27-08-2012, 10:51 AM
A Technique to Improve the Image Quality in Computer Tomography
A Technique.pdf (Size: 452.24 KB / Downloads: 37)
Abstract
The purpose of this paper is to assess the effect of
different noise reduction filters on computed tomography (CT)
images. In particular, denoising filters based on the combination of
Gaussian and Prewitt operators and on anisotropic diffusion are
proposed. Simulation results show that the proposed techniques
increase the image quality and allow the use of a low-dose CT
protocol.
INTRODUCTION
COMPUTED tomography (CT) is a radiographic inspection
method that generates a 3-D image of the inside of
an object from a large series of 2-D images taken on a crosssectional
plane of the same object. In most clinical conditions,
CT has been necessary in adjunction to conventional radiography.
Generally speaking, conventional radiographs depict a
3-D object as a 2-D image, and theirmain limitation is that overlying
tissues are superimposed on the image. CT overcomes
this problem by scanning thin slices of the body with a narrow
X-ray beam (produced by an X-ray tube), which rotates around
the body of the stationary patient.
As it is well known in the medical world, the X-ray passing
through the patient is picked up by a row of detectors. The
tube and the detectors are positioned on opposite sides of a
ring (gantry) rotating around the patient [1]. The CT image
is derived from a large number of systematic observations at
different viewing angles; then, a 3-D image is reconstructed
from the resulting projection data by means of a processor.
Each image pixel corresponds to a CT number measured in
Hounsfield units (HU) [2], ranging in [−1000; 1000], which
represents how much of the initial X-ray beam is absorbed by
the tissues at each volume element (voxel) of the body. The
absorption coefficient varies according to the density of the
tissues.
RADIATION DOSE AND IMAGE QUALITY
CT accounts for 47% of whole medical radiation, although
it represents only 7% of total radiology examinations. Hence,
the development of techniques for reducing the radiation dose
becomes essential, particularly in pediatric applications [4].
In conventional radiography imaging, it is usually clear when
overexposure has taken place. This is not true in CT, because the
amount of radiation adsorbed by the patient depends on many
technical parameters, which can automatically be controlled
by CT scanners to balance the high image quality and the
exposure dose. Then, it is possible that the differences between
an adequate image and a high-quality image (obtained with
higher exposure) are not so immediately evident [5]. Unfortunately,
as the radiation increases, the associated risk of cancer is
increased, although this is extremely small. The potential health
risks associated with the radiation dose have motivated the
American College of Radiology to publish guidelines to ensure
that CT imaging protocols are optimized for the diagnostic
image quality at the lowest radiation dose possible [6].
CT IMAGE NOISE
CT images are intrinsically noisy, and this poses significant
challenges for image interpretation, particularly in the context
of low-dose and high-throughput data analysis.
CT noise affects the visibility of low-contrast objects. By
using well-engineered CT scanners, it is reasonable to neglect
the electronic noise caused by electronic devices [12]. Then, in
the CT image, the primary contributor to the total noise is the
quantum noise, which represents the random variation in the
attenuation coefficients of the individual tissue voxels [13]. In
fact, it is possible that two voxels of the same tissue produce
different CT values.
A possible approach to reduce the noise is the use of large
voxels, which absorb a lot of photons, assuring a more accurate
measurement of the attenuation coefficients. However, the use
of large voxels increases blurring and limits the visibility of fine
details.
Some image elaboration techniques allow one to significantly
reduce the radiation dose without compromising the
image quality. These techniques work as filters, reducing random
noise and enhancing structures. This way, it is possible
to obtain, at the same time, high-quality images and low-dose
radiation.
MATERIALS AND METHODS
In this paper, 20 high-dose chest CT images supplied by the
Radiologist staff of “G. Moscati” Taranto Hospital have been
examined. In particular, our attention was pointed to chest examinations
due to high frequency by radiologists investigating
chest pathology, as well as the good availability of this type
of images. In fact, in the chest, CT is generally better than
medical imaging analysis such as MRI for the hollow viscera.
Moreover, lung is the only organ whose vessels can be traced
without using contrast media, and this simplifies the image
elaboration.
All images (512 × 512 pixels) were in Digital Imaging
and Communications in Medicine format, which represents the
standard in radiology and cardiology imaging industry for data
exchange and image-related information. This standard groups
information into data sets, including important characteristics
such as image size and format, acquisition parameters, equipment
description, and patient information [16].
The examined images were acquired by means of a helical
CT scanner with the following acquisition setting: the tube
voltage peak is 120 kVp, the tube current is 375 mA, and the
slice thickness is 7.5 mm. Image visualization was performed
by using the standard windowing parameters for chest CT, i.e.,
windowing center of 30 HU and windowing width of 350 HU.
Each image was corrupted by additive zero-mean white
Gaussian noise to simulate a low-dose CT image. To this aim,
we have simulated the reduction in the tube current level by
adopting an amount of noise in agreement with the results
of previous studies about simulation of dose reduction in CT
examinations [18], [19].
CONCLUSION
In this paper, an analysis of denoising techniques applied
to CT images has been presented with the aim of increasing
the reliability of CT examinations obtained with low-dose
radiation.
First, the main technical parameters influencing the radiation
dose and their implications for diagnostic quality were
investigated.
Successively, the main causes of CT noise and its statistical
properties were analyzed.
Finally, some image filters to reduce the noise contribution
were proposed. In particular, a combination of Gaussian and
Prewitt filters was initially tested, obtaining a RMS of 10%.
Successively, a filtering technique based on anisotropic diffusion
was applied. Several simulations have been carried out to
choose the best filter parameters. This way, it has been possible
to decrease the relative error to about 6%.