Note: Descriptions are shown in the official language in which they were submitted.
l~Z~
04 9 6T
BONE DENS I TOMETER
The invention is an x-ray densitometer suitable
to measure bone density, or density of bone-liXe
materials, in the human body, particularly in the spine
and hip. Such measurements are useful, e.g. for
determining whether patients are affected by
osteoporosis. The invention uses a measurement
technique which is an improveMent over a related
technique called dual-photon absorptiometry.
10 Dual-photon absorptiometry is based on the use of
radioisotopic sources to provide photons of two
different energies whereas the present invention uses an
x-ray tube switched between two different voltages in
order to generate a collimated beam of two different
15 energies.
An x-ray source is capable of producing an
intensity of radiation about 1000 times greater than
conventional radioisotopic sources used for bone density
measurements. If an x-ray source were successfully
20 incorporated in a bone densitometer, an improvement in
measurement time, resolution, accuracy, precision, and
minimization of radiation dose might be effected. Prior
efforts to use x-ray sources for bone densitometers have
not been altogether successful. A ma~or purpose of the
25 present invention is to provide a succe~sful bone
densitomQter and to achieve improved performance in all
of the important categories by taking advantage of the
high radiation intensity produced by an x-ray source.
The invention achieves this objective by use of
30 an x-ray tube which is moved in a 2-dimensional raster
scanning pattern with a fixed relationship between the
tube and a collimator and detector which move with it,
1;2~ZI~
~ 2 --
with alternating high and low voltage levels being
applied to the x-ray tube.
In order to take optimal advantage of the high
photon intensity provided by x-ray sources, the
invention overcomes certain proble~s associated with
using x-ray tubes for bone densitometry. Although x-ray
sources are more intense than radioisotopic sources,
they are also less stable because they vary (drift) with
changes in the voltage and current supplied to them. In
10 addition, x-ray tubes produce photons that have a broad
range of energies whereas radioisotopic sources
typically produce photons with only a few energies.
These and other problems are met by a system
which employs two reference detectors instead of one,
15 means for providing frequent bone calibration, e.g. on
every scan line, preerably on every point, and use of
an integrating detector in the raster scan.
An important feature of the invention is a
calibration technique which determines the location of
20 bone and then calibrates the system on the basis of
x-ray data produced from non-bone areas that lie close
to the location of bone. Such calibration not only
accomodates drift of the x-ray source but also other
variations encountered, such as variation in body
25 thickness from patlent to patient.
To summarize, according to one aspect of the
invention, a bone densitometer is provided for measuring
density of bone-like material in a patient who is held
in fixed position, comprising an x-ray tube means having
30 a power supply, detector means arranged on the opposite
side of the patient to detect x-rays attenuated by the
patient, means to effectively expose portions of the
patient having bone and adjacent portions having only
flesh (non-bone body substance), means for causing the
beam to pass through a bone-like calibration material in
the course of the exposure, and signal processing means
responsive to the output of the detector means to
provide a representation of bone density of the patient
(e.g., an x-ray film-like picture of the patient,
showing bone density distribution or calculated values
representing bone density of the patient), the signal
processing means adapted to respond to data based upon
10 x-rays attenuated by the calibration material in regions
having only flesh to calibrate the output of the
detector means, thereby enabling the accommodation of
drift in the x-ray tube, differences in patient
thickness and other system variations.
According to another aspect of the invention, a
bone densitometer, e.g., having the features described
above, further comprises a pencil beam collimator
arranged to form and direct a pencil beam of x-rays
through the patient, and the detector means, on the
20 opposite side of the patient, is aligned with the
collimator, the x-ray tubes, pencil bQam collimator and
detector means adapted to be driven in unison in an X-Y
raster scan pattern relative to the patient, and
rotating means cause the beam to pass through bone-like
25 calibration mean8. According to this aspect of the
invention, the signal processing means is not limited to
responding to data based upon x-rays attenuated by the
calibration material in regions having only flesh to
calibrate the output of the detector means.
According to another aspect of the invention, a
bone densitometer incorporates features of both of the
above described aspects of this invention.
In preferred embodiments of these aspects of
the invention, the power supply of the bone densitometer
l~?Z~310
is adapted to apply alternate high and low voltage
levels to the x-ray tube; control means for the
frequency of the voltage is related to the speed at
which the x-ray tube, collimator and detector means are
S driven in scan motion and the beam width produced by the
collimator to apply alternating high and low voltage
levels to the x-ray tube at a frequency sufficiently
high that at least one pair of high and low level
exposures occurs during the short time period during
10 which the pencil beam traverses a distance equal to
about one beam width, preferably the bone densitometer
being adapted to produce pairs of high and low voltage
pulses at a rate of the order of sixty per second, the
x-ray tube, collimator and detector means being driven
15 along the scan at a rate of the order of one inch per
second and the collimator produces a pencil beam of
between about one and three millimeters in diameter; the
x-ray beam passes through the bone-like calibration
material at least onCQ per scan line of a scan pattern
20 for a period equal to at least the time during which one
pixel of resolution is traversed, preferably the x-ray
beam passes through the bone-like calibration material
for the duration of every other high and low voltage
pulse pair; the detector means comprises an integrating
25 detector controlled to integrate the detected signal
repeatedly over short time periods relative to the time
required to advance the x-ray scan pattern by one pixel
of resolution, preferably an analog to digital converter
being provided to convert each integrated value to a
30 digital signal and a dlgital computer means is provided
for producing the representation of bone density Oe the
patient by processing the stream of the digital signals;
and a reference system is provided having at least two
reference detectors each provided with a different
` - lZ~
absorber, the reference system adap~ed to correct for
both x-ray tube current and voltage changes, preferably
the system adapted to correct the detected signal
substantially on the basis of a function of the signals
produced by the reference detectors and, where there are
two of the reference detectors, the function being
substantially a straight line defined by the detected
signals of the reference detectors.
The present invention makes it possible to
10 perform bone density measurements more rapidly and with
better resolution and accuracy than prior devices.
Because it does not use radioisotopic sources, the user
does not need to handle and replace radioactive
materials which are dangerous and are strictly
lS controlled by federal licensing regulations.
In the drawings:
Figure 1 is a diagrammatic illustration of the
preferred embodiment;
Figure 2 represents a patient's spine with
20 superposed scan pattern: Figure 2a illustrates the scan
pattern employed by the preferred embodiment while
Figure 2b illustrates an alternative scan pattern,
Figure 3 is a block diagram of the electronic
control and measuring system of the peeferred embodiment;
Figure 4 is a plan view of the calibration disc
employQd in the preferred embodiment;
Figure 5 is an illustration of the voltage
levels produc~d during the high energy level ~HEL) and
low energy level ~LEL) phases of energization of the
30 x-ray tube as modified by the synchronized calibration
wheel having "bone" and "no bone" quadrants;
Figure 6 shows a plot of a function derived
from thQ '`bone" and "no bone" pulse pairs for a single
traverse of the spine; and
l~Z~3iO
-- 6 --
Figure 7 is a flow diagram of the calculations
performed by the computer for examination of a spine,
Descri~tion of the Preferred Embodiment
Figure 1 shows the basic components of the
x-ray densitometer. X-ray tube 1 carried on x-y table
arrangement 2 is energized by power supply 11 which is
designed to alternate its voltage output rapidly between
two levels called the "High Energy Level" (HEL) and the
"Low Energy Level" (LEL). The HEL is typically lSO
10 kilovolts and the LEL is typically 75 kilovolts. The
x-rays emitted by the x-ray tube are collimated to form
a pencil beam B by collimator 3. The pencil beam passes
through a calibration disc 12 which rotates at a rate
which is synchronized with the rate at which the power
lS supply ll switches between the HEL and LEL. The role
played by the calibration disc will be described below.
The x-ray pencil beam passes through, e.g., a
female adult patient 4 under examination for possible
presence of osteoporosis and impinges on a main
20 radiation detector 5. Two reference detectors 7 and 8
which are similar in design to the main detector 5 are
also shown in Figure 1. The reference detectors 7 and a
monitor the flux emitted by x-ray tube 1 and provide
information used to correct the signal measured by main
25 detector 5 for variations in the x-ray tube current and
voltage.
The reference detectors 7 and 8 measure
radiation from the x-ray tube 1 after it has traversed
one of two x-ray absorbers 9 and 10. The two absorbers
30 9 and lO are of substantially different thicknesses
which are typically chosen to be representative of the
x-ray attenuation of a thin patient and a heavy patient
respectively. By using two reference detectors with
different absorbers it is possible to monitor changes
l~Z~iO
simultaneously in both the x-ray tube current and
voltage. The reference detector measurements are used
to correct the measurements made with the main detector
S in order to compensate for these changes in tube
current and voltage. The manner in which these
corrections are made is described below.
X-ray tube 1, collimator 3, reference detectors
7 and 8 and absorbers 9 and lO, calibration disc 12, and
main detector s are all mechanically continuously
10 scanned in the X direction across the body during which
time signals from the main detector S and reference
detectors 7 and 8 are digitized and stored in computer
system 13. After each scan from right to left or left
to right in Figure l the assembly briefly stops moving
15 in the X direction, and is indexed a small amount in the
Y direction, out of the plane of Figure 1. As a result
of these motions, the pencil x-ray beam B undergoes a
rectangular scanning pattern 18 such as shown in Figures
2 and 2a. A modified rectangular scanning pattern shown
20 as parallelogram pattern l9 in Figure 2b might also be
used to measure a bone such as the neck or the femur,
which is set at an angle in the human body.
Figure l illustrates the fixed relationship
between the x-ray source l and main detector 5 during
25 the scanning period throughout which patient 4 lies
stationary on patient table 20. X-ray tube l,
collimator 3, reference detectors 7, ~ and absorbers 9
and lO, and calibration disc 12 are mounted together in
a single assembly called the source assembly 22. This
30 assembly is mounted in turn below the patient on a
conventional X-Y table 2. Separate stepping motors and
lead screws are used to move the X-Y table in the
X-direction and Y-direction respectively. The stepping
2~10
-- 8 --
motors and lead screws are of a type well known in the
art and are not shown,
The main detector 5 is mounted above the
patient and in the preferred embodiment shown is rigidly
attached by means of C-arm 21 to the source assembly 22
so that x-ray pencil beam B and main detector 5 have a
fixed relationship throughout the scan. The main
detector 5, in alternative embodiments, could be driven
with its own drive system in either the X-direction,
10 Y-direction or both so long as it maintains the same
fixed relationship to pencil beam 3.
In Figure 2, a representation of the patient's
spine 6 and adjacent portions of the body is shown. In
general, pencil beam B scans from side to side across
15 the patient's spine, and through flesh on either side of
the spine, but does not pass beyond the outer dimensions
of the adult patient 4. The total distance scanned from
side to side ~i.e. in the X-direction) is typically 5
inches and the total distance scanned from head to toe
20 (i.e. in the Y-direction) is typically 5 inches.
During the scanning period, the signals from
detector 5 and from reference detectors ~ and 8 are
digitized and stored in computer system 13. It is
possible to calculate the bone density at each point in
25 the scan pattern from these data using a method
described in more detail below. 30th the raw data and
the calculated bone density can be displayed as an image
using any one of a number of devices well known in the
art. Such an image will resemble a conventional x-ray
30 image or radiograph. In a preferred embodiment, the
computer system 13 contains one such device known as a
display processor which displays the image acquired in
this manner on a television screen. Devices such as a
l~Z~3:10
g
display processor or other computer peripherals such as
laser printers are well known devices for displaying
images from digital data.
Figure 3 is an electronic block diagram of the
bone densitometer showing the relationship between the
different components of the system. X-ray photons
striking the crystals in the scintillation detectors 5,
7, 8 generate optical radiation which is converted by
the detector photomultiplier tubes into electrical
10 cùrrents. These in turn are amplified and converted to
voltage levels by individual amplifiers 30. The
amplifier outputs are integrated by respective
integrators 32 for time periods that are controlled by
the system timing control 36 about which more will be
15 said. The output of the three integrators are digitized
by an analog-to-digitial converter 34 and stored for
processing in a small computer 13 such as an I3M PC or
AT computer system.
The system timing control 36 synchronizes the
20 x-ray power supply pulsing and the signal integrators.
For example, just before a HEL voltage is applied to the
x-eay tube, all three integrators are reset to zero. As
the HEL is applied to the x-ray tube, radiation is
emitted and all three integrators begin to integrate
25 signal8. A short time later (typically l/120 second),
the system timing control terminates the HEL voltage
level, terminating the emission of x-radiation. This is
immediately followed by a signal generated by the timing
control which terminates the integration of detector
30 signals and causes the A/D converter to digitize the
integrator output and transfer the digital value to the
computer system. A similar sequence of timing signals
is then generated for the next LEL pulse after which the
cycle is repeated.
-- 10 --
The system timing control 36, in addition to
the functions described above, also provides a means to
synchronize, via synchronizing circuit 37, the
calibration disc motor 12a to assure that the rotation
frequency of the calibration disc 12 and the x-ray
pulsing frequency are locked together so that the timing
relationships illustrated in Figure 5 are maintained.
The timing control in a preferred embodiment also
provides a pulse sequence to the stepping motor
10 controller 33 which drives X and Y direction motors 2a
and 2b and assures that every scan line in the x-ray
image has exactly the same number and phasing of x-ray
pulses.
The computer 13 provides scan distance
15 instructions to the stepping motor controller 38 and
scan initiation instructions to the system timing
control 36 and allows the operator to initiate,
manipulate, and terminate the raster scan motion and
x-ray generation by means of a standard keyboard. It
20 records the digitized detector information, calculates
bone density for each point in the raster scan pattern,
and displays the resulting image using a standard
computer display processor and television monitor.
Hardcopy vQrsions o the calculated and displayed bone
25 density can be obtained with a standard printer
interfaced to the computer.
~EAM SIZE AND SWITCHING FREQUENCY
During the scanning of the x-ray pencil beam B,
the voltage on the x-ray tube 1 is switched between the
30 HEL and LEL. A typical speed used to continuously scan
across the patient from side to side is 1 inch per
second and a typical switching frequency for the x-ray
tube power supply is 60 cycles per second. In this case
there will be 60 HEL pulses alternating with 60 LEL
iO
-- 11 --
pulses generated during each one second of scanning.
The signals from detectors 5, 7 and 8 are recorded
separately for the HEL pulses and for the LEL pulses.
For a scan speed of 1 inch per second, one HEL/LEL pair
of measurements is made for every 1/60 of an inch (0.016
inch) traversed by the pencil beam.
The cross sectional area of the pencil beam B
is determined by the opening in collimator 3 and is
typically 1-3 mm (0.040-0.120 inch) so that there are
10 typically 2 1/2 to 8 HEL/LEL pulse pairs per beam
width. One of the important features of the present
invention is that there is at least about one pulse pair
per beam width, As a result, the small region of the
body sampled by the pencil beam during the HEL
15 measurement will be essentially the same as the small
region of the body sampled by the LEL measurement made
1/120 second later.
USE OF INTEGRATING DETECTORS
The detectors used in the preferred x-ray
20 densitometer are of a type generally known as
scintillation detectors, although use of a number of
other types of detectors is also possible. A
scintillation detector consists of a crystal material
coupled to a photomultiplier tube. The crystal serves
25 to convert x-ray radiation to optical radiation and the
photomultiplier tube converts the optical radiation to
an electronic signal. Solid state photodiodes coupled
to x-ray ~luorescent screens, ionization chambers, and
other devices might also serve as radiation detectors
30 for the present invention.
In dual-photon bone densitometers using
radioisotopes, scintillation detectors are also used as
radiation detectors. However, in these devices, the
detectors must detect individual x-ray photons and sort
~2~10
- 12 -
these photons into two separate channels corresponding
to high-energy photons and low-energy photons. This
requirement for performing a spectrum analysis on
individually detected photons is dictated by the fact
that the isotopic source emits both high energy and low
energy photons simultaneously.
In the present invention, the scintillation
detector used need not perform a spectrum analysis task
by sorting photons into high and low-energy channels
10 because the high-energy and low-energy photons are not
emitted simultaneously. Rather the HEL and LEL voltages
are generated alternating in time. The high-energy and
low-energy photons are integrated separately over the
duration of the HEL and LEL pulses respectively and are
15 therefore recorded at different times.
The ability to use energy integrating detectors
rather than photon counting detectors is an important
feature of the x-ray densitometer because it makes it
possible to complete a patient scan in a short time. In
20 order to measure bone density to a given accuracy, it is
necessary to detect a resulting minimum number of
photons because the statistical accuracy of a
measurement is related, as is well known, to the square
root of the number of detected photons. For example, at
25 least 50 - 100 million photons are typically detected in
a bone density measurement of the spine.
The x-ray densitometer of the present invention
completes a measurement scan in as little as 2 to 5
minutes, In order to record as many as 100 million
30 photons in 2 mlnutes, the detector must record on the
order of 1 million photons per second. By using energy
integrating detectors rather than pulse counting
detectors, the x-ray densitometer can record photon
fluxes of 1 million per second or higher. (An energy
~2~?Z~
-- 13 --
integrating detector can easily record photon fluxes as
high as lO0 million photons per second.) The USQ of
alternating high and low voltage pulsing of the x-ray
tube, coupled with integrating detectors to measure the
high energy and low energy signals produced, is an
important feature of the present invention because it
makes short scan times possible, which implies a shorter
visit by patients and better use of capital equipment.
THE REFERENCE DETECTORS AND A~SOR~ERS
By use of two reference detectors, with
different absorbers, small changes in both the x-ray
tube current and applied voltage are effectively
monitored along with the signals from the main
detector. Just as the main detector integrates photons
15 and supplies separate values for the HEL pulse and the
LEL pulse, the reference detectors also supply separate
reference values for each HEL pulse and LEL pulse. Each
HEL or LEL measurement recorded by the main detector is
corrected according to the following method.
Let Pl and P2 be the percentàge changes in
output signal (from a HEL or ~EL pulse) measured by the
first and second reference detectors respectivQly due to
a small variation in x-ray tube current and voltage.
Let Tl and T2 be the thic~nesses of body tissue that
25 attenuate the HEL or LEL pulses the same amount as do
the first and second absorbers shown in Figure l
respectively. Each main detector signal corresponds to
a body thickneæs T0, and is corrected by a percentage P0
which i6 given by the formula ~P0-Pl) -
30~P2-Pl)/~T2-Tl)~T0-Tl). ~This can be recognized as the
formula for a straight line fit between the measured
values of Pl and P2.) This correction compensates the
main detector measurement for changes in both x-ray tube
current and voltage. This feature enables the main
l~Z~10
- 14 -
detector signal to be corrected for fluctuations in
x-ray tube current and voltage that would otherwise
degrade the accuracy of the bone density calculation,
unless the x-ray power supply is quite stable over the
duration of one patient scan. In the case that the
power supply is sufficiently stable, the use of the
reference detectors may be omitted. Longer term
variations in x-ray tube current and voitage are
compensated by the use of the calibration disc now to be
described.
THE CALIBRATION DISC
Figure 4 is a plan view of the calibration disc
12 which is mounted such that the region of the disc
near the circumference interrupts pencil beam B as the
disc rotates. The calibration disc is synchronized to
the switching frequency of the high voltage power
supply. In a preferred embodiment, the power supply
produces HEL and LEL pulses which are derived from the
main power line frequency of 60 Hertz. The HEL and LEL
pulses generated by the power supply in this embodiment
are shown in Figure 5. One pair of HEL and LEL pulses
are generated every 1/60 of a second.
~ n the preferred embodiment, the calibration
disc is driven with a synchronous motor which rotates at
a rate o exactly 30 revolutions per second and which is
ad~usted in phase such that four pre-defined quadrants
of the disc ~labeled Ql, Q2, Q3, and Q4 in Figure 4)
correspond to the HEL and LEL levels being generated by
the power supply. More specifically, when Quadrant 1 is
30 obstructing the pencil beam the voltage level is HEL,
when Quadrant 2 is obstructing the pencil beam the
voltage level is LEL, when Quadrant 3 is obstructing the
beam the voltage level is again HEL, and finally when
2~1~
-- 15 --
Quadrant 4 is obstructing the beam the voltage level is
again LEL. The desired synchronization between the
quadrants of the calibration disc and the HEL/LEL
voltage pulses is illustrated in Figure 5.
In this preferred embodiment, the circumference
of Quadrants 1 and 2 consists of a material which has
the same x-ray attenuation characteristics as bone.
Both quadrants contain exactly the same amount of the
bone-like calibration material 15 which typically
amounts to about 1 gram per square centimeter of
material. As a result, every other HEL/LEL pulse pair
recorded by the main detector is attenuated by a
constant thickness of calibration bone, as the
calibration bone rotates in and out of the x-ray beam.
Using the calibration disc in this manner, four distinct
types of measurements are periodically recorded from the
main detector. These types and their abbreviations are
1) HEL and no calibration bone ~H), 2) LEL and no
calibration bone (L); 3) HEL with calibration bone (HB),
and 4) LEL wi~h calibration bone (LB). The four groups
of measurements H, L, HB, and LB conditions are
illustrated in Figure 5.
One additional function of the calibration disc
is worth noting. It is possible to make USQ of the same
disc for the additional purpose of providing different
x-ray filtration for the HEL and LEL x-ray beams. For
example, ln a preferred embodiment, the LEL beam is left
unflltered whereas the HEL beam is filtered with 1 mm of
copper. The purpose of the copper filtration is to
attenuate the HEL beam, which is typically of
considerably higher intensity than the LEL beam because
it suffers less attenuation in tissue. By attenuating
the HEL beam, it is possible to avoid unnecessary x-ray
- 16 -
exposure to the patient and thereby lower the dose
without substantially affecting the accuracy of the
final bone measurement.
In Figure 4, in a preferred embodiment,
Quadrants 1 and 3 contain the copper filtration in the
form of a constant thickness sheet of copper 20 for the
HEL beam and Quadrants 2 and 4 contain no filtration (or
possibly another filtration material) for the LEL beam.
The use of a rotating wheel to provide different x-ray
filtrations is another advantage of the present
invention, although the main purpose of the calibration
disc is to provide continuous calibration of the
densitometer as explained below.
CALCULATION OF BONE DENSIT~
The method used to calculate bone density with
high accuracy is based on dual-photon absorptiometry
calculations as described in prior publications, (See
for examplè: "Noninvasive Bone Mineral Measurements,"
by Heinz W~ Wahner, William L, Dunn, and B. Lawrence
Riggs, Seminars ln Nuclear Medicine, Vol. XIII, No. 3,
1983.) The x-ray densitometer described here uses a
modification of the established method which makes use
of the calibeation disc bone-like material to obtain an
absolute reference for making accurate and repeatable
measurQments of the real bone. The calibration dlsc
measurements automatically and continuously calibrate
the values calculated for bone density and thereby
compensate for any short or long term drift in the x-ray
detection electronics or other system variations, as
30 well as differences in patient thickness.
Figure 6 shows schematically a plot of the
function F=ln(L)-k*ln(H) for both the "bone" and
"no-bone" pulse pairs for a single traverse across the
spine. It is important for this calibration to traverse
Z~310
- 17 -
across at least some portions of the patient having only
flesh adjacent to the spine. (Using the notation
defined above to be more precise, the calibration bone
version of F, F(B) is equal to ln(LB)-k~ln(HB) and the
no calibration bone version of F, F(NB), is equal to
ln(L)-k*ln(H).) In these formulae, the symbol, ln,
indicates the natural logarithm function and the letter,
k, is equal to the ratio of the attenuation coefficient
of tissue for the LEL pulse to the attenuation
coefficient of tissue for the HEL pulse. The values for
H, L, HB, and LB used to calculate F in these formulae
are the x~ray beam attenuation values measured by the
main detector after corrections derived from the
reference detector measurements have been applied. The
reference corrections applied in this manner are given
by the values for PO as described above.
The Wahner et al. publication cited
demonstrates that the value of the functions F(B) and
F(NB) will be a constant if there is no bone or
bone-like material (or other high atomic number
material) in the beam path through the patient. This is
strictly true if k is constant across the scan line,
independent of patient thickness, and can be made to
hold in practice by measuring any dependence of k on
patient thicknQss and using the corrected value of k to
calculate the function F. In Figure 6, the resulting
constant values obtained when the beam is on either side
of the spine ~i.e. passing through portions of the body
having flesh, without bone) are labeled Calibration
Baseline and Normal Baseline corresponding to the plots
of the functions F(B) and F(NB) respectively. The
increase in the value of the function, F, over and above
each baseline level when the x-ray beam scans across the
1;2~2~10
- 18 -
spine has been shown to be directly proportional to the
amount of bone or bone-like material in the path of the
beam.
In Figure 6, the separation value 16 between
the Normal Baseline and the Calibration ~aseline can be
calculated by finding the numerical average of the
difference between F(B) and F(NB) for measurements made
through portions of the body having only flesh on either
side of the bone. This separation value is an important
parameter in the operation of the densitometer because
it calibrates the bone measurements in the spine
directly against the known density of the bone-like
calibration material which is used in the calibration
disc measurements. Thus, the separation value carrects
for both x-ray source drift, and for variations between
different patients e.g. patient thickness, and other
system variations. In Figure 6, for example, suppose
that a particular set of H/L measurements results in an
F value (17 in Figure 6) equal to 2.46 at the spine, and
that the average separation value 16 at adjacent
portions of the body has a value of 1.23. Then the bone
density in the spine at the point where F equals 2.46
will be exactly (in this example) 2.00 times the density
of the bone-like calibratlon material used in the
calibration disc.
Using the separation value 16 as a calibration
constant, the spine bone density can be calculated for
all measuremsnt pairs H/L recorded during the entire two
dimensional raster scanning of the patient without the
need for precise mechanical positioning of the x-ray
beam or the calibration bone. The resulting two
dimensional array of bone density values can be
displayed as an image using readily available computer
peripherals.
2~1~
-- 19 --
Using the calibration disc to perform
calibration measurements is the presently preferred
method of calibrating the x-ray densitometer. Because
calibration measurements are made numerous times for
every scan line in the bone density image, one obtains
assurance of proper calibration regardless of the
location of the bone in the scan pattern. Other means
of performing calibration measurements are possible and
are comprehended by certain broader aspects of this
invention. For example, if the pencil beam were brought
to a rest at the end of each scan line or at the start
or completion of the entire scan a piece of bone-like
material could be inserted in the beam and calibration
data could be taken, Alternatively, the bone-like
calibration material could be inserted into the beam on
every alternate scan line. In another embodiment, a
small amount of calibration material is placed under a
part of the patient having only flesh without bone,
After the baselines have been determined for
each æcan line and the bone density image has been
created, it is also useful to detect automatically the
outer boundary of the spine for each scan line, From
~hese data it is possible to calculate a number of
useful parameters, For example, the integral of the
bonQ density values (expressed in units of grams per
square centimeter) between two pre-determined scan lines
~usually chosen to correspond for spine measurements to
a fixed number of vertebral bodies) is called the total
bone mineral content ~expressed in grams) and is one
often quoted parameter. The integral of bone density
values over the entire region of interest divided by the
area of the bone as determined by the outer boundaries
- 20 -
of the spine is the area averaged bone mineral density
(expressed in grams per square centimeter) and is
another frequently cited parameter.
The means for determining baseline values for
individual scan lines, the means for calculating the
boundary of the bone in the bone density image, and
other details of the bone density calculation are well
known and are not novel to the present invention. The
means for intercepting the x-ray beam with a bone-like
1~ calibration material and calculating two baselines at
portions of the body having only flesh, to calibrate the
system, has many advantages.
Figure 7 is a flow diagram illustrating the
steps performed by computer system 13 to calculate bone
lS mineral content of the spine while Fig. 7a diagrams the
steps of the calibration routine, block 3 of Fig. 7 and
Fig. 7b diagrams the routine for calculation of bone
mineral content.
For calibration, step 3, the program examines
the data and decides which points in the scan form part
of the spine and which do not. In other words, it
distinguishes "spine" points from "flesh" points. It
achieves this very simply by setting respective
thresholds for F~N) and F~NB) values and defines all
values above the threshold a9 "spine" points and all
values below the threshold as "flQsh" points.
The program then subtracts the F~NB) points
that overlay 1esh from the F~B) points that overlay
flesh to calculatQ the separation value SL shown in
Figure 6 as distance 16. The average of the separation
values averaged over all such subtracted pairs of points
is then used as a calibration constant for Step 15 of
Figure 7 to calibrate the bone mineral content, see Fig.
7b. If the separation value changes by two percent, for
~ 21~
- 21 -
example, from one day to another because of x-ray tube
drift the~ the uncalibrated bone mineral content will
change by two percent but the calibrated bone mineral
content will remain constant.
Most of the steps shown in Figure 7 are
self-explanatory but some additional explanation is
helpful. After the separation value, calculated in Step
3, is subtracted from the F values for "bone", there is
no longer any distinction between the F values for
"bone" and "no-bone" and these two sets of F values may
be merged into a single set in order to increase the
spatial resolution achieved by doubling the number of
points used to create the bone image (Steps 4 and 5).
In Step 7, the edges of the spinal vertebral
bodies are found because Fig. 7 illustrates the flow
sequence for measuring bone mineral density of the
spine. For measurements of other bones, such as the
hip, a different edge detection routine would be used.
~etween Steps 8 and 9 and between Steps 11 and 12, the
operàtor is given an opportunity to modify the
parameters calculated by the computer and to correct the
parameters based on viewing the bone mineral image
display. In Step 13 the operator dQcides which
vertebral bodies will be included in the region of
interest used to calculate bonQ mineral content and bone
mineral density,
What i8 claimed is:
