Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Introduction

Filmless computer Xray apparatuses and even digi


tal computer Xray apparatuses are being widely intro
duced into domestic medical practice. Moreover, there
are many approaches to digital Xray imaging described
in the literature at length [8, 12].
The goal of this work was to describe a widely used
approach to digital Xray imaging: luminescent screen
optical systemcharge coupled device (CCD) matrix
amplifieranalogtodigital converter. CCD cooling
should increase the signal/noise ratio of resulting image,
thereby increasing the diagnostic capacity of the system
without increasing radiation load. The goal of this work
was to substantiate the necessity of CCD cooling, which
improves characteristics of Xray image.
Formation of Useful Signal and CCD Noise
The incident Xray quanta received during expo
sure time at input screen element constitute the useful
signal S
in
:
S
in
N

. (1)
The variance
2
S
in
of input signal or variance of the
number of quanta
2
N

in Poisson statistics is:

2
S
in

2
N

= N

. (2)
Transformation of Xray signal gives resulting digital
signals for each pixel (image element) I
m
. The statistical
characteristics of the signal are: mean value I
0
and vari
ance
2
I. Because each stage of image transformation
introduces additional noise, the relationship between
mean value and variance is:

2
I I
0
+
2
CCD
, (3)
where
2
CCD
contains both intrinsic CCD noise and noise
of amplifier and analogtodigital converter. The number
> 1 is associated with the quantum efficiency of the
device [7]. The numerical value of is determined by the
number of preliminary amplification stages and signal
transformation system.
It follows from Eq. (3) that intrinsic CCD noise sig
nificantly deteriorates image quality at low signal ampli
tude. Therefore, suppression of intrinsic CCD noise is
very important. The resulting useful signal from a CCD
matrix is additionally corrupted by the following factors:
dark charge (spontaneous generation of electronhole
pairs in accumulation and transfer modes), the tempera
ture dependence of the process being described by the fol
lowing equation:
q
T
(T) = [q

tr
(
0
) + i

acc
(
0
)t
exp
]exp[ln2(
0
)/T], (4)
where
0
is initial temperature; T is temperature change
corresponding to dark charge doubling; q

tr
(
0
) is dark
charge generated during transfer; i

acc
(
0
) is dark current
during signal accumulation. In addition, transformation
chain noise also contributes substantially to final discrete
signal noises as a function of the number of forming elec
trons. This noise or output reading noise (
2
N
re
)
1/2
is
determined by the device design.
It follows from Eq. (4) that contribution of the CCD
to useful signal variance is:

2
CCD
= q

() +
2
N
re
. (5)
The statistical characteristics of the system are deter
mined by mean value Isignal and variance of output sig
nal
2
I calculated from Eq. (3).
Biomedical Engineering, Vol. 40, No. 1, 2006, pp. 1620. Translated from Meditsinskaya Tekhnika, Vol. 40, No. 1, 2006, pp. 1519.
Original article submitted December 27, 2004.
16
00063398/06/40010016 2006 Springer Science+Business Media, Inc.
Laboratoriya Rentgenovskoi Meditsinskoi Tekhniki, Ltd., Kharkov,
Ukraine; Email: morgun@lrmt.kharkov.com
A Study of the Necessity for Cooling of ChargeCoupled Devices
in XRay Imaging Systems
O. N. Morgun, K. E. Nemchenko, and Yu. V. Rogov
Cooling of ChargeCoupled Devices in XRay Imaging Systems 17
Contrast Sensitivity as an Objective Quality Parameter of
an Imaging System
The ability of the imaging system to resolve objects of
given size and low contrast is a qualitative characteristic
of various devices. This characteristic was used in the first
models of analog video systems to assess the quality of X
ray image intensifiers [4], and it is being used for deter
mining the contrast sensitivity of digital Xray imaging
systems [13, 5, 10].
An objective method of determination of contrast
sensitivity was suggested in [6]. This method is an alter
native to the subjective method of determination of con
trast sensitivity by a group of experts. In fact, this method
is equivalent to the Rouz method [9] and can be described
as follows. Let an object of given size be in the plane of an
image. According to [11], the useful signal S is the differ
ence between mean signal I calculated from M pixels
occupied by the object
and mean signal I
0
calculated from the rest of the image:
S = I
0
I. (6)
If the human vision analyzer averages signals of M
neighboring pixels and represents signal S to the whole
object, the S distribution variance with respect to collec
tive values of Mneighboring pixels is:

2
S = (1/M)
2
I. (7)
Let the signal/noise ratio of the object be:
= SNR = S/(
2
S)
1/2
= (I
0
I)(M)
1/2
/(
2
I)
1/2
=
= K(M)
1/2
I
0
/(
2
I)
1/2
, (8)
where K = (I
0
I)/I
0
is object contrast.
It was shown in [9] that this value described the abil
ity of the human vision analyzer to detect a given object.
This value has been assessed experimentally [10] and the
oretically [9]. It follows from these estimates that the crit
ical threshold of object resolution is:
SNR(threshold)
c
5. (9)
Notation
c
was introduced in [1, 2, 5].
The value at given dose, radiation contrast, and
test object size can be calculated from Eq. (8). The device
quality can be assessed from the results of comparison of
this value with critical values
c
or similar values calcu
lated for other devices. Thus, (discernibility of thresh
old contrast) is, in fact, an objective and quantitative
comparative characteristic of Xray imaging systems.
In this work parameter is used as an objective and
quantitative comparative characteristic of Xray image
quality.
Experimental Methods
Experimental study of the dependence of contrast
sensitivity of Xray imaging system on the chargecou
pled device crystal temperature was the main goal of this
work. The test objects used in experiments were a set of
objects with variable thickness and size. The parameter
introduced in the previous section for quantitative evalu
ation of contrast sensitivity was used to eliminate the sub
jectivity factor.
Experiments were performed at anode voltage 70 kV,
different exposure doses, and two temperature condi
tions: room temperature (20C) and working temperature
mode of CCD (<0C). A 20mm Al filter was used.
Exposure dose was measured at the Xray detector site.
Aluminum disks of different size were used as objects.
The following parameters were determined: mean
signal of image I
0
, signal behind object I, and signal vari
ance
d
2
. The test object image size was M pixels. The
object resolution under various experimental conditions
was calculated from Eq. (8).
The boundary value
0
= 10 of parameter was
shown to allow the object size and shape to be clearly
resolved. The value >
0
was shown to allow both loca
tion and shape of the object to be clearly resolved even if
initial location of the object was unknown.
A decrease in is accompanied by the loss of object
shape, but its location can still be determined provided
that the object size is known. A real object is lost among
artifacts (false objects) upon approaching the limiting
value
c
= 5. In this case it can be assumed that the
object is visible in the image. However, this is true only if
exact location of the object is known. The shape of the
object is not resolved and the error in the size determina
tion is large.
Thus, the value
0
= 10 rather than
c
= 5 was
selected to be the limiting resolution value for experimen
tal data processing. Previously derived Eqs. (8), (3), and
(5) were recast as a general equation for object contrast
K = (I
0
I)/I
0
and size M at given exposure (mean value
I
0
) and CCD temperature T (factor of CCD noise

2
CCD
(T)):
18 Morgun et al.
= (I
0
I)(M)
1/2
/(
2
CCD
(T) + I
0
)
1/2
=
= Kd(/4)
1/2
I
0
/(
2
CCD
(T) + I
0
)
1/2
, (10)
where d is disk diameter.
This equation provides theoretical substantiation of
the results obtained in this work and explanation of exper
imental data.
Analysis of Experimental Results
The temperature dependence of resolution of an
object of given size was studied in the first stage of this
work. Decreasing the CCD temperature was shown to
increase the resolution of lowcontrast objects and
decrease noise. The dependence of object resolution
parameter on exposure dose at given contrast (1%) and
size (disk diameter 20 mm) is shown in Fig. 1. The object
size corresponded to 100 pixels. Experimental results are
shown as circles. The curves were calculated from Eq.
(10). Curves 1 and 2 are for the cooled and noncooled
CCD, respectively. Open and filled circles are cooled and
noncooled CCD, respectively.
Increase in the parameter upon increasing the dose
(Fig. 1) is consistent with the wellknown fact that reso
lution of lowcontrast objects increases with dose. The
difference between curves 1 and 2 corresponding to
cooled and noncooled CCD, respectively, is evidence for
the effect of cooling on resolution. For example, at dose
1.5 mR the object is poorly resolved by the thermal CCD
( = 6
c
), but well resolved by the cooled CCD (
14 >
0
).
The effect of cooling on the dependence of resolu
tion of an object on object size and contrast at constant
dose was studied in the second stage of this work. The
dependence of resolution of object on the product of
contrast multiplied by diameter at constant dose 1 mR is
shown in Fig. 2. Curves 1 and 2 are for cooled and non
cooled CCD, respectively. Open and filled circles are for
cooled and noncooled CCD, respectively. It follows
from Fig. 2 that CCD cooling increases the resolution
parameter and changes the slope of the curve.
Therefore, CCD cooling allows not only lowcontrast
and/or size objects to be resolved, but also objects of sim
ilar contrast to be distinguished from each other. In fact,
this property increases the dynamic range of the device.
The third stage of this work was devoted to verifica
tion of object resolution improvement upon CCD cooling
in the case of equal exposure, i.e., wellresolved object
size and/or contrast decrease with the cooled CCD with
Fig. 1. Dependence of object resolution parameter on exposure
dose at given contrast (1%) and size (disk diameter 20 mm).
Experimental results are shown as circles. The curves were calcu
lated from Eq. (10). Here and in Figs. 24 curves 1 and 2 are for
cooled and noncooled CCD, respectively. Open and filled cir
cles are for cooled and noncooled CCD, respectively.
Dose, mR
Fig. 2. Dependence of resolution of object on product of con
trast multiplied by diameter at constant dose 1 mR.
Kd, %pixel
Cooling of ChargeCoupled Devices in XRay Imaging Systems 19
respect to the noncooled CCD. The dependence of con
trast of wellresolved object on object size at constant
dose is shown in Fig. 3. Curves 1 and 2 are for cooled and
noncooled CCD, respectively.
It follows from Fig. 3 that at given contrast objects of
smaller size are resolved by the cooled CCD matrix,
because the given value of K for the cooled CCD (curve 1)
corresponds to the value of d 50% smaller than for the
noncooled CCD (curve 2). Similarly, at given object size
the objects of smaller contrast are resolved better.
The fourth stage was most important for reduction of
exposure dose at constant object resolution with the
cooled CCD. The question was: to what extent the expo
sure dose could be reduced by CCD cooling without
deterioration of resolution of object of given size and con
trast?
The dependence of dose required for clearly resolv
ing the object on its contrast at given object size is shown
in Fig. 4. It follows from Fig. 4 that the same resolution of
object with given factor Kd with the noncooled CCD
(curve 2) is observed at dose level three times higher than
with the cooled CCD (curve 1). Therefore, reliable reso
lution using the noncooled CCD requires three times
larger dose than when using the cooled CCD.
It should also be noted that threshold contrast
improvement induced by CCD cooling was attained with out deterioration of other characteristics, as was observed
in case of objectives with high lightgathering power or
CCD with larger pixels, which deteriorate the frequen
cycontrast characteristic of the device.
Conclusion
Experimental and theoretical studies were obtained
to substantiate the necessity of cooling of chargecou
pled devices for improvement of Xray image character
istics.
Theoretical analysis of signal processing from the X
ray quanta to digital computer image gave Eq. (10) for the
quantitative parameter of Xray imaging quality.
Satisfactory fit of theoretical data to experimental results
allows Eq. (10) to be used in further construction of new
imaging systems.
The theoretical and experimental results revealed
that objects of certain contrast and size unresolvable using
a noncooled CCD could be resolved at the same dose
using a cooled CCD.
The experimental results obtained in this work
showed that CCD matrix cooling decreased radiation
load on patient without deterioration of diagnostic capac
ity. The CCD matrix cooling also improves the system
parameters observed at fixed radiation exposure dose.
Fig. 3. Dependence of contrast of object on object size at = 10
and constant dose 1 mR.
K, %
d, pixel
Fig. 4. Dependence of dose required for clearly resolving the
object ( = 10) on its contrast at given object size.
Kd, %pixel
Dose, mR
20 Morgun et al.
The results reveal the necessity of cooling of CCD
matrix crystals in the Xray imaging system under consid
eration.
REFERENCES
1. N. N. Blinov and A. I. Mazurov, Med. Tekh., No. 5, 310 (1999).
2. N. N. Blinov, E. B. Kozlovskii, and S. I. Luzin, Med. Tekh., No. 5,
2426 (1999).
3. L. V. Vladimirov, Med. Radiol., 26, No. 6, 4448 (1981).
4. GOST 2614184. XRay Image Intensifiers of Medical XRay
Apparatuses. General Technical Requirements and Methods of
Testing [in Russian], Moscow (1984).
5. B. M. Katner, Med. Tekh., No. 5, 1013 (1999).
6. O. N. Morgun, K. E. Nemchenko, and Yu. V. Rogov, Med. Tekh.,
No. 5, 1921 (2003).
7. O. N. Morgun, K. E. Nemchenko, and Yu. V. Rogov, Med. Tekh.,
No. 5, 69 (2003).
8. O. I. Nedavnii and V. A. Udod, Defektoskopiya, No. 8, 6282 (2001).
9. A. Rose, Vision: Human and Electronic [Russian translation],
Moscow (1977).
10. P. A. Connoly, In: Quality Assurance and Patient Radiation
Protection in Diagnostic Radiology, Berlin (1996).
11. H. Gfirtner, In: Quality Assurance and Patient Radiation
Protection in Diagnostic Radiology, Berlin (1996).
12. B. Hemdal, I. Andersson, A. Thilander, et al., In: Recent
Technical Developments and Their Clinical Potential (2002), pp.
20022008.

You might also like