59
Page No.
Chapter 2: Literature survey
2.1 Introduction 60
2.2 Literature Survey on Image Processing and its Applications 60
2.3 Literature Survey on MATLAB based Image Processing
Medical Applications 69
2.4 Literature Survey on ANN based Image Processing Applications 71
2.5 Literature Survey for the Detection of Tuberculosis 74
2.6 Motivation for the Present Work 80
2.7 Objectives for the Present Research Work 81
References 82
60
2.1 Introduction
This chapter deals with the literature survey of digital image processing and applications
of the image processing using different techniques.This chapter also describes the
literature survey on MATLAB based image processing medical applications, literature
survey on ANN based Image processing applications, literature Survey for the detection
of tuberculosis using medical and technical methods. From the elaborated literature
survey the motivations for the present work is presented. References are provided at the
end of the chapter.
2.2 Literature survey on image processing and its applications
Many of the techniques of digital image processing, or digital picture processing as it
often was called, were developed in the 1960s at the Jet Propulsion Laboratory,
Massachusetts Institute of Technology, Bell Laboratories, University of Maryland. A few
researches such as application to satellite images, wire-photo standards conversion,
medical imaging, videophone, character recognition, and photograph enhancement were
also carried out[1].
SuezouNakadateet al [2] discussed the use of digital image processing techniques for
electronic speckle pattern interferometry. A digital TV-image processing system with a
large frame memory allows them to perform precise and flexible operations such as
subtraction, summation, and level slicing. Digital image processing techniques made it
easy compared with analog techniques to generate high contrast fringes.
Satoshi Kawataet al [3] discussed the characteristics of the iterative image-restoration
method modified by the reblurring procedure through an analysis in frequency space. An
iterative method for solving simultaneous linear equations for image restoration has an
inherent problem of convergence. The introduction of the procedure called “reblur”
solved this convergence problem. This reblurring procedure also served to suppress noise
amplification. Two-dimensional simulations using this method indicated that a noisy
image degraded by linear motion can be well restored without noticeable noise
amplification.
61
William H [4] highlighted the progress in the image processing and analysis of digital
images during the past ten years. The topics included digitization and coding, filtering,
enhancement, and restoration, reconstruction from projections, hardware and software,
feature detection, matching, segmentation, texture and shape analysis, and pattern
recognition and scene analysis.
David W. Robinson [5] presented the application of a general-purpose image-processing
computer system to automatic fringe analysis. Three areas of application were examined
where the use of a system based on a random access frame store has enabled a processing
algorithm to be developed to suit a specific problem. Furthermore, it enabled automatic
analysis to be performed with complex and noisy data. The applications considered were
strain measurement by speckle interferometry, position location in three axes, and fault
detection in holographic nondestructive testing. A brief description of each problem is
presented, followed by a description of the processing algorithm, results, and timings.
S V Ahmed [6] discussed the work prepared by concentrating upon the simulation and
image processing aspects in the transmission of data over the subscriber lines for the
development of an image processing system for eye statistics from eye .
P K Sahooet al [7] presented a survey of thresholding techniques and updated the earlier
survey work. An attempt was made to evaluate the performance of some automatic global
thresholding methods using the criterion functions such as uniformity and shape
measures. The evaluation was based on some real world images.
Marc Antoniniet al [8] proposed a new scheme for image compression taking psycho-
visual features in to account both in the space and frequency domains. This new method
involved two steps. First, a wavelet transform in order to obtain a set of bi orthogonal
subclasses of images; the original image is decomposed at different scales using a
yramidal algorithm architecture. Second, according to Shannon's rate distortion theory,
the wavelet coefficients are vector quantized using a multi resolution codebook.
Furthermore, to encode the wavelet coefficients, a noise shaping bit allocation procedure
was proposed which assumes that details at high resolution are less visible to the human
eye. Finally, in order to allow the receiver to recognize a picture as quickly as possible at
minimum cost, a progressive transmission scheme was presented. It is showed that the
wavelet transform is particularly well adapted to progressive transmission.
62
Harpen MD [9] presented a wavelet theory geared specifically for the radiological
physicist. As a result, the radiological physicist can expect to be confronted with
elementsof wavelet theory as diagnostic radiology advances into teleradiology, PACS,
and computer aided feature extraction and diagnosis.
Salem Saleh Al-amriet al [10] attempted to undertake the study of segmentation image
techniques by using five threshold methods as Mean method, P-tile method, Histogram
Dependent Technique (HDT), Edge Maximization Technique (EMT) and visual
Technique and they are compared with one another so as to choose the best technique for
threshold segmentation techniques image. These techniques are applied on three satellite
images to choose base guesses for threshold segmentation image.
Wiecek B.et al [11] proposed a new image processing tools for conversion thermal and
visual images, mainly for application in medicine and biology. A novel method for area
and distance evaluation based on statistical differencing was discussed. In order to
increase the measurements accuracy, the interpolation and sub pixel bitmap processing
were chosen.
Patnaiket al [12] presented an image compression method using auto-associative neural
network and embeddedzero-treecoding. The role of the neural network (NN) is to
decompose the image stage by stage, which enabled analysis similar to wavelet
decomposition. This works on the principle of principal component extraction (PCE).
Network training is achieved through a recursive least squares (RLS) algorithm. The
coefficients are arranged in a four-quadrant sub-band structure. The zero-treecoding
algorithm is employed to quantize the coefficients. The system outperformed the
embeddedzero-tree wavelet scheme in a rate-distortion sense, with best perceptual quality
for a given compression ratio.
Shanhui Sun Christian Bauer et al [13] presented a fully automated approach for
segmentation of lungs in CT datasets. The method was specifically designed to robustly
segment lungs with cancer masses and consists of three processing steps. First, a ribcage
detection algorithm is utilized to initialize the model-based segmentation method. Second,
a robust active shape model matching approach is applied to roughly segment the outline
of the lungs. Third, the outline of the matched model is further adapted to the image data
by means of an optimal surface finding approach. The method was evaluated on the
63
LOLA11 test set, consisting of 55 chest CT scans with a variety of different lung diseases
and scan protocols. Compared to a reference standard, mean average and median
volumetric overlap scores of 0 949 and 0, 990 were achieved respectively. Several
examples demonstrated the ability of our method to successfully segment lungs with
cancer masses.
Sonalet al [14] presented various types of image compression techniques. There are
basically two types of compression techniques. One is Lossless Compression and other is
Lossy Compression Technique. Comparing the performance of compression technique is
difficult unless identical data sets and performance measures are used. Some of these
techniques are obtained good for certain applications like security technologies. Some
techniques perform well for certain classes of data and poorly for others.
Suzuki Ket al [15 ] developed an image-processing technique for suppressing the contrast
of ribs and clavicles in chest radiographs by means of a multiresolution massive training
artificial neural network (MTANN). An MTANN is a highly nonlinear filter that can be
trained by use of input chest radiographs and the corresponding "teaching" images.
"bone" images obtained by use of a dual-energy subtraction technique as the teaching
images were employed. A validation test database consisting of 118 chest radiographs
with pulmonary nodules and an independent test database consisting of 136 digitized
screen-film chest radiographs with 136 solitary pulmonary nodules collected from 14
medical institutions are used in this study.
Weixingwanget al [16] presented the newly developed ridge detection algorithm to
diagnose indeterminate nodules correctly, allowing curative resection of early-stage
malignant nodules and avoiding the morbidity and mortality of surgery for benign
nodules. The algorithm was compared to some traditional image segmentation algorithms.
All the results are satisfactory for diagnosis.
Md. FoisalHossainet al [17] presented an enhancement technique based upon a new
application of contrast limited adaptive histograms on transform domain coefficients
called logarithmic transform coefficient adaptive histogram equalization (LTAHE). The
method is based on the properties of logarithmic transform domain histogram and contrast
limited adaptive histogram equalization. A measure of enhancement based on contrast
measure with respect to transform was used as a tool for evaluating the performance of
64
the proposed enhancement technique and for finding optimal values for variables
contained in the enhancement. The algorithm's performance was compared quantitatively
to classical histogram equalization using the aforementioned measure of
enhancement.Experimental results were presented to show the performance of the
proposed algorithm alongside classical histogram equalization.
Wenhong Li . Collet al [18] presented the paper on currency classification system using
image processing techniques. The processing effect and recognition accuracy of RMB is
an important part in the paper currency classification system. According to the
characteristics of RMB images, the paper uses the theory of digital image processing and
pattern recognition to put forward the method of RMB image processing based on the
processing and recognition of the part of the RMB serial numbers, the arithmetic of linear
perception based on rewards and punishment method and the extraction method of serial
numbers character. Through the experiment on the paper currency classification system
which uses the CIS sensor as the image acquisition, it testifies that this method of
recognition has a high feasibility and recognition accuracy.
Li Minxiaet al [19] designed defect extraction by image segmentation. Firstly, on the
basis of wavelet analysis, a new wavelet adaptive threshold denoising method based on
genetic algorithm optimization was proposed. Secondly, an algorithm of multi-scale
morphological to local contrast enhancement was designed. Finally, background is
simulated, and the defect regions were extracted using algorithm of digital subtraction.
The experimental results indicated that the methods can achieve automatic extraction of
defect region, which is always a good foundation for flaw feature parameter extraction
and choice.
Kanwal, N.et al [20] deals with contrast enhancement of X-Ray images and presents here
a new approach for contrast enhancement based upon Adaptive Neighborhood technique.
A hybrid methodology for enhancement has been presented. Comparative analysis of
proposed technique against the existing major contrast enhancement techniques has been
performed and results of proposed technique are promising.
Noorhayati Mohamed Noor et al [21] presented the enhancement capability of adaptive
histogram equalization (AHE) on the soft tissue lateral neck radiograph for suspected fish
bone ingestion. Embedded fish bone lodge in the throat is not easily visible in
65
unprocessed plain radiograph. Serious complication may cause perforation of the lodged
and inflammation that can progress to abscess. Due to the high resolution, the images
were cropped before being processed using adaptive histogram equalization. The quality
of the image was assessed and evaluated during pre and post processing by the
radiologists. The result showed AHE as a promising contrast enhancement for detection
of fish bone in soft tissue at the lateral neck radiographs.
Lu Zhang et al [22] described about Diffraction-enhanced imaging (DEI) and the
capability of DEI to observe different types of tissues was investigated.It is a synchrotron
based imaging technique, which generates high spatial resolution and contrast of both
calcified and soft tissues. This technique not only provided the visualization of absorption
information like conventional X-ray imaging, but also refraction and scattering properties.
In this study the MIR is used to extract information from a series of DEI images.
Md. FoisalHossainet al [23] proposed a method of medical image enhancement based
upon non-linear technique and the logarithmic transform coefficient histogram
equalization using EME as a measure of performance. The performance of this algorithm
was compared to a classical histogram equalization enhancement technique. This method
improves visual quality of images that contain dark shadows due to limited dynamic
range of imaging like X-ray images. Experimental results ascertained that the proposed
technique outperform commonly used enhancement technique like the histogram
equalization qualitatively and quantitatively.
HasanDEMIREL[24] introduced a new face recognition technique based on the gray-
level co-occurrence matrix(GLCM). GLCM represents the distributions of the intensities
and the information about relative positionsof neighboring pixels of an image. Two
methods were being proposed to extract feature vectors using GLCM for
faceclassification. The first method extracts the well-known Haralick features from the
GLCM, and the secondmethod directly uses GLCM by converting the matrix into a vector
that can be used in the classification process. The results demonstrated that the second
method, which uses GLCM directly, is superior to the first method that uses the feature
vector containing the statistical haralick features in both nearest neighbour and neural
networks classifiers. The proposed GLCM based face recognition system not only
outperforms well-known techniques such as principal component analysis and linear
66
discriminant analysis, but also has comparable performance with local binary patterns and
Gabor wavelets.
Pu J et al [25] presented a shape “break-and-repair” strategy for medical image
segmentation and applied it to the segmentation of human lung and pulmonary nodules in
this study. In this approach, the regions that may cause any problems in segmentation
were removed and then estimated using implicit surface fitting based on RBFs. Its most
important characteristic is the capability of segmenting anatomical structures depicted on
medical images in a unified framework within a single pass. The preliminary assessment
results are encouraging and demonstrated the feasibility, generality, and robustness of this
strategy in segmentation.
Hongsheng Li et al [26] proposed a novel predictive model, active volume model (AVM),
for object boundary extraction. It is a dynamic “object” model whose manifestation
includes a deformable curve or surface representing a shape, a volumetric interior
carrying appearance statistics, and an embedded classifier that separates object from
background based on current feature information. The model focused on an accurate
representation of the foreground object’s attributes, and does not explicitly represent the
background. They showed, however, the model is capable of reasoning about the
background statistics which can detect when change is sufficient to invoke a boundary
decision.
Sadeer G. Al-Kindiet al [27] proposed a novel hybrid and repetitive smoothing-
sharpening (HRSST) technique and its impacts are assessed to beneficially enhance
sonogram and mammogram images. The technique aimed to gain and combine the
advantages of both the sharpening process that aims to highlight sudden changes in the
image intensity, with the advantages of iterative image smoothing, which is usually
applied to remove random noise from digital images. Nevertheless the developed
technique also eliminated the drawbacks of each of the two sharpening and smoothing
techniques resulting from their individual application in image processing field. The
proposed technique was tested on both breast ultra-sound (BUS) as well as breast X-ray
mammograms. Results showed that the proposed methodology has high potential to
advantageously enhance the image contrast hence giving extra aid to radiologists to detect
and classify sonograms and mammograms.
67
Sandeep Kumar et al [28] provided a frame work for denoising the enhanced image based
on prior knowledge on the Histogram Equalization. Many image enhancement schemes
like Contrast limited Adaptive Histogram Equalization (CLAHE), Equal area dualistic
sub-image histogram equalization (DSIHE), Dynamic Histogram equalization (DHE)
Algorithm were implemented and compared after the denoising using the wavelet
thresholding. The Performance of all these Methods with the denoising has been analyzed
and a number of Practical experiments of real time images have been presented. From the
experimental results, it is found that all the three techniques with the denoising yields
Different aspects for different parameters.
XiaoyanXu [29] implemented the embedded zero tree wavelet algorithm (EZW), which is
a simple, yet remarkably effective, image compression algorithm. The experiment was
done on a set of standard images and the results show the good performance of this
algorithm compared to some other compression scheme. EZW has proved to be a very
effective image compression method based on the mean-square error (MSE) distortion
measure. Coding results shown in this paper illustrats the performance of this
improvement.
Arpita Mittal et al [30] reported that as till date there is no proven cure for the disease i.e.
Rheumatoid Arthritis(RA), hence close monitoring of the disease is important in the
medical treatment of this disease . An application of image processing techniques for
identification of most common disease (RA) is opted. In this paper Fingers and Knee
images of the patient having RA have been analyzed through Morphological Image
processing techniques. The processed images find their application in the field of Medical
Science and can be beneficial for doctors in identification of disease stages from
monitoring point of view.
Jagadeeshet al [31] presented the preprocessing methods of the leukemic blast cells image
in order to generate the features well characterizing different types of cells. The solved
problems include: the segmentation of the bone marrow aspirate by applying the
watershed transformation, selection of individual cells, and feature generation on the
basis of texture, statistical and geometrical analysis of the cells.
KimmiVermaet al [32] did a research which made the use of software with edgedetection
and segmentation methods, which gave the edge pattern and segment of brain and the
68
brain tumor itself. In this research, it has provided a foundation of segmentation and
edges reviewed with an emphasis placed on revealing the advantages and disadvantages
of these methods for medical imaging applications. The use of image segmentation in
different imaging modalities is also described along with the difficulties encountered in
each modality.
Ashraf Anwar et al [33] introduced an inexpensive, user friendly general-purpose image
processing tool and visualization program specifically designed in MATLAB to detect
much of the brain disorders as early as possible. The application provided clinical and
quantitative analysis of medical images. Minute structural difference of brain gradually
results in major disorders such as schizophrenia, Epilepsy, inherited speech and language
disorder, Alzheimer's dementia etc. Here the main focusing is given to diagnose the
disease related to the brain and its psychic nature (Alzheimer’s disease). Medical
imaging is expensive and very much sophisticated because of proprietary software and
expert personalities.
Pallavi T. Suradkar [34] reviewed image analysis studies aimed at automated diagnosis or
screening of malaria infection in microscope images of thin blood film smears.
Md.AmranHossenBhuiyanet al [35]reportedthat in order to achieve an effective way to
identify skin cancer at an early stage without performing any unnecessary skin biopsies,
digital images of melanoma skin lesions were investigated. To achieve this goal, feature
extraction was considered as an essential weapon to analyze an image appropriately In
this paper, different digital images have been analyzed based on unsupervised
segmentation techniques. Feature extraction techniques are then applied on these
segmented images. After this, a comprehensive discussion has been explored based on the
obtained results. Signal and imaging investigations are currently a basic step of the
diagnostic, prognostic and follow-up processes of heart diseases.
R K Samantarayet al [36] has presented an effective way to achieve a high-level
integration of signal and image processing methods in the general process of care, by
means of a clinical decision support system(CDSS), and has discussed the advantages of
such an approach. In particular, significant and suitably designed image and signal
processing algorithms are introduced to objectively and reliably evaluate important
features that, in collaboration with the CDSS, could facilitate decisional problems in the
69
heart failure domain. Further-more, additional signal and image processing tools enrich
the model base of the CDSS.
S.Kannadhasanet al [37] described a method which is not only effectively detecting the
presence of cancer cells but also it is reducing the overall time taken for diagnosis by
carrying the whole process under biotelemetry. On the other hand, biotelemetry is mostly
used for one dimensional signals thus in this project it extended for transferring two
dimensional signals i.e., image if it happen so then complex or time consuming diagnosis
process completes in short duration. The telemetry link was provided by Zigbee
transceivers and diagnosis was carried with the help of digital image processing
technique.
HardikPandit [38]discussed an application of digital image processing and analysis
techniques., which can be useful in healthcare domain to predict some major diseases of
human beings. The application is an image processing system, which works on the basis
of medical palmistry. The images of human palm form input to the system. Then, system
applies digital image processing and analysis techniques on input images to identify
certain features in the image. By using knowledge base of medical palmistry it analyzes
certain features in image and predicts probable disease.
2.3 Literature survey on MATLAB based image processing medical applications
M Bister [39] illustrated some of the important points with fast implementations of
bilinear interpolation, watershed segmentation and volume rendering with MATLAB, as
MATLAB has often been considered an excellent environment for fast algorithm
development but is generally perceived as slow and hence not fit for routine medical
image processing, where large data sets are now available e.g., high resolution CT image
sets with typically hundreds of 512x512 slices. Yet, with proper programming practices –
vectorization, pre-allocation and specialization – applications in MATLAB could run as
fast as in C language.
JiřiBlahuta [40] presented a processing of medical ultrasound images with MATLAB.
This processing is useful to potential diagnosis of Parkinson´s disease in brain-stem
area.Furthermore introduced DICOM standard for medical imaging and modern 3D/4D
scanning for high level and accuracy of diagnoses that was higher than traditionally 2D
scanning.
70
Joaquim Jose Furtado et al [41] aimed to realize the image classification using MATLAB
software. The image was classified using three and five classes, with a population size of
20 and time of 30, 50 and 100. The gotten results showed that the time seems to affect the
classification more than the number of classes.
S. AllinChristeet al [42] presented an efficient architecture for various image filtering
algorithms and tumour characterization using Xilinx System Generator (XSG). This
architecture offered an alternative through a graphical user interface that combines
MATLAB, Simulink and XSG and explored important aspects concerned to hardware
implementation. Performance of this architecture implemented in SPARTAN-3E Starter
kit (XC3S500E-FG320) exceeds those of similar or greater resources architectures. The
proposed architecture reduced the resources available on target device by 50%.
NasrulHumaimiMahmoodet al [43] reported a survey of image processing algorithms
that have been developed for detection of masses and segmentation techniques. 35
students from university campus participated in the Development of Biomedical Image
Processing Software Package for New Learners Survey investigating the use of software
package for processing and editing image. Composed of 19 questions, the survey built a
comprehensive picture of the software package, programming language, workflow of the
tool and captured the attitudes of the respondents. The result of this study showed that
MATLAB is among the famous software package and its result is expected to be
beneficial and able to assist users on effective image processing and analysis in a newly
developed software package.
Ching Yee Yong et al [44] made a survey of image processing algorithms that were
developed for detection of masses and segmentation techniques. The result of this study
showed that MATLAB is among the famous software package; more than 60% of the
respondents prefer to use MATLAB for their image processing work. The Microsoft
Photo Editor is the second popular software for images editing process. More than 30% of
respondents are very likely to use a ready-to-use package for processing image rather than
given source code. The result is expected to be beneficial and is able to assist users on
effective image processing and analysis in a newly developed software package. A
preliminary image processing tool prototype that was developed is also being presented
in the paper.
71
Deepak Kumar Garget al [45] discussed a method that involves processing of ECG paper
records by an efficient and iterative set of digital image processing techniques for the
conversion of ECG paper image data to time series digitized signal form, resulting in
convenient storage and retrieval of ECG information. The method involved are
calculation of Heart rate, QRS Width and Stability (variation in R-R peaks) from the
extracted signal. Comparison of the above calculated parameters with the manually
calculated parameters showed an accuracy of 96.4%, thus proving the effectiveness of the
process. The author also proposed the development of fuzzy based ECG diagnosis system
that assists the doctors in diagnosis.
ShiruiGao [46] emphasized the MATLAB based medical image processing tools. It
includes the theoretical background and examples. Through MATLAB this paper made
the introduction of the post-imaging quality in medical technology and medical imaging.
It also introduces the medical image processing technology and describes the image
processing and processing technologies, including the organ contours, interpolation,
filtering, and segmentation techniques. In medicine, the DICOM image data processing
using MATLAB is also widely used in this type of image processing.
BhausahebShindeet al [47] proposed a method to improve the accuracy of MRI, Cancer,
X-ray and Brain images for easy diagnosis. For this experimental work they took different
medical images like MRI, Cancer, X-ray, and Brain and calculated standard derivations
and mean of all these medical images after finding Gaussian noise and then applied
median filtering technique for removal of noise. After removing noise by using median
filtering techniques, again standard derivations and mean are evaluated. The results,
achieved were more useful and they proved to be helpful for general medical practitioners
to analyze the symptoms of the patients with ease.
2.4 Literature survey on ANN based Image processing applications
Artificial Neural Network is a branch of Artificial intelligence and has been accepted as a
new technology in computer science. Neural Networks are currently a 'hot' research area
in medicine, particularly in the fields of radiology, urology, cardiology, oncology and etc.
It has a huge application in many areas such as education, business; medical, engineering
and manufacturing . Artificial neural networks are finding many uses in the medical
72
diagnosis application. Neural Network plays an importat role in a decision support
system.
HeryPURNOMOet al [48 ] explored the performance testing for PCNN method
compared to the classical standard method for tuberculosis detection. There was
significant improvement in processing time and Diagnosis percentage, which the image is
processed first with Adaptive White Gaussian noise (AWGN) for reliability testing of the
method.
Juan A et al [49] described some important aspects of recent visual cortex-based ANN
models and finally discussed about the conclusions reached throughout the process.
N Ganesanet al [50] made an attempt to make use of neural networks in the medical
field (carcinogenesis (pre-clinical study)). In carcinogenesis, artificial neural networks
have been successfully applied to the problems in both pre-clinical and post-clinical
diagnosis. The main aim of research in medical diagnostics is to develop more cost-
effective and easy–to-use systems, procedures and methods for supporting clinicians. It
has been used to analyze demographic data from lung cancer patients with a view to
developing diagnostic algorithms that might improve triage practices in the emergency
department. For the lung cancer diagnosis problem, the concise rules extracted from the
network ,achieve a high accuracy rate of on the training data set and on the test data set.
Dilip Roy Chowdhuryet al [51] reported the use of artificial neural networks in predicting
neonatal disease diagnosis. The proposed technique involves training a Multi Layer
Perceptron with a BP learning algorithm to recognize a pattern for the diagnosing and
prediction of neonatal diseases. A comparative study of using different training algorithm
of MLP, Quick Propagation, Conjugate Gradient Descent, shows the higher prediction
accuracy. The Back propagation algorithm was used to train the ANN architecture and the
same has been tested for the various categories of neonatal disease. About 94 cases of
different sign and symptoms parameter have been tested in this mode l. This study
exhibits ANN based prediction of neonatal disease and improved the diagnosis accuracy
of 75% with higher stability.
QeetharaKadhimet al [52] presented a method to evaluate artificial neural network in
disease diagnosis. Two cases were studied. The first one is acute nephritis disease; data is
the disease symptoms. The second is the heart disease; data is on cardiac Single Proton
73
Emission Computed Tomography (SPECT) images. Each patient classified into two
categories: infected and non-infected. Classification is an important tool in medical
diagnosis decision support. Feed-forward back propagation neural network is used as a
classifier to distinguish between infected or non-infected person in both cases. The results
of applying the artificial neural networks methodology to acute nephritis diagnosis based
upon selected symptoms showed abilities of the network to learn the patterns
corresponding to symptoms of the person.
Hasanet al [53] introduced a new face recognition technique based on the gray-level co-
occurrence matrix (GLCM). GLCM represents the distributions of the intensities and the
information about relative positions of neighboring pixels of an image. Two methods
were proposed to extract feature vectors using GLCM for face classification. The first
method extracts the well-known Haralick features from the GLCM, and the second
method directly uses GLCM by converting the matrix into a vector that can be used in the
classification process. The results demonstrated that the second method, which uses
GLCM directly, is superior to the first method that uses the feature vector containing the
statistical Haralick features in both nearest neighbor and neural networks classifiers. The
proposed GLCM based face recognition system not only outperforms well-known
techniques such as principal component analysis and linear discriminate analysis, but also
has comparable performance with local binary patterns and Gabor wavelets.
MussaratYasminet al [54] summarized overview of research and development held in
recent past highlighting the role of Neural Networks in advancement of Medical Imaging.
Zhenghao Shi [55] reviewed the application of artificial neural networks in medical image
preprocessing, in medical image object detection and recognition. Main advantages and
drawbacks of artificial neural networks were discussed. By this survey, the paper tried to
answer what the major strengths and weaknesses of applying neural networks for medical
image processing would be.
74
2.5 Literature Survey for the Detection of Tuberculosis
2.5.1 Clinical methods
Tuberculosis is diagnosed by finding Mycobacterium tuberculosis bacteria in a clinical
specimen taken from the patient. While other investigations may strongly suggest
tuberculosis as the diagnosis, they cannot confirm it.
A complete medical evaluation for tuberculosis (TB) must include a medical history, a
physical examination, a chest X-ray and microbiological examination (of sputum or some
other appropriate sample). It may also include a tuberculin skin test, other scans and X-
ray, surgical biopsy.
A definitive diagnosis of tuberculosis can only be made by culturing Mycobacterium
tuberculosis organisms from a specimen taken from the patient (most often sputum, but
may also include pus, CSF, biopsied tissue, etc.). A diagnosis made other than by culture
may only be classified as "probable" or "presumed". For a diagnosis negating the
possibility of tuberculosis infection, most protocols require that two separate cultures both
test negative.[56]
2.5.1.1 Sputum
Sputum smears and cultures are done for acid-fast bacilli if the patient is producing
sputum.[57] The preferred method for this is fluorescence microscopy (auramine-
rhodamine staining), which is more sensitive than conventional Ziehl-Neelsen
staining.[58] In cases where there is no spontaneous sputum production, a sample can be
induced, usually by nebulized inhalation of a saline or saline with bronchodilator solution.
A comparative study found that inducing three sputum samples is more sensitive than
three gastric washings.[59]
2.5.1.2 Abreugraphy
A variant of the chest X-Ray, abreugraphy (from the name of its inventor, Dr. Manuel
Dias de Abreu) was a small radiographic image, also called miniature mass radiography
(MMR) or miniature chest radiograph. Though its resolution is limited (it doesn't allow
the diagnosis of lung cancer, for example) it is sufficiently accurate for diagnosis of
tuberculosis [60].
75
Much less expensive than traditional X-Ray, MMR was quickly adopted and
extensively utilized in some countries, in the 1950s. For example, in Brazil and in Japan,
tuberculosis prevention laws went into effect andabout 60% of the population to
undergoMMR screening.
The procedure went out of favor, as the incidence of tuberculosis dramatically
decreased, but is still used in certain situations, such as the screening of prisoners and
immigration applicants.
2. 5 .1.3 Immunological test
ALS Assay
Antibodies from Lymphocyte Secretion or Antibody in Lymphocyte Supernatant or ALS
Assay is an immunological assay to detect active diseases like tuberculosis, cholera,
typhoid etc. Recently, ALS assay nods the scientific community as it is rapidly used for
diagnosis of Tuberculosis. The principal is based on the secretion of antibody from in
vivo activated plasma B cells found in blood circulation for a short period of time in
response to TB-antigens during active TB infection rather than latent TB infection [61].
2.5.1.4 Nucleic acid amplification tests (NAAT)
This is a heterogeneous group of tests that use the polymerase chain reaction (PCR)
technique to detect mycobacterial nucleic acid. These test vary in which nucleic acid
sequence they detect and vary in their accuracy. The two most common commercially
available tests are the amplified mycobacterium tuberculosis direct test (MTD, Gen-
Probe) and Amplicor (Roche Diagnostics) [62]. In 2007, a systematic review of NAAT
by the NHS Health Technology Assessment Program concluded that "NAAT test
accuracy to be far superior when applied to respiratory samples as opposed to other
specimens. Although the results were not statistically significant, the AMTD test appears
to perform better than other currently available commercial tests."[63].
As the sputum test is the universally accepted medical test for the detection of TB, so, we
have chosen the test as the clinical result
76
2.5.2 Technical Methods
Plikaytis BDet al [64] presented a computerized pattern recognition model used to
speciation mycobacterium based on their restriction fragment length polymorphism
(RFLP) banding patterns. Thirty-nine independent strains of known origin, not included
in the probability matrix, were used to test the accuracy of the method in classifying
unknowns: 37 of 39 (94.9%) were classified correctly. An additional set of 16 strains of
known origin representing species not included in the model were tested to gauge the
robustness of the probability matrix. Every sample was correctly identified as an outlier,
i.e. a member of a species not included in the original matrix.
S APatil [65] presented a computer algorithm for texture analysis of TB chest radiograph.
Algorithm included important steps, like image acquisition, image pre-processing, lung
field segmentation, and features extraction. Total 49 images were used during experiment
to estimate 1st and 2nd order texture features. Gray Level Co-occurrence Matrix (GLCM)
technique is used to estimate texture features .
K. Veropouloset al [66] presented a method of developing an automated method for the
detection of tubercle bacilli in clinical specimens, principally sputum smears to improve
the diagnostic process. A preliminary investigation is presented here, which makes use of
image processing techniques and neural network classifiers for the automatic
identification of TB bacilli on Auramine stained sputum specimens. The developed
system showed a sensitivity of 93.5% for the identification of individual bacilli. As there
are usually fairly numerous TB bacilli in the sputum of patients with active pulmonary
TB. The overall diagnostic accuracy for sputum smear positive patients was expected to
be very high. Potential benefits of automated screening for TB are rapid and accurate
diagnosis, increased screening of the population, and reduced health risk to staff
processing slides.
Rachna H. B et al [67] proposed an algorithm based on image processing technique for
identification of TB bacteria in sputum, as the availability of expertise, time and cost are
the constraints of the human intervention based examinations. The method is based on
Otsu thresholding and k-means clustering approach. The performance of clustering and
thresholding algorithms for segmenting TB bacilli in tissue sections is compared. The
developed automated technique showed good accuracy and efficiency.
77
P. Sadaphalet al [68] demonstrated the proof of principle of an innovative computational
algorithm that successfully recognized Ziehl-Neelsen (ZN) stained acid-fast bacilli (AFB)
in digital images. Automated, multi-stage, color-based Bayesian segmentation identified
possible ‘TB objects’, removed artifacts by shape comparison and color-labeled objects as
‘definite’, ‘possible’ or ‘non-TB’, bypassing photo micrographic calibration.
Superimposed AFB clusters, extreme stain variation and low depth of field were
challenges. This novel method facilitated electronic diagnosis of TB, permitting wider
application in developing countries where fluorescent microscopy is currently
inaccessible and unaffordable. [49]
Stefan Jaeger et al [69] described the medical background of TB detection in chest X-
rays and presented a survey of the recent approaches using computer-aided detection.
After a thorough research of the computer science literature for such systems or related
methods, 16 papers were identified, including our own, written between 1996 and early
2013. These papers showed that TB screening is a challenging task and an open research
problem. They reported on the progress to date and described experimental screening
systems that have been developed.
Manuel G foreroet al [70] developed a new autofocus algorithm and a new bacilli
detection technique with the aim to attain a high specificity rate and reduce the time
consumed to analyze sputum samples. This technique is based on the combined use of
some invariant shape features together with a simple thresholding operation on the
chromatic channels. Some feature descriptors were extracted from bacilli shape using an
edited dataset of samples. A k-means clustering technique was applied for classification
purposes and the sensitivity vs specificity results were evaluated using a standard ROC
analysis procedure.
Ajay Divekar [71] developed a stepwise classification (SWC) algorithm to remove
different types of false positives, one type at a time, and to increase the detection of TB
bacilli at different concentrations. Based on the Shannon cofactor expansion on Boolean
function for classification, Both bacilli and non-bacilli objects are first analyzed and
classified into several different categories including scanty positive, high concentration
positive, and several non-bacilli categories: small bright objects, beaded, dim elongated
objects, etc. The morphological and contrast features were extracted based on a prior
clinical knowledge. The SWC is composed of several individual classifiers. Individual
78
classifier to increase the bacilli counts utilizes an adaptive algorithm based on a
microbiologist’s statistical heuristic decision process. Individual classifier to reduce false
positive is developed through minimization from a binary decision tree to classify
different types of true and false positive based on feature vectors. Finally, the detection
algorithm was tested on 102 independent confirmed negative and 74 positive cases. A
multi-class task analysis showed high accordance rate for negative, scanty, and high-
concentration as 88.24%, 56.00%, and 97.96%, respectively.
Jeannette Chang [72] presented an algorithm for automated TB detection in smear images
taken by digital microscopes such as Cell Scope, a novel low-cost, portable device
capable of bright field and fluorescence microscopy. Automated processing on such
platforms could save lives by bringing healthcare to rural areas with limited access to
laboratory-based diagnostics. Though the focus of the study was the application of
automated algorithm to Cell Scope images, the method may be readily generalized for use
with images from other digital fluorescence microscopes. The algorithm applies
morphological operations and template matching with a Gaussian kernel to identify TB-
object candidates. Then moment, geometric, photometric, and oriented gradient features
were used to characterize these objects and perform discriminative, support vector
machine classification. Then the algorithm was tested on a large set of CellScope
fluorescence images from sputum smears collected at clinics in Uganda (594 images
corresponding to 290 patients). The object- level classification is highly accurate, with
Average Precision of 89:2% _ 2:1%. For slide-level classification, the algorithm
performed at the level of human readers, demonstrating the potential for making a
significant impact on global healthcare.
The main drawback with these methods are that the work is trying to replace a clinical
test. They are working on the detection of bacilli using different techniques such as
neural network, which is bit complicated as for further classification we need to use
another neural network which becomes the usage of multiple neural network. To use this
method we need a high resolution image which is a bit expensive.
The main problem in the texture analysis of chest radiographs is the complex
“background” of superimposed normal anatomical structures to which the analysis must
be somehow insensitive.
79
The enhancement method [14] is quite sufficient in diagnosing the fish bone quickly.
However the enhanced images not only enhance the fish bone but also enhance the noise
that is present in the radiographs images. Hence, all these factors motivated the author to
design and develop a novel technique for TB detection.
2.6 Motivation for the Present Research Work
Tuberculosis (TB) is one of the most important public health problems worldwide. There
are 9 million new TB cases and nearly 2 million TB deaths each year. Case-finding and
the management of pulmonary tuberculosis is an essential target of tuberculosis control
programs. However, pulmonary tuberculosis (PTB) is becoming more and more of a
serious problem, particularly in countries affected by epidemics of human
immunodeficiency virus (HIV)-TB co-infection. The diagnosis of PTB using prompt and
accurate methods is a crucial step in the control of the occurrence and prevalence of TB.
However, the diagnosis of PTB is quite complex, so there is no unified standard at
present. Frequently, there is over diagnosis and missed diagnosis and it is a thorny
question in the field of TB control. That is why TB as the parameter is chosen. If it is not
diagnosed in the early stage it may lead to death. The image processing techniques along
with the usage of Artificial Neural Network in instrumentation is used in this work for the
design of diagnosing system and hence provides better way for the further treatment.
After elaborate literature survey we found that most of the studies on TB diagnosis
reported were on conventional clinical test based or technical methods separately for
identifying the TB. These tests involve identification of presence/absence of bacillus
which is a cause for TB. Besides, no studies reported on the severity (percentage) of TB
in a patient.
In addition, the manual screening for the bacillus identification involves a labour
intensive task with a high false negative rate. Automatic screening will entail several
advantages, like a substantial reduction in the labour workload of clinicians, improving
the sensitivity of the test and a better accuracy in diagnosis by increasing the number of
images that can be analyzed by the computer.
Though the disease seems to be simple but it is very much infectious and has to be cured
in time or in other words in early stage. So, as far as the treatment is concerned, diagnosis
becomes the primary and crucial stage of the disease. There are many ways of diagnosing
80
the TB. One of them is the sputum examination. This test is accepted worldwide. X-ray
analysis is another technique of diagnosing TB. In this method, the radiologist or a
consultant physician has to take the print of X-ray and analyze the presence of the disease
(TB). The second method is quite expensive, tedious and may not yield precise results.
This has motivated the author to design a system which can diagnose the presence of TB
without taking the X-ray film print. And also according to the survey made with senior
doctors, it is learnt that the diagnosis of TB for inexperienced doctors is very difficult.
The cost level can also be reduced as the X ray print need not be taken. At the same time
according to the survey made, people have used either X-ray of chest or the sputum result
or the image of the smear to find out the bacilli. But, in the proposed system sputum as
well as X-ray image which has led to more accuracy in identifying and predicting the
percentage of TB has been used.
As the present systemwill diagnosis of Pulmonary Tuberculosis with the help of X-ray &
sputum results which are mandatory for a specialist to detect TB, So, the system is not
very expensive. At the same time the junior doctor who is not experienced can use the
system for the diagnosis pulmonary TB.
Over all, the aim is to design a system which helps the patient through the doctor. Also, as
the X-ray system is already digitized ithas beenmade very convenient for taking the
softcopy of the images in diagnosing the TB along with the sputum examination result.
These have motivated to carry out this research work. Hence, the proposed design is a
novel approach combining the conventional (sputum analysis) and modern (X-ray
analysis) methods to diagnose and thus better treat the fatal TB in human beings.
81
2.7 Objectives for the Present Research Work
To collect the lung X-ray images of PTB and normal patients.
To preprocess the X-ray images.
To extract the features from the X- ray images.
To design an ANN for further investigation.
To train the ANN with the extracted features.
To design a GUI for user.
To test unknown X-ray images for the detection of PTB.
To check the severity of PTB.
Detailed study on the methodology i.e., design and development of overall system is dealt
in the next chapter.
82
References
[1] Azriel Rosenfeld, “Picture Processing by Computer”, New York: Academic Press,1969.
[2] SuezouNakadate, ToyohikoYatagai, and Hiroyoshi Saito “Electronic speckle patterninterferometry using digital image processing techniques”Applied Optics, vol. 19,Issue 11, 1980, pp. 1879-1883.
[3] Satoshi Kawata and YoshikiIchioka “Iterative image restoration for linearlydegraded images. II. Reblurring procedure,” Journal of the Optical society ofAmerica(JOSA), vol. 70, 1980, pp. 768–772.
[4] William H. Carter San Diego “Evaluation of Peak Location Algorithms WithSubpixel Accuracy For Mosaic Focal Planes” Processing of Images and Data fromOptical Sensors, Conference Volume 0292, 1981.
[5] David W. Robinson “Automatic fringe analysis with a computer image-processingsystem” Applied Optics, vol. 22, Issue 14, 1983, pp. 2169-2176.
[6] S. V. Ahamed ,V. B. “An image processing systemforeye statistics from eyediagrams” Lawrence IAPR Workshop on CV- SpealHarclware and IndustrialApplications October 12-14. 1988. Tokyo.
[7] P. K. Sahoo, S. Soltaniand A. K. C. Wong “A Survey of thresholding Techniques”Computer vision, graphics, and image processing,vol. 41,1988, pp. 233-260.
[8] Marc Antonini, Michel Barlaud“Image Coding Using Wavelet Transform”IEEEtransactions on image processing, vol. 1, no.2. APRIL 1992.
[9] Harpam MD “An introduction to wavelet theory and application for the radiologicalphysicist”. Med Phys. 1998, vol. 25,no.10, pp.1985-93.
[10] Salem Saleh Al-amri ,N.V. Kalyankar and KhamitkarS.D “Image Segmentation byUsing Thershod Techniques” Journal of computing, vol. 2, Issue 5, May 2010,pp.83-86.
[11] Wiecek, B., Danych, R. ; Zwolenik, Z.Jung, A.“Engineering in Medicine andBiology Society”, Proceedings of the 23rd Annual International Conference of theIEEE, vol. 3, 2001, pp. 2805 – 2807.
[12] Patnaik, S.Pal, R.N. “ Image compression using auto-associative neural networkand embedded zero-tree coding”, IEEE Third Workshop on WirelessCommunications Wireless Communications, 2001,pp. 388-390.
[13] Shanhui Sun Christian Bauer and ReinhardBeichel“Robust Active Shape ModelBased Lung Segmentation in CT Scans”, LOLA11 Challenge pp. 213 -223.
[14] Sonal, Dinesh Kumar “A study of various image compression techniques”.www.rimtengg.com/coit2007/proceedings/pdfs/43.pdf
83
[15] Suzuki K, Abe H, MacMahon H, Doi K, “Image-processing technique forsuppressing ribs in chest radiographs by means of massive training artificial neuralnetwork (MTANN)”. IEEE Transactions on Medical Imaging., 2006, vol. 25,no.4,pp. 406-416.
[16] WeixingWang,Shuguang Wu “A Study on Lung Cancer Detection by ImageProcessing” international conference on Communications,Circuits and SystemsProceedings, 2006, pp. 371-374.
[17] Md. FoisalHossain, Mohammad Reza Alsharif “Image Enhancement Based onLogarithmic Transform Coefficient and Adaptive HistogramEqualization”International Conference on Convergence Information Technology,21-23 November, 2007, pp. 1439 – 1444.
[18] Wenhong Li,Yonggang Li,KexueLuo, “Application of Image ProcessingTechnology in Paper Currency Classification System”, IEEE transactions 22-24Oct. 2008, pp. 1- 5.
[19] Li Minxia, ZhengMeng “A Study of Automatic Defects Extraction of X-ray WeldImage Based on Computed Radiography System” International Conference onMeasuring Technology and Mechatronics Automation - ICMTMA, 2011.
[20] Kanwal, N. , Girdhar, A. ; Gupta, S “Region Based Adaptive ContrastEnhancement of Medical X-Ray Images”, Bioinformatics and BiomedicalEngineering, (ICBBE) 2011 5th International Conference, pp. 1-5.
[21] Noorhayati Mohamed Noor, Noor Elaiza Abdul Khalid. “Fish Bone ImpactionUsing Adaptive Histogram Equalization (AHE)” Proceedings of the SecondInternational Conference on Computer Research and Development 2010, IEEEComputer society Washington, pp.163-167.
[22] Lu Zhang, Dongyue Li, ShuqianLuo , “Information extraction of bone fractureimages based on diffraction enhanced imaging” International Conference ofMedical Image Analysis and Clinical Application (MIACA) 10-17 June 2010, pp.106 -108.
[23] Md.FoisalHossain, Mohammad Reza Alsharif, and Katsumi Yamashita “MedicalImage Enhancement Based on Nonlinear Technique and Logarithmic TransformCoefficient Histogram Matching”, IEEE/ICME International Conference onComplex Medical Engineering July,2010, pp. 13-15.
[24] HasanDemirel “Co-occurrence matrix and its statistical features as a new approachfor face recognition”Turk J ElectricalEngeneeringand Computer Society, vol.19,No.1, 2011,pp.97-107.
[25] Pu J, Paik DS, Meng X, Roos JE, Rubin GD“Shape “Break-and-Repair” Strategyand Its Application to Automated Medical Image Segmentation”,IEEE transactionson visualization and computer graphics, vol.17, no. 1, january 2011.
[26] TianShen, Hongsheng LiXiaolei Huang “ Active Volume Models for MedicalImage Segmentation” Medical Imaging, IEEE Transactions on vol. 30, Issue 3March 2011, pp. 774 – 791.
84
[27] Sadeer G. Al-Kindi, “Breast Sonogram and Mammogram Enhancement UsingHybrid and Repetitive Smoothing-Sharpening Technique”, 1st Middle EastConference onBiomedical Engineering (MECBME), 21-24 Feb. 2011, pp. 446 –449.
[28] Sandeep Kumar, PuneetVerma “Comparison of Different Enhanced ImageDenoising with Multiple Histogram Techniques”, International Journal of SoftComputing and Engineering (IJSCE), vol. 2, Issue. 2, May 2012.
[29] XiaoyanXu, “Embedded Zero Tree as Image Coding”, School of Engineering,UniversityofGuelph.qh.eng.ua.edu/classes/spring2007/...files/.../project_report_jpeg2000.pdf
[30] Arpita Mittal, Sanjay Kumar Dubey , “Analysis of MRI Images of RheumatoidArthritis through Morphological Image Processing Techniques” IJCSI InternationalJournal of Computer Science Issues, vol. 10, Issue 2, No. 3, March 2013,pp.118-122.
[31] S.JagadeeshDr.E.NagabhooshanamDr.S.Venkatachalam , “Image processing basedapproach to cancer cell prediction in blood samples”International Journal ofTechnology and Engineering Sciences, vol. 1, pp.1- 4.
[32] KimmiVerma, AruMehrotra, VijayetaPandey, Shardendu Singh, “Image processingtechniques for the enhancement of brain tumour patterns”.
[33] Ashraf Anwar &ArsalanIqbal, “Image Processing Technique for Brain AbnormalityDetection” International Journal of Image Processing (IJIP), vol. 7, Issue 1, 2013,pp. 51-61.
[34] Pallavi T. Suradkar , “Detection of Malarial Parasite in Blood Using ImageProcessing”International Journal of Engineering and Innovative Technology (IJEIT),vol. 2, Issue 10, April 2013.
[35] Md.AmranHossenBhuiyan, Ibrahim Azad, Md.KamalUddi“ Image Processing forSkin Cancer Features Extraction” International Journal of Scientific & EngineeringResearch,Volume 4, Issue 2, February 2013.
[36] R K Samantaray, T K Mohanta , “Image Processing for Decision Support in HeartFailure”.Researcher 2013, Volume 5, no.4, pp.1-8.
[37] S.Kannadhasan, N.BasheerAhamed, M.RajeshBaba , “Cancer Diagonsis with thehelp digital Image Processing using ZIGBEE Technology”, International Journal ofEmerging Trends in Electrical and Electronics (IJETEE), Volume 1, Issue 2, March2013.
[38] HardikPandit ,Dr. D M Shah “Application of Digital Image Processing and Analysisin Healthcare Based on Medical Palmistry”International Conference on IntelligentSystems and Data Processing ICISD 2011 Special Issue published by InternationalJournal of Computer Applications (IJCA), pp. 56-59.
[39] M Bister*, “Increasing the speed of medical image processing in MATLAB”,Biomedical Imaging and Intervention Journal , pp. 2-12.
85
[40] JiřiBlahuta, TomašSoukup, PetrČermak, ”Image processing of medical diagnosticneurosonographical images in MATLAB” Recent Researches in Computer Science,ISBN: 978-1-61804-019-0, pp.85-90.
[41] Joaquim Jose Furtado ,ZhihuaCai , Liu Xiaobo, “Digital image processing:supervised classification using genetic algorithm in matlab toolbox” Report andOpinion, 2010, Volume2, no.6, pp. 53-61.
[42] Mrs. S. AllinChriste, Mr.M.Vignesh, Dr.A.Kandaswamy, “An efficient FPGAimplementation of MRI image filtering and tumour characterization using xilinxsystem generator” International Journal of VLSI design & CommunicationSystems (VLSICS), Volume 2, No.4, December 2011.
[43] NasrulHumaimiMahmood, Ching Yee Yong, Kim Mey Chew and Ismail Ariffin,“Image Processing Software Package in Medical Imaging”, A review InternationalJournal of Computational Engineering Research IJCER, Volume 2, Issue No.1 , Jan-Feb 2012, pp. 199-203.
[44] Ching Yee Yong, Kim Mey Chew, NasrulHumaimiMahmood and Ismail Ariffin“Image Processing Tools Package in Medical Imaging in MATLAB”, Internationaljournal of education and information technologies, Volume 6 Issue 3, 2012.
[45] Deepak Kumar Garg ,seema Sharma, swetha Bharadwaj , Diksha Thakur , “ECGPaper Records Digitization through Image Processing Techniques”, InternationalJournal of Computer Applications Volume 48,No.13, June 2012.
[46] ShiruiGao, “Research on Medical Image Processing Method Based on theMATLAB”, Informatics and management science Lecture notes in electricalengineering Volume. 208, 2013, pp. 687-694.
[47] Bhausaheb Shinde*, DnyandeoMhaske, MachindraPatare, A.R. Dani “NoiseDetection and Noise Removal Techniques in Medical Images”.
[48] HeryPURNOMO, Hiroshi, HASEGAWA, Kazuo, SHIGETA and HideyaTAKAHASHI, “Pulse Coupled Neural Network for Identifying the Tuberculosis onHuman Lung” Volume 44, 2003, pp. 23-30.
[49] Juan A. Ramírez-Quintana, Mario I. Chacon-Murguia and Jose F. Chacon-Hinojos,“Artificial Neural Image Processing Applications: A Survey”, Engineering Letters,2009, volume 20, issue 1, pp 68-80.
[50] Dr. N. Ganesan, Dr.K. Venkatesh ,Dr. M. A. Rama ,”Application of NeuralNetworks in Diagnosing Cancer Disease Using Demographic Data”, InternationalJournal of Computer Applications, Volume. 1, No. 26, pp.76-85.
[51] Dilip Roy Chowdhury, MridulaChatterjee& R. K. Samanta, “An Artificial NeuralNetwork Model for Neonatal Disease Diagnosis”, International Journal of ArtificialIntelligence and Expert Systems (IJAE), Volume 2, Issue 3, 2011.
[52] QeetharaKadhim, Al-Shayea, “Artificial Neural Networks in Medical Diagnosis”,“IJCSI International Journal of Computer Science Issues”, Volume 8, Issue 2, ISSN(Online):1694-0814, March 2011.
86
[53] AlaaELEYAN, HasanDemirel , “Co-occurrence matrix and its statistical features asa new approach for face”, Turk journal Electrical Engineering and Computerscience Volume. 19, No 1,2011, pp.97-107.
[54] MussaratYasmin, Muhammad Sharif and SajjadMohsin, “Neural Networks inMedical Imaging Applications: A Survey”, World Applied SciencesJournal,Volume 22,No.12, 2013, pp. 85-96.
[55] Zhenghao Shi and LifengHe , “Application of Neural Networks in Medical ImageProcessing” Proceedings of the Second International Symposium on Networkingand Network Security , 4th April. 2010, pp. 023-026.
[56] Burke and Parnell. “Minimal Pulmonary Tuberculosis” Canadian MedicalAssociation Journals, 59:348.
[57] Steingart K, Henry M, Ng V, et al. "Fluorescence versus conventional sputum smearmicroscopy for tuberculosis: a systematic review". (2006). Lancet InfectDiseasesVolume 6, No.9, pp. 570–581.
[58] Brown M, Varia H, Bassett P, Davidson RN, Wall R, Pasvol G "Prospective studyof sputum induction, gastric washing, and bronchoalveolar lavage for the diagnosisof pulmonary tuberculosis in patients who are unable to expectorate". Clin InfectDis, Volume44 ,No.11, 2007, pp.1415–1420.
[59] Drobniewski F, Caws M, Gibson A, Young D "Modern laboratory diagnosis oftuberculosis". Lancet Infect Disseases, 2003,Volume.3,No.3, pp. 141–147.
[60] Moore D, Evans C, Gilman R, Caviedes L, Coronel J, Vivar A, Sanchez E, PiñedoY, Saravia J, Salazar C, Oberhelman R, Hollm-Delgado M, LaChira D, Escombe A,Friedland J "Microscopic-observation drug-susceptibility assay for the diagnosis ofTB". (2006). N Engl J Med Volume 355, No. (15):2006 1539–50.//www.ncbi.nlm.nih.gov/pmc/articles/PMC1780278/.
[61] Rossi, S. E.; Franquet, T.; Volpacchio, M.; Gimenez, A. Aguilar, G. "Tree-in-BudPattern at Thin-Section CT of the Lungs: Radiologic-Pathologic Overview".Radiographics,Volume 25, No. 3, 2005, pp. 789–801.
[62] CDC - Guidelines for Using the QuantiFERON-TB Gold Test for DetectingMycobacterium tuberculosis Infection, United States.
[63] Dinnes J, Deeks J, Kunst H, Gibson A, Cummins E, Waugh N, Drobniewski F,Lalvani A "A systematic review of rapid diagnostic tests for the detection oftuberculosis infection" Health Technology Assess, Volume 11,No.3, 2007, pp. 1–314.
[64] Plikaytis BD, Plikaytis BB, Shinnick TM. ”Computer-assisted pattern recognitionmodel for the identification of slowly growing mycobacterium includingMycobacterium tuberculosis”. Journal of General Microbiology1992, 138(11).
[65] PatilS.A “Texture analysis of TB X-ray images using image processingtechniques”, Journal of Biomedical and Bioengineering, Volume. 3, Issue 1, 2012,pp.53-56.
87
[66] K. Veropoulos, C. G. Lear month, B. J. Simpson “The Automated Identification ofTubercle Bacilli using Image Processing and Neural Computing Techniques”,Proceedings of the 8th International Conference on Artificial Neural Networks,Skövde, Sweden, 2-4 September 1998, Volume 2, pp. 797-802.
[67] Rachna H. B., M. S. MallikarjunaSwamy,”Detection of Tuberculosis Bacilli usingImage Processing Techniques”, International Journal of Soft Computing andEngineering (IJSCE), Volume. 3, Issue. 4, September 2013.
[68] Ziehl-Neelsen stains P. Sadaphal, J. Rao, G. W. Comstock, M. F. Beg “Imageprocessing techniques for identifying Mycobacterium tuberculosis “, INT JTUBERC LUNG DIS 12(5): © 2008, pp.579–582.
[69] Stefan Jaeger, AlexandrosKarargyris, SemaCandemir, Jenifer Siegelman, Les Folio,Sameer Antani, George Thoma, Clement J. McDona , “Automatic screening fortuberculosis in chest radiographs,”a survey Quantitative imaging in Medicine andSurgery”, Volume 3, No 2, April 2013, pp. 89-99.
[70] Manuel G Forero, Filipsrobek, Gabriel Christobal, “Real-Time Imaging” - Specialissue on imaging in bioinformatics, Part III archive Volume 10 Issue 4, August 2004pp. 251-262.
[71] Ajay Divekar, CorinaPangilinan, Gerrit Coetzee, TarlochanSondh, Fleming Y. M.Lure, and Sean Kennedy, “Automated detection of tuberculosis on sputum smearedslides using stepwise classification”, Proceeding SPIE Medical ImagingConference, 2012, pp.1-9.
[72] Jeannette Chang “Automated Tuberculosis Diagnosis Using Fluorescence Imagesfrom a Mobile Microscope” thesis.