Application of analysis approach in Noise Estimation in Panoramic X- rays images using image...

Canadian Journal on Medicine Vol. 2, No. 3, July 2011

38

Application of analysis approach in Noise Estimation in Panoramic X-

rays images using image processing program (MatLab) Yousif Mohamed Yousif Abdallah*, Arwa Ahamed Almoustafa, Gehan Elhadi, Mamdouh Mohammed, Osman Khalafallah

& Tayseer Khalid

College of Medical Radiologic Science, Sudan University of Science and Technology, Khartoum, Sudan

Tel: +249911628004,+249126641065, Email [email protected]

Abstract — This research presented an appropriate

approach for the robust estimation of noise statistic in dental

panoramic x-rays images. To achieve maximum image quality

after denoising, a new, low order, local adaptive Gaussian

Scale Mixture model is presented, which accomplishes

nonlinearities from scattering. State of art methods use multi

scale filtering of images to reduce the irrelevant part of

information, based on generic estimation of noise. The usual

assumption of a distribution of Gaussian and Poisson

statistics only lead to overestimation of the noise variance in

regions of low intensity (small photon counts), but to

underestimation in regions of high intensity and therefore to

non-optional results. The analysis approach is tested on 50

samples from a database of 50 panoramic X-rays images and

the results are cross validated by medical experts.

In this thesis, prominent constraints are firstly preservation

of image's overall look; secondly preservation of the

diagnostic content in the image and thirdly detection of small

low contrast details in diagnostic content of the image. As

shown in previously, state of the art methods provide non-

convincing results. The new approach is funded on an attempt

to interpret the problem from the view of blind source

separation (BSS), thus to see the panoramic image as a simple

mixture of (unwanted) background information, diagnostic

information and noise.

Key Words — image quality, Noise, blind source separation,

matlab

I. INTRODUCTION

The “classic” 18-film intraoral radiograph examination of

the teeth is relatively comfortable for the patient, and the

overall radiation dose is generally acceptable; however, it is a

time-consuming endeavor, and in some cases provides a less-

than-optimum depiction of individual teeth. Nevertheless, it

offers valuable information as a basis for treatment planning, is

far superior to purely clinical observations, and often gives

impetus for more extensive examination strategies. Finally, the

18-film“full series” exam provides a sense of security for both

dentist and patient, and enhances confidence through

communication. On the other hand, clear radiographic

depiction of not only the teeth, but also the jaws, the

temporomandibular joints, and the alveolar lobes of the

maxillary sinuses will reduce the risk of an incomplete and

possibly incorrect examination, which in the worst-case

scenario could lead to malpractice. In this regard, the

panoramic radiograph always leads to a broadening of

horizons because it improves the dentist’s knowledge of

radiographic anatomy and thus improves her/his skill in

distinguishing between and among normal and pathologic

conditions. This, in conjunction with a better understanding of

the interrelationships of systemic medical problems and

dental/oral problems can open new avenues for treatment

planning (1)

.

The question of whether the treating dentist her/himself

actually takes the radiographs or delegates this task to

auxiliary personnel is really a question of responsibility for the

radiographic quality that is achieved, and the radiation dose

that this requires. If, for whatever reason, this assignment is

delegated, the person who bears the primary responsibility

must see to it that the auxiliary personnel are well-trained and

have received legal certification (2)

.

They must not only become expert in the production of

high-quality radiographs, but must also be knowledgeable

regarding the dental indications and the procedures for

protecting patients and staff from excessive radiation

exposure. Auxiliary dental personnel who are given the

responsibility of taking radiographs must remain current in all

continuing education standards in order to insure high

radiographic quality and the lowest possible radiation dose for

every patient. This is critical. An image defined as it may be

applied to a picture such as a photograph, a painting or a

sketch which has a real physical existence. But it may be also

being applied to an idea or concept which has a mental rather

than physical existence. If asked to imagine an object such as

apple, a mental image of an apple comes to mind. The

existence of this visual image helps us to grasp the concept of

an apple. Of course, a photograph of an apple represents only

one aspect of an apple: its visual appearance. It provides no

representation of the taste, smell or feel of the characteristics

of an apple which we remember from past experiences (3)

.

There are two components are superimposed in the image.

The presence of noise limits the amount of information which

can be extended from image. In particular, the finer details of

structure may be lost by being swamped by the effect of noise.


39

Noise may see an unlikely term apply to a visual image which

is silent. The use of the term probably originates from radio

engineering, where the quality reception of radio signals is

often impaired by background noise (i.e. hisses and whistles),

particularly when the origin signal is rather weak and the radio

receiver is being used with maximum amplification. The effect

of such noise on the information communicated via a radio

link is very similar to the effect of noise on the information

carried by an image. Under optimum conditions the magnitude

of the signal is very much greater than the magnitude of the

noise. The signal-to-noise ratio is said to be high. Under

adverse conditions the signal-to-noise ratio is law, and much

information is lost. It is not uncommon for a radiographic

image to be fogged, perhaps due to an accidental exposure to

scattered radiation in X-ray room, or to poor conditions of

storage. When such a radiograph is examined its image

appears as viewed through a mist or fog. The information

content of the image is reduced. Some information is lost

entirely, whilst that which remains is more difficult to see. The

fogging contains no information about the subject of

radiograph. Such fogging is an example of one form of image

noise. Another type of noise may be seen on a television

image. In this case, the image appears as if viewed through a

snowstorm. The screen display a very large number of small

white specks which appear superimposed on the image. Again,

the effect is to reduce the information content of the image.

Such as appearance may be due to inferior design of the

electronic circuitry employed, causing electronic noise.

However, in many cases the basic is that the signal being

displayed is too weak, and those speaks represent the absence

of information, rather like the spaces left by missing pieces in

a jigsaw puzzle. It is possible for a similar effect to occur in

the image on a radiograph, expect that in this case the image

appears to have a large number of very small black specks

superimposed on it, giving it a grainy or mottled appearance

commonly known as quantum mottle or quantum noise. To be

able to identify a feature on an image, it must appear different

its surroundings. On a radiograph, a structure must be of a

different optical density (shade of grey) from adjacent

structures. On television image, the structure must be of a

different luminance (brightness). The term contrast is used to

describe these differences are small; it is difficult to identify

the structure will stand out well from its surroundings and it is

said the contrast is high (good). If the differences are small, it

is difficult to identify the structure against its background. The

contrast is said to be low (poor) (3).

II. QUANTITATIVE IMAGE ANALYSIS

This is the fastest growing application area of medical

workstations. The number of applications is vast, every major

vendor has research activities in this area, and many research

institutes are active. To quote from the scope of ‘Medical

Image Analysis’, one of the most influential scientific journals

in the field: “The journal is interested in approaches that utilize

biomedical image datasets at all spatial scales, ranging from

molecular/cellular imaging to tissue / organ imaging. While

not limited to these alone, the typical biomedical image

datasets of interest include those acquired from: magnetic

resonance, ultrasound, computed tomography, nuclear

medicine, X-ray, optical and confocal microscopy, video and

range data images. The types of papers accepted include those

that cover the development and implementation of algorithms

and strategies based on the use of various models (geometrical,

statistical, physical, functional, etc.) to solve the following

types of problems, using biomedical image datasets:

representation of pictorial data, visualization, feature

extraction, segmentation, inter-study and inter-subject

registration, longitudinal / temporal studies, image-guided

surgery and intervention, texture, shape and motion

measurements, spectral analysis, digital anatomical atlases,

statistical shape analysis, computational anatomy (modeling

normal anatomy and its variations), computational physiology

(modeling organs and living systems for image analysis,

simulation and training), virtual and augmented reality for

therapy planning and guidance, telemedicine with medical

images, telepresence in medicine, telesurgery and image-

guided medical robots, etc.” (4)

Segmentation is a basic necessity for many subsequent

viewing or analysis applications. Mostly thresholding and

2D/3D region growing are applied, but these often do not give

the required result. Proper segmentation is notoriously

difficult. There are dozens of techniques, such as model-based

segmentation, methods based on statistical shape variations

(‘active shape models’), clustering methods in a high-

dimensional feature space (e.g. for textures), histogram-based

methods, physical models of contours (‘snakes’, level sets),

region-growing methods, graph partitioning methods, and

multiscale segmentation. The current feeling is that fully

automated segmentation is a long way off, and a mix should be

made between some (limited, initial) user-interaction (semi-

automatic segmentation). Feature detection is the finding of

specific landmarks in the image, such as edges, corners, T-

junctions, highest curvature points, etc. The most often used

mathematical technique is multi-scale differential geometry (5)

.

It is interesting that the early stages of the human visual

perception system seem to employ this strategy. Image

enhancement is done by calculating specific properties which

then stand out relative to the (often noisy) background.

Examples are the likeness of voxels to a cylindrical structure

by curvature relations (‘vesselness’), edge preserving

smoothing, coherence enhancing, tensor voting, etc.

Commercial applications are, e.g. MUSICA (‘Multi-Scale

Image Contrast Amplification’, by Agfa), and the Swedish

Context Vision Enhancement is often used to cancel the noise

increasing effects of substantially lowering the X-ray dose,

such as in fluoroscopy and CT screening for virtual

colonoscopy. Quantitative MRI is possible by calculating the

real T1 and T2 figures from the T1 and T2 weighted


40

acquisitions, using the Bloch equation of MRI physics. Multi-

modal MRI scans can be exploited for tissue classification:

when different MRI techniques are applied to the same

volume, each voxel is measured with a different physical

property, and a feature space can be constructed with the

physical units along the dimensional axes: e.g. in the

characterization of tissue types in atherosclerotic lesions with

T1, T2 and proton density weighted acquisitions, fat pixels

tend to cluster, as do blood voxels, muscle voxels, calcified

voxels, etc (Fig. 1).

Fig.1 . A Multimodality MRI of atherosclerotic plaque in the human

carotid artery: (w1) T1-weighted 2D TSE, (w2) ECG gated proton

density-weighted TSE, (w3) T1-weighted 3D TFE, (w4) ECG-gated

partial T2-weighted TSE, (w5) ECG-gated T2-weighted TSE. b Feature

space for cluster analysis. c Classification result. From Hofman et al.

(2006)

Pattern recognition techniques like neural networks and

Bayesian statistics may find the proper cluster boundaries.

Shape can be measured with differential geometric properties,

such as curvature, saddle points, etc. It is often applied when,

e.g. in the automated search for (almost) occluded lung vessels

in pulmonary emboli, polyps on the colon vessel wall,

measuring the stenotic index, spiculated lesions in

mammography, etc. A popular method is based on ‘active

shape models’, where the shape variation is established as so-

called shape eigenmodes by analyzing a large set of variable

shapes and performing a ‘principal component analysis’, a well

known mathematical technique. The first eigenmode gives the

main variation, the second the one but largest, etc. Fitting an

atlas or model-based shape on a patient’s organ or segmented

structure becomes by this means much more computationally

efficient. Temporal analysis is used for bolus tracking (time-

density quantification), functional maps of local perfusion

parameters (of heart and brain), contrast- enhanced MRI of the

breast, cardiac output calculations by measuring the volume of

the left ventricle over time, multiple sclerosis lesion growth

/shrinkage over time, regional cardiac wall thickness variations

and local stress/strain calculations, and in fluoroscopy, e.g. the

freezing of the stent in the video by cancellation of the motion

of the coronary vessel (7)

.

III. DENTAL RADIOGRAPHY:

1. Intraoral radiography:

Bitewing radiography is a lateral view to demonstrates the

crowns of the teeth and both the maxilla and mandible.

Periapical radiography is a lateral projection displaying both

the crown and he root of the tooth and the surrounding bone.

Occlusal radiography comprises an umber of views in which

the film is positioned in the occlusal plane.

2. Extra – oral radiography:

Dental panoramic radiography: produce an image of both

jaws. Oblique lateral radiography demonstrates large area of

the maxilla and mandible. Cephalometry used to demonstrate

the facial bone for use in orthodontic, orthogenetic and implant

treatment. There are a number of technologies used for digital

imaging in planer radiography. They can be divided into CR

and DDR. CR is, in first appearance, similar to use of a

film\screen system. The CR plate is in cassette, which will fit

the table and vertical Bucky trays and can be used with mobile

equipment. The plate is then scanned in a reading system

similar in size to a daylight processor. This therefore makes

the change to digital radiography easier. A DDR system entail

more changes in x-ray couch and vertical Bucky design and

often changes to the x-ray tube assembly. Unlike the

removable CR cassette, the DR plate or detector is fully

integrated into the exposure equipment. The patient is radio

graphed and the image appears on the acquisition workstation

in a few seconds. The image can be optimized and then sent

for reporting or repeated if necessary.

IV. PRELIMINARIES AND DEFINITIONS OF IMAGE

A digital image is defined as a tow-dimensional array of

numbers that represents the real, continuous spatial signal is

sampled at regular intervals and the intensity is quantized to a

finite number of the array is referred to as a picture element or

pixel. The digital image is defined as a spatially distributed

intensity signal f(m ,n),where f is the intensity of the pixel, and

m and define the position of the pixel, along a pair of

orthogonal axes usually defined as horizontal and vertical. We

shall assume that the image has M rows and N columns and

that the digital image has P quantized levels of intensity (gray

levels) with values ranging from 0 to P-1.

The histogram of an image, commonly used in image

enhancement and image characterization, is defined as a vector

that contains the count of the number of pixels in the image at

each gray level. A useful image enhancement operation is

convolution using local operators, also known as Kernels.

Considering a Kernel w (k, l) to be an array of (2k+1+2+1)

coefficients where the point (k, l) = (0,0) is the center of the

Kernel, convolution of the image with the Kernel is defined

by:


41

G(m, n) = w(k, l) * f(m, n) (1)

Where G(m, n) is the out come of the convolution or out put

image. To convolute an image with a kernel, the kernel is

centered on an image pixel (m, n), the point-by-point products

of the kernel coefficients and corresponding image pixels are

obtained, and the subsequent summation of these products is

used as the pixel value of the out put image g (m, n) is

obtained by operating the same operation on an pixels of

original image. A convolution kernel can be applied to an

image in order to effect of specific enhancement operation or

change in the image characteristics. This typically results in

desirable attributes being amplified an d undesirable attributes

being suppressed. The specific values of the kernel coefficients

depend on the different types of enhancement that may be

desired. Attention is needed at the boundaries of the image

where parts of the kernel extend beyond the input image. One

approach is to simply use the portion of the kernel that

overlaps the input image. This approach can, however, lead to

artifacts in the boundaries of the out put image. The Fourier

transform F (u, v) of an image f(m, n) is defined as Where u

and v are the spatial frequency parameters. The Fourier

transform provides the spectral representations of an image,

which can be modified to enhance desired properties. The

forward or inverse Fourier transform of an N×N image,

computed directly with the preceding definitions, requires a

number of complex multiplications and additions proportional

to N2. By decomposing the expressions and eliminating

redundancies, the fast Fourier transform (FFT) algorithm

reduces the number of operations to the order of N

NLog2N (2)

The computational advantage of the FFT is significant and

increases with increasing N. when N =64 the number of

operation are reduced by an order of magnitude and when N =

1024, by two orders of magnitudes (3)

.

V. PIXEL OPERATIONS:

In this section we present methods of image enhancement that

depend only up on the pixel gray level and don’t take in to

account the pixel neighborhood or whole image

characteristics. Compensation for Nonlinear Characteristics of

Display or print Media Digital images are generally displayed

on cathode ray tube (CRT) type display systems or printed

using some type of photographic emulsion. Most display

mechanisms have non-linear intensity characteristics that result

in a non-linear intensity profile of the image when it is

observed on the display. This effect can be described

succinctly by the equation

e(m, n) = c(m, n) (3) where f(m, n) is the acquired intensity image, e(m, n)

represents the actual intensity output by the display system, in

order to correct for the nonlinear characteristics of the display,

a transform that is the inverse of the displays nonlinearity must

be applied.

G(m, n)= T(e(m, n))≈ C-1

(c(m, n)) (4)

G(m, n) ≈ f(m, n)

Where T () is a nonlinear operator which is approximately

equal to C-1

(), the inverse of the display system operator, and

G (m, n) is the output image. Determination of the

characteristics of the nonlinearity could be difficult in practice.

In general, if a linear intensity wedge is imaged, one can

obtain a test image that captures the complete intensity scale of

the image acquisition system. However, an intensity

measurement device that is linear is then required to assess the

output of the display system, in order to determine its actual

nonlinear characteristics. Intensity scaling is the method of

image enhancement that can be used when the dynamic range

of the acquired image data significantly exceeds the

characteristics of the display system, or vice versa. It may also

be the case that image information is present in specific narrow

intensity bands that may be of special interest to the observer.

Intensity scaling allows the observer to focus on specific

intensity bands in the image by modifying the image such that

the intensity band of interest spans the dynamic range of the

display.

For example, if f1 and f2 are known to define the intensity

band of interest, a scaling transformation may be defined as

e= (5)

. (f max)

Where e is an intermediate image, g is the output image, and f

max is maximum intensity of the display. These operation may

be seen through the images present an image with details in the

intensity band from 90 to 170 that may be of interest to, for

example a gum specialist. The image, however, is displayed

such that all gray level in the range 0 to 255 is seen (3)

.

Although intensity scaling can be very effective in enhancing

image information present in specific intensity bands, often

information is not available a priori to identify the use full

intensity bands. In such cases, it may be more use full to

maximize the information conveyed from the image to the user

by distributing the intensity information in the image as

uniformly as possible over the available intensity band. This

approach is based on the approximate realization of


42

information – theoretic approach in which the normalized

histogram of the image is interpreted as the probability density

function of the intensity of the image. In histogram

equalization, the histogram of the input image is mapped to a

new maximally- flat histogram. The histogram is defined as

h(i), with 0 to p-1 gray levels in the image. The total number

of pixels in the image, M*N, is also the sum of all the values in

h(i). Thus, in order to distribute most uniformly the intensity

profile of the image, each bin of the histogram should have a

pixel count of (M*N)/p. It is, in general, possible to move the

pixel with a given intensity to another one, resulting in an

increase in the pixel count in the new intensity bin. On the

other hand, there is no acceptable way to reduce or divide the

pixel count at a specific intensity in order to reduce the pixel

count to the desired (M*N) /P. In order to achieve

approximate uniformity, the average value of the pixel count

over a number of pixel values can be made close to the

uniform level.

A simple and readily available procedure for distribution

of the pixels in the image is based in the normalized

cumulative histogram, defined as

H (j) = 1/M. N , j=0,1,… p – 1. (6)

The normalized cumulative histogram can be used as a

mapping between the original gray levels in the image and the

new gray levels required for enhancement. The enhanced

image g(m, n) will have a maximally uniform histogram if it is

defined as

G (m, n)= (p-1). H (f(m, n)) (7)

Local operators enhance the image by providing a new value

for each pixel in a manner that depends only on those pixels

and others in neighborhood around it. Many local operators

are linear spatial filters implemented with the kernel

convolution, some are non linear operators, and other imparts

histogram equalization with in a neighborhood. In this section

we present a set of established standard filters commonly used

for enhancement. These can be easily extended to obtain

slightly modified results by increasing the size of the

neighborhood while maintaining the structure and function of

the operator (8)

.

A) Noise Suppression by Mean Filtering

Mean filtering can be achieved by convolving the image with

(2K+1×2L+1) Kernel where each coefficient has a value equal

to the reciprocal of the number of coefficients in the Kernel.

For example, L=K=1, we obtain

W (k, l) =,

Referred to as the 3×3 averaging kernel or mask. Typically,

this type of smoothing reduces noise in the image, but at the

expense of the sharpness of edges [4, 5, 12, 13]. Note that the

size of the kernel is a critical factor in the successful

application of this type of enhancement. Image details that are

small relative to the size of the kernel are significantly

suppressed, while image details are significantly larger than

the kernel size are affected moderately. The degree of noise

suppression is related to the size of the kernel, with greater

suppression achieved by larger kernels. B) Noise Suppression by Median Filtering

Median filtering is a common nonlinear method for noise

suppression that has unique characteristics. It does not use

convolution to process the image with a kernel of coefficients.

Rather, in each position of the kernel frame, a pixel of the

input image contained in the frame is selected to be out put

pixel located at the coordinates of the kernel center. The kernel

frame is centered on each pixel (m, n) of the original image,

and the median value of pixels with in the kernel frame is

computed. The pixel at the coordinates (m, n) of the out put

image is set to this median value. In general, median filters

don’t have the same smoothing characteristics as the mean

filter [4,5,8,9,15]. Features that are smaller than half the size

of the median filter kernel are completely removed by the

filter. Large discontinuities such as edges and large changes in

image intensity are not affected in terms of gray level intensity

by the median filter, although their positions may be shifted by

a few pixels. This nonlinear operation of the median filter

allows significant reduction of specific types of noise. For

example, "shot noise" may be removed completely from an

image without attenuation of significant edges or image

characteristics.

C) Edge Enhancement

Edge enhancement in images is of unique importance because

the human visual system uses edges as a key factor in the

comprehension of the contents of an image. Edges in different

orientations can be selectively identified and enhanced. The

edge-n enhanced images may be combined with the original

image in order to preserve the context. Horizontal edges and

lines are enhanced with image can result in an output image

with both positive and negative values. An enhanced image

with only positive pixels can be obtained either by adding an

offset or by taking the absolute value of each pixel in the

output image. If we are interested in displaying edge – only

information, this may be a good approach. On the other hand,

if we are interested in enhancing edges that are consistent with

kernel and suppressing those that or not, the output image may

be added to the original input image. This addition will most

likely result in a nonnegative image. Edge enhancement


43

appears to provide greater contrast than the original imagery

when diagnosis pathologies. Edges can be enhanced with

several edge operators other than those just mentioned and

illustrated (3)

.

D) Local- Area Histogram Equalization

A remarkably effective method of image enhancement is local-

area histogram equalization, obtained with a modification of

the pixel operation. Local area histogram equalization applies

the concepts of the whole- image histogram equalization to

small, overlapping local areas of the image. It is a nonlinear

operation and significantly increases the observability of subtle

features in the image. The method formulated as shown next is

applied at each pixel (m, n) of the input image. Local –area

histogram equalization is a computationally intensive

enhancement technique. The computational complexity of the

algorithm goes up as the square of the size of the kernel. It

should be noted that since the transformation that is applied to

the image depends on the local neighborhood only, each pixel

is transformed in a unique way. This results in higher visibility

for hidden details spanning very few pixels in relation to the

size of the full image. A significant limitation of this method is

that the mapping between the input and the output images is

nonlinear and highly no monotonic. This means that it is no

appropriate to make quantitative measurements of pixel

intensity on the output image, as the same gray level may be

transformed one way in one part of the image and a completely

different way in another part (4)

.

E) Noise Suppression by Image Averaging

Noise suppression using image averaging relies on three basic

assumptions; firstly; that a relatively large number of input

images are available, secondly; that each input images has

been corrupted by the same type of additive noise, and thirdly;

that the additive noise is random with zero mean value and

independent of the image. When assumptions hold, it may be

advantageous to acquire multiple images with the specific

purpose of using image averaging since with this approach

even severely corrupted images can be significantly enhanced

F) Change Enhancement by Image Subtraction

Image subtraction is generally performed between two images

that are not captured under the same or very similar condition

may need to be registered this may be the case if the images

have been acquired at different times or under different setting.

The output image may be have a very small dynamic range and

may need to be rescaled to the available display range given

tow image (7)

.

VI. NOISE

Noise injects a random or stochastic component into an image

many sources of noise in an image, noise add or subtracts to a

measurement value such that the recorded measurement differs

the actual value mean:

Standard deviation - measure of variability, either naturally

occurring or random fluctuation.

Gaussian Probability Distribution Function

e.g., the height of 1000 third grade school children.

Plot a histogram.

In Gaussian (normal) distribution, the mean describes where

the center of the curve is positioned alone the X- axis and

standard deviation governs how wide the bell curve is, many

commonly encountered measurements of people and things

make for this kind of normal distribution (Gaussian)

Gaussian probability density function

Generic Gaussian PDF with mean =100 and STDev=20

Probability of observing an observation in a range : integrate

area (forG):

1σ=68%

2σ=95 %

3σ=99%

Error bars and confidence intervals are plotted to include

2σ=95%.

VII. MATERIALS AND METHODS

This study was conduct in College of Medical Radiological

Sciences - Khartoum Dental Center and Elnasery Dental Clinic

in period of March 2011 to June 2011. For panoramic images

each film will scan using digitizer scanner then treat by using

image processing program (MatLab), where the enhancement

and contrast of the image was determined. The scanned image

was saved in a TIFF file format to preserve the quality of the

image. The data analyzed used to enhance the contrast within

the soft tissues, the gray levels which can be redistributed both

linearly and nonlinearly using the gray level frequencies and

noise estimation of the original panoramic images.

In Figure 3-1 explains the block diagram about the flow

proposed research paper. Researchers use in this study a

method for image preprocessing of X-ray radiographic images

step by step process. This flow diagram explains the flow of


44

work. Collecting the X-ray images from Govt. Hospitals and

converting this image RGB into Gray color.

Figure 1. Flow chart of image analysis

Panoramic images were filtered by two different filtering

algorithms. The filtering algorithms are used for anisotropic

filtering and median filtering algorithm. The output results

were compared in PSNR and MSE values. The output of

filtering image applied for Image enhancement. That given

image improves the subjective quality of contrast, noise

reduction and edge sharpening. The targets for image

enhancing are better contrast, sharpness of detail and visibility

of features. Several algorithms are Histogram Equalization

(CLAHE). Here researchers applied contrast limited adaptive

histogram equalization algorithm. The best PSNR value of X-

ray image output is given into input of Contrast Limited

Adaptive Histogram Equalization (CLAHE) The data analyzed

by using statistical package, Statistical Package for Social

Studies (SPSS) under windows using t test to measure the

significant difference between contrast within the soft tissues,

the gray levels in both enhanced and unenhanced images and

noise variance (10).

VIII. THE RESULTS

This is an experimental study deals with propose a new

approach for independently identically Distributes (IID) noise

in panoramic images and reduction of the redundancy in the

image data using image processing technique (MatLab version

R2009a). In addition to highlight the role of the proposed

approach (noise variance) by preservation of the image's

overall look, preservation of the diagnostic content in the

image and detection of small and low contrast details in the

diagnostic content of the image and to highlight the role of

using image processing technique in Radiology. The target

population amount for this study was 50 patients present at the

area of the study Table 1.

TABLE 1. SHOWS GENDER DISTRIBUTION IN SAMPLE:

Gender Frequency Percentage

Male 30 60

Female 20 40

Figure 2. Gender Distribution in sample

Thus, in real X-rays images one can find a significant

coherence of the noise with the image content. To illustrate the

problem two panoramic image acquisitions are dedicated; one

image ( figure 3) form OPG scan technique and other image

form another X-ray imaging (figure 4). The figure 3 shows an

idealistic view of panoramic X-rays in figure 4, which needs to

be enhanced and free from the mixture of noise.

Figure 3. Original OPG radiograph (before processing) used to study the

application of analysis approach in Noise Estimation in Panoramic X-

rays images using image processing program (MatLab). (The contrast of

this image was enhanced so that it could be better displayed in the

printed document).


45

Figure 4. Original OPG radiograph (before processing) used to study the

application of analysis approach in Noise Estimation in Panoramic X-

rays images using image processing program (MatLab). (The contrast of

this image was enhanced so that it could be better displayed in the

printed document).

Preprocessing is to improve their quality of images. If these

images are too nosy or blurred they should be filtered and

sharpened. In image processing, filters are mainly used to

suppress either the high frequencies in the image, i.e.

smoothing the images or the low frequencies, i.e. enhancing or

detecting edges in the image. Due to various factors the images

are in general poor in contrast. Researchers applied image

preprocessing to remove artifacts and degradations such as

blurring and noise. A variety of smoothing filters have been

developed that are not linear. While they cannot, in general, be

submitted to Fourier analysis, their properties and domains of

application have been studied extensively. For this reason

researchers applied anisotropic filtering and median filtering.

In study method anisotropic and median filtering algorithms

were used. Anisotropic filter that smoothes noisy regions in the

image while respecting edge boundaries. The another filter

median used to reduce noise in an image, somewhat like the

mean filter (it is a simple , intuitive and easy to implement

method of smoothing images, i.e. reducing the amount of

intensity variation between one pixel and the next. It is often

used to reduce noise in images). The median filter is normally

used to reduce noise in an image, somewhat like the mean

filter. However, it often does a better job than the mean filter

of preserving useful detail in the image. Like the mean filter,

the median filter considers each pixel in the image in turn and

looks at its nearby neighbors to decide whether or not it is

representative of its surroundings. Instead of simply replacing

the pixel value with the mean of neighboring pixel values, it

replaces it with the median of those values. The median is

calculated by first sorting all the pixel values from the

surrounding neighborhood into numerical order and then

replacing the pixel being considered with the middle pixel

value. (If the neighborhood under consideration contains an

even number of pixels, the average of the two middle pixel

values is used.) Figure 5 illustrates an example calculation.

Figure 5. Sample matrix value taken from figure 4-1.

Neighborhood values: 77,77,79,79,79,79,79.79,81

Median Value is: 79

Figure 5 calculating the median value of a pixel neighborhood.

As can be seen, the central pixel value of 81 is rather

unrepresentative of the surrounding pixels and is replaced with

the median value: 79. A 3×3 square neighborhood is used here


46

larger neighborhoods will produce more severe smoothing.

Anisotropic Filtering and Median Filtering algorithms are

applied in ten different X-ray images and calculated the

equivalent MSE and PSNR values. The table-1 shows the

value of MSE and PSNR. The original picture figure 5 RGP is

changed into GRAY color Figure 4. This gray color image

applied into anisotropic filtering method the output is in Figure

6 and also applied in median filtering algorithms the output is

in Figure 7.

Figure 6. Shows Figure 3 OPG radiograph after applying median filter

(The contrast of this image was enhanced so that it could be better

displayed in the printed document).

Figure 6. Shows Figure 4 OPG radiograph after applying median filter

(The contrast of this image was enhanced so that it could be better

displayed in the printed document).

Histogram equalization is a method in image processing of

contrast adjustment using the image's histogram. This method

usually increases the local contrast of many images, especially

when the usable data of the image is represented by close

contrast values. Through this adjustment, the intensities can be

better distributed on the histogram. This allows for areas of

lower local contrast to gain a higher contrast without affecting

the global contrast. Histogram equalization accomplishes this

by effectively spreading out the most frequent intensity values.

Given an image, improve the subjective quality of Contrast,

Noise reduction and Edge sharpening. Fast implementation of

Contrast Limited Adaptive Histogram Equalization (CLAHE)

enhances the contrast of an image by transforming the pixel

values of the image. It operates on small pixel regions (tiles),

rather than the entire image. Each tile's contrast is enhanced,

so that the histogram of the output region approximately

matches the specified histogram. The neighboring tiles are

then combined using bilinear interpolation in order to

eliminate artificially induced boundaries. The contrast,

especially in homogeneous areas, can be limited in order to

avoid amplifying the noise which might be present in the

image.

ACKNOWLEDGMENT

My acknowledgements and gratefulness at the beginning

and at last is to god who gave us the gift of the mind. Profound

thanks and gratitude to everyone who encouraged me to

complete this paper. From the formative stages of it, to the

final draft, I owe an immense debt of gratitude to Miss Nada

Faisal Yousif Mohamed, and Prof. Khalid Hassan Eltom.

Their sound advice and guidance were invaluable. Their work

has stimulated and fostered my efforts in producing this paper. .

REFERENCES

[1] Adelson, E.H., Bergen, J.R. 1991. "The plenoptic function

and the elements of early vision", In Computation Models

of Visual Processing, M. Landy and J.A. Movshon, eds.,

MIT Press, Cambridge, 1991, pp. 3-20.

[2] Arvo, J., 1994, The Irradiance Jacobian for Partially

Occluded Polyhedral Sources, Proc. ACM SIGGRAPH,

ACM Press, pp. 335-342.

[3] Yousif M.Y. Abdallah, 2010, Computed verification of

Light and radiation Field Size on Coblat-60, Lambert

Publisher Press, Germany, P.p. 34-66

[4] Ball, J., Moore, A., 1997, Essential physics for

radiographers, 3rd edition, Blackwell Scientific, Oxford.

[5] Ball, J., Price, T., 1995, Chesney's radiographic imaging,

6th edition, Blackwell Scientific, Oxford.

[6] Buehler, C., Bosse, M., McMillan, L., Gortler, S., Cohen,

M., 2001, Unstructured Lumigraph rendering, Proc. ACM

SIGGRAPH, ACM Press.

[7] Farr, R., Allisy-Roberts, P., 1997, Physics for medical

imaging, W.B. Saunders, London.

[8] Fritsch D.S.; Chaney E.L.; McAuliffe M.J.; Raghavan S.;

Boxwala A.; Earnhart J.R.D., 1995,International Journal

of Radiation Oncology, Biology, Physics, Volume 32,

Number 971, , pp. 217-217.

[9] Georgiev, T., Zheng, C., Nayar, S., Curless, B., Salesin,

D., Intwala, C., 2006, Spatio-angular Resolution Trade-

offs in Integral Photography, Proc. EGSR 2006.

[10] Levoy, M., Ng, R., Adams, A., Footer, M., Horowitz, M.,

2006, Light field microscopy", ACM Transactions on

Graphics (Proc. SIGGRAPH), Vol. 25, No. 3.

[11] Matusik, W., Pfister, H., 2004, 3D TV: a scalable system

for real-time acquisition, transmission, and

autostereoscopic display of dynamic scenes, Proc. ACM

SIGGRAPH, ACM Press.

[12] Yang, J.C., Everett, M., Buehler, C., McMillan, L., 2002,

A real-time distributed light field camera, Proc.

Eurographics Rendering Workshop 2002.

.

BIOGRAPHIES


47

DR. YOUSIF MOHAMED YOUSIF ABDALLAH

Job position: B.Sc, M.Sc., PhD RT (Student)

Lecturer & Radiotherapy Technician

Address: Radiotherapy & Nuclear Medicine

Department, College of Medical Radiological

College- Sudan University of Science and Technology-Crossing

elguish St. and Elshahid St,, Khartoum, Sudan

e-mail: [email protected], [email protected]

web: http://sustech.edu/Colleges/College of Medical Radiologic

Science/Staff

ARWA AHAMED ALMOUSTAFA

Date of Birth: 4/12/1992,

Place of birth: Khartoum (Sudan)

Civil Status: Single, Nationality: Sudanese

Address: Radiotherapy & Nuclear Medicine Department, College of

Medical Radiological College- Sudan University of Science and

Technology-Crossing elguish St. and Elshahid St,, Khartoum, Sudan

GEHAN ELHADI




Address: Diagnostic Radiology Department, College of Medical

Radiological College- Sudan University of Science and Technology-

Crossing elguish St. and Elshahid St,, Khartoum, Sudan

MAMDOUH MOHAMMED




Address: Diagnostic Radiology Department, College of Medical

Radiological College- Sudan University of Science and Technology-

Crossing elguish St. and Elshahid St,, Khartoum, Sudan

OSMAN KHALAFALLAH







TAYSEER KHALID







REFERENCES

PROF. KHALID HASSAN ELTOM



Civil Status: Married, Nationality: British

Job position: MBCB, DMRT, FRCR, Consultant Clinical

Oncologist




Mobile: +249-911628004, +249-126641065

e-mail: [email protected]


Science/Staff

DR. MOHAMED ELFADIL MOHAMED GAR-ELNABI


48



Civil Status: Married, Nationality: Sudanese

Job position: B.Sc, M.Sc., PhD (Med. Phy.), Head of

Radiotherapy and Nuclear Medicine Department




Mobile: +249-924773235

e-mail: [email protected]


Science/Staff

ENG. AZIS-ELDIEN GAFAR MOHAMED




Job position: IT engineering, B.Sc, M.Sc., Computer Science

& Information Technology, Computer Science & Information

Technology Department,

Address: National Cancer Institute, University of Gazeria, Madani,

Sudan

Application of analysis approach in Noise Estimation in Panoramic X- rays images using image...

Documents

Transcript of Application of analysis approach in Noise Estimation in Panoramic X- rays images using image...