanner at 500 msec scan speed [with or without "triggering"]
showing a very good correlation between calcium scores with EBCT [see AHA
abstracts, 1998]. The calcium
scores ranged from zero to several thousand. However, when the data are examined for scores <100, there
is wide scatter. It is a
statistical fact that a correlation across a very broad range can be
significant, but the spread about the line of correlation can be quite wide, as
is the case with these data. Recently
a set of data was reported by Carr who used a GE scanner in 36 patients and
reported a similar "high correlation" to EBCT with a broad range of [Agatston]
calcium scores. However,
correlation and precision are two different matters.
There have been actually very few data published on comparisons between EBCT
and spiral CT and there are data coming forward which demonstrate the
limitations of scanning at slow scan speeds.
Budoff has recently reported [AHA 11/99] that direct comparison between
HCT and EBCT scanning showed significant variation when the calcium scores were
low to intermediate in magnitude.
An example of imaging a moving object using different "shutter
speeds" is shown below in Figures 3 and 4.
This "toy" was moved the equivalent of an average coronary
artery diameter (2.5 mm) in one second (25 mm/sec).
Achenbach has actually noted that the ballistic movements of each
coronary artery during the cardiac cycle are not the same.
The RCA moves at a rate of 65 mm/sec, while the LAD moves only at 27
mm/sec. The LCX motion is between
these at 42 mm/sec. ). These data
would suggest that the amount of blurring during cardiac motion may in some
instances be even more apparent using a spiral CT approach than noted in Figures
3 and 4. Direct imaging of a
standard "resolution" phantoms during simulated cardiac motion for EBCT
[100 msec per image] and the General Electric Lightspeed
[500 msec per image] is shown on Figure 5.
As can be seen, when the speed of cardiac motion exceeded 20 mm/sec there
was true "aliasing" of imaging in a manner analogous to the Nyquist
sampling limit for pulsed Doppler velocimetry.
Based upon the currently available data and data under development at a
variety of laboratories, imaging using spiral CT for coronary artery calcium has
little data to support its clinical utilization and its use for quantification
of calcium volume as well as a means to follow disease progression is currently
questionable. and most definitely misinforming the public who believe that their
doctors are using the best methods to perform individual diagnostic studies.
In conclusion, the claims by the spiral CT makers that HCT is "just
as good" as EBCT are unfounded and unsubstantiated at present.
Many users of HCT do not tell their doctors or their patients that they
are using a technology different from EBCT.
At present HCT is NOT an ultrafast CT, and the use of "subsecond"
scanning to imply that this is the same as rapid scanning using an electron beam
is misleading, and in my opinion, dishonest.
Proposing imaging using HCT for quantification of coronary calcium
should, at best, be considered experimental at the present time, and not
currently ready for clinical practice. One
or two papers using different types of HCT with several abstracts and other
limited data [despite the promise of more to come] is insufficient compared to
the scientific and published approaches done by myself and colleagues regarding
investigations of EBCT from the pathology and clinical laboratories. The proposition that HCT and EBCT are equivalent is "less
than honest". More importantly
it is misdirecting the radiologists and/or cardiologists performing these
analyses [who have no formal training from the handful of us that are qualified
to instruct], confusing the referring physicians, and not providing the patient
with the best available technology to assess their potential for heart disease.
