Computer Aided Detection Algorithm For Digital Mammogram Images - A Survey

International Journal of Computer Trends and Technology (IJCTT) volume 8 number 3 Feb 2014
ISSN: 2231-2803 http://www.ijcttjournal.org Page138

Computer Aided Detection
Algorithm for Digital
Mammogram Images A Survey
Bommeswari Barathi
1
, Siva Kumar.R
2
, Karnan.M
3
1
P.G Scholar, Computer and Communication Engineering, Tamilnadu College of Engineering, Coimbatore,
India.
2
Professor, Dept. of Information Technology, Tamilnadu College of Engineering, Coimbatore, India.
3
Professor and Head. Dept. of Computer Science and Engineering, Tamilnadu College of Engineering,
Coimbatore, India.

ABSTRACT: In worldwide, Breast cancer is one
of the leading disease among the women, under the
age group of 15- 54. An automatic detection of
micro calcifications is performed in magnetic
resonance imaging system. Here discuss about,
preprocessing and enhancement, feature
extraction, segmentation, classification and
analysis steps in the stage of preprocessing and
enhancement, the medical MR images are
enhanced by the computer aided detection
algorithm. Segmentation performed by using K
means clustering algorithm and then feature
extracted by gray scale co-occurrence matrix
(GLCM). Classification is done by support vector
machine. Finally analysis determines using
receiver operating characteristics.
Keywords-Magnetic resonance image (MRI), pre
processing and enhancement, segmentation,
feature extraction, classification.
I. INTRODUCTION
1. Breast Cancer
Breast cancer is one of the most common
disease among women and that lead to causes death
under the age group of 15-54.In 1996, the
American cancer society estimated that 184,300 are
diagnosed by breast cancer, in that around 44,300
are women, Where as another study showed that
approximately 720,000 new cases will be
diagnosed per year which results about 20% of all
malignant tumor cases. The world health
organizations international agency for research on
breast cancer estimates more than 150,000 women
die due to breast cancer each year in world wide.
Since breast cancer is leading disease in world
detection at the right time is crucial. The early
detection will leads to the chance of survival.
Screening mammography is only method that
available in current for early and potentially
curable for breast cancer.

2. Mammography
Mammography is high-resolution x-ray
imaging of the compressed breast which involves
radiation transmi ssi on through the tissue
and the projection of anatomical structures on a
film screen or image sensor. Though the x-ray
imaging projection is a reduction in anatomical
information from a 3D organ to a 2D
film/image. Hence the Two imaging projections
of each breast, cranio caudal (CC) and medio
lateral oblique (MLO) are routinely obtained
which indicates three dimensions to understand
the overlapping structures. High quality
mammogram along with high spatial resolution
through the adequate contrast separation allows the
radio logiest for observing the fine structure. Thus
the both method shows that mortality rate decrease
by 30% of all women age 50 and older has regular
mammograms. Breast cancer usually appears by
distributed ducal structures.
Breast cancer consists of three major
types. They are 1.Circumscribed/oval masses,
2. Spiculated lesions and 3.Microcalcification.The
above three types deals with malignant and benign
lesions. Malignant lesions consist of more irregular
shape than the benign lesions. Circumscribed
masses deal in the form of compact and roughly
elliptical. Spiculated lesions consist of central
tumor mass and it is surrounded by radiating
pattern of linear spicules. Micro calcifications
visible as bright dot spots are the form of clusters.
These represented as calciumdeposits from call
recreation and necrotic cellular debris. Two types
of micro-calcification are, 1. Benign micro-
calcifications have the features of high uniform
density with smooth and sharply outlined. 2.
Malignant micro-calcification visible as irregular
shape and distributed variably.
II. DETECTION
To automatically detect the breast cancer
through MRI, here introduced the computer aided
diagnosis (CAD) system. In computers aided
detection, the receiver operating characteristic

(ROC) is used to evaluate its performance in the
system and resulted in description of detection or
diagnostic approximately. In computer aided
detection, there are two classes are available. They
are, 1) One class is either cancer non abnormal
class (True positive) and 2) The other class is
normal class (True negative).

Table 1- An Overview of Detection Techniques
Methods Remarks
Back propagation neural
network[14]
Used to detect the
micro calcification
clusters and show
the result of 84.3%
Position emission
mammography Camera
[39]
To detect the breast
cancer is the earlier
stage.
Nuclear magnetic moment
vectors [18]
To point the random
directions of the
atoms and align
them direction to the
direction of
magnetic field.
Forward Fourier
transformand inverse
forward transform.[8]
To find the spacing
between the object
and the image. The
shows the overlap
with the results to
each image contain
zero is normal.
Micro-calcification
algorithm, wavelet
transform, fuzzy shell
clustering [15]
Helps to improve the
diagnostic accuracy
and enhances the
quality of
mammograms using
multi reduction
analysis and detect
through the
introduction of shape
information and thus
it helps to canyon
with the
experimental results
to confirm the
activity.
Region of interest
identification fuzzy shell
clustering nodular
extraction. [4]
Used to identify the
region that encircles
with the micro-
calcification clusters
and helps to reduce
the iteration by
analysis the
detection between
the valid micro-
calcification region
and invalid one.
Bayesian network by
directed acyclic graph
[20]
To find the breast
cancer at the early
stage and helps to
take order into an
account and find
their dependencies
relation is the single
model. Hence
obtained the results
are improved by the
approaches results.
Plasmon coupling effects
[23]
To identify the
breast cancer bio-
markers by colour
changes and then
generate to analysis
by using dark field
microscope.
Computer aided diagnosis
schemes. [27]
Used to develop the
detection of primary
signatures of these
diseases along with
masses and micro-
calcification.
III. DATABASE (IMAGE ACQUISITION)
All the mammograms are used to obtain
by the digital database for screening
mammography (DDSM). The DDSM database
comprised of digitized mammograms by associated
ground truth and other information. The main
purpose of database used for providing a large set
of mammograms that are free, which can also be
used by researchers to evaluate along with the
comparison of performance in the computer aided
detection (CAD) Algorithm. The database contains
2620 cases available along with 43 volumes of
each case having four views, (medio lateral,
oblique and cranio caudal views of left and right
breasts).
IV. PREPROCESSING AND
ENHANCEMENT
The stage of preprocessing and
enhancement is the simplest model of medical
image processing. This process helps to reduce the
noise and improves the quality of digital images.
1. Preprocessing
Preprocessing helps to identify different
scale of signal intensities of different images.
Preprocessing function involve the operation to
analysis the data and extract the information the
preprocessing techniques such as adaptive filter,
histogram equalization, weighted K mean
clustering are compared.
Table 2 - An Overview of Preprocessing
Techniques
Methods Remarks
Fuzzy c means [32] To remove the diffusion.
Contrast enhanced Capacity to correlate the

MRI [11] features is
mammograms by
enhancing the regions
and Used to classify the
3D distribution of
micro-calcification
structures and also
correlates to the
enhancement
characteristics of ducts.
Computer aided
detection Algorithm
[17]
Used to collect the
resources to provide
large set of
mammograms is the
digital format
Computer image
processing techniques
[42]
Used to enhance the
images and leads to
segmentation by using
the region of interest.
Adaptive filter,
histogram
equalization,
histogrammodified
local contrast
enhancement [34]
Used to improve the
quality of images by
removing the noise and
then enhances by
adjusting the value of
parameter
Gaussian smoothing
with median filtering
[9]
Used to enhance the
images with high quality
of demonising
2. Enhancement
Enhancement methods used to improve
the visual appearance of images from magnetic
resources (CT) scan, positron emission tomography
(PET) and contrast enhancing based on height of
interesting tissue such as fibrous or glandular
tissues and cancerous tissues, where it allows for
more accurate analysis of the mammograms.
Enhancement techniques like histogram modified
local contrast enhancement, local range
modification and discrete wavelet are used.
Table 3 - An Overview of Enhancement
Techniques
Methods Remarks
K mean and fuzzy c
mean algorithm[32]
To improve the
performance in noisy
images and also to
result in terms of
speed, robustness and
accuracy.
Conceptual geometry
projection process [11]
Helps to biased to
voxels with high
enhancement to
produce the projection
indemnity
Pharmacokinetic model
[4]
Used to chosen the
actual enhancement
characteristics
Micro calcification Helps to improve the
algorithmwavelet
transform, fuzzy shell
clustering [5]
diagnostic accuracy
and enhances the
quality of
mammograms using
multi reduction
analysis and detect
through the
introduction of shape
information and thus it
helps to canyon with
the experimental results
to confirmthe activity
Gray scale histogram
equalization, wavelet
transform, digital
database screening
mammograms [7]
Used to construct the
images by representing
the indemnity values of
pixels based on the
image histogram and
modifies with visual
artifacts and then
decompose the images
with high level of
reductions
Region of interest
identification, fuzzy
shell clustering nodular
extraction [8]
with the micro
and helps to reduce the
iteration by analysis the
detection between the
valid micro
calcification region and
invalid one. Thus
among images 95%
micro calcification are
detected correctly and
5% were failed to find,
though they are not
nodular in structure.
Therefore concluded
the better way to detect
micro calcification
without nodular
structure.
Histogram
modification [48]
To control over the
level of enhancement
that deals with the
adjustment of features.
Hence the work
extended to test the
presence of
mammogramimages in
the presence of noise
and can be employed as
a pre processing step
for analysis of micro
calcification in
mammograms
Haar wavelet transform
[28]
Helps to the accelerate
the signals of damaged

primitive feature using
wavelet co efficient for
single degree of
freedomsystem.
Computer assisted
diagnosis software [16]
Used to examine the
mammograms. Hence
helps to detect the
small calcium deposits
and there by enhances
the quality of image
with high reductions
Histogrambased
method, effective multi
peak generalised
histogramequalisation
[3]
Used to segment the
image and compose
with the original
images and helps locate
the peaks.
Local range
modification
redundant discrete
wavelet linear
stretching and
shrinkage algorithm
[10]
Used to visible the
small calciumdeposits
Incremental fuzzy
mining techniques [37]
Used to solve the
complexities among the
higher dimensional
noisy data.
Particle swarm
optimization [25]
Used to solve the
problem influenced by
the dimension and
enhances the swarm
size, neighbourhood,
acceleration co
efficient and random
values
Markov random field
[35]
Used to filter the noise
and enhance the
iteration through the
clustering process and
thus result with =0.5
on the noise.
Anisotropic diffusion
and weighted k means
clustering [30]
Enhancing the contrast
among the images by
reducing the noise.
Pixel intensity
transformation, spatial
filtering, log arithmetic
transformation, local
and global threshold
techniques , contrast
stretching, histogram
equalization [15]
Used to segment the
transformation among
the MR images and
extract the features by
filtering from unwanted
signal noise and
distortion

V. SEGMENTATION
In the stage of segmentation, the images
are partition or separated into region of similar
attribute or characteristics. The main aimto extract
the features fromthe images data, through which a
description, interpretation or understanding the
features. The images can be segmented based on
the intensity, region, and threshold values. The
usage of K- mean algorithm is simplest elegant
segmentation method and it deals with the
unsupervised clustering algorithm that used to
classifies the input data points into multiple classes
depended on their distances. Thus the high pass
sharpened images are segmented to provide
thresholds the image in four of the common
methods used to perform the regimentation
methods are,1) Fuzzy logic, 2) Expectation
Algorithm, 3) Vector quantization,4) K means
clustering, 5) Matching property. It is issued to
segment the image and then proceed to detect the
part, thereby helps to collect information fromthe
image and to achieve the segmentation method.
Table 4 - An Overview of Segmentation
Techniques
Methods Remarks
Functional MRI,cardiac
MRI, magnetic
resonance
angiography[32]
Used to segment the
images that can be
used is various
diagnostics
Computational
intelligence [9]
Efficiently used for
segmentation
Fuzzy logic, expectation
maximization algorithm
[3]
To collect information
from the image and to
achieve the
segmentation method.
Ant colony optimization
algorithm[8]
To solve the
optimization problem
and segment the MR
images.
Vector quantization, k
mean clustering,
matching property
[27][6]
Used to segment the
images and then
proceed to the
detection part.
Gray level co
occurrence matrix,
water shed algorithm
kekres median
codebook generation
algorithm[17]
Used to segment the
image into the equal
exact size and then
generate the desired
size of the resulted
rate is 68.5%.
Fuzzy c mean
segmentation methods,
kernel function based on
FCM methods [16]
Used to segment the
breast regions by using
breast MRI images
and helps to cover the
entire images.
Computer image
processing techniques
[42]
Used to enhance the
images and leads to
segmentation by using
the regions of interest.
Region based methods,
edge based methods [3]
Used to separate the
region by locating the
boundaries from

different regions and
also grouping by
individual pixels.
Laplacian kernel
method [2]
Helps to locate the
edges through regions.
Image processing
threshold, edge based
and watershed
algorithm[33]
Used to segment the
mammogram breast
cancer image based on
time consuming and
simplicity
K mean clustering [6] Used to partition the
cluster
Threshold techniques,
hybrid techniques [34]
Used to postulate all
pixel values that lies in
the certain range.
Fuzzy c means along
with feature extraction
techniques and texture
based segmentation [37]
Used to segment into
three features such as
entry, standards
deviation and number
of pixels.
Bayesian maximum
likelihood classifier [21]
Used to slice the
regions and helps to
interactive with
thresholding values
and leads to the
membership value less
than the threshold
value of 0.8.
Granulometry
normalized size
destruction [35]
Used to pattern the
spectrum in the MR
images
Wavelet transformation
and k mean clustering
using intensity based
segmentation and
expectation maximum
algorithm[45]
Used to divided based
on the intensity value
and helps to sharpen
the images and
validates to detect
tumour region froman
MRI images
Soft computing
approaches using edge
based techniques [43]
Process of partitioning
takes place and among
the multiple regions or
set of pixels.
Multiphase level set
approach [38]
Used to minimize the
set of functions
towards the desired
image features like
object boundaries
Fuzzy c mean clustering
algorithmgeneralized
with objective functions
least squares [11]
Used to segment the
aggregate clusters by
using three norms such
as diagonal,
mahalonobis,
Euclidean and
adjusting the weighted
factor.
Wards clustering
methods, fuzzy k mean
clustering with
membership function
Clusters segments
small sets of
compound to highlight
and identify the multi
[24] cluster membership
along with the outlier
compounds.
Pixel intensity
transformation, spatial
filtering, log arithmetic
transformation, local
and global threshold
techniques, contrast
stretching, histogram
equalization [15]
Used to segment the
transformation among
the MR images and
extract the features by
filtering from
unwanted signal, noise
and disortation.
Kekres median code
book generation and
kekres fast code book
generation [13]
Used to segment only
the proper tumour spot
in the MR images.
VI. FEATURE EXTRACTION
Feature extraction is main important step
in which ultimate the performance of the systemis
determined by optimal parameter of the classifier,
to intrinsic separability of the feature vectors. The
attributes of the normal poses the real challenges
due to the complexity of normal tissues fact in the
normal mammograms is not well defined. Hence
breast cancer cannot be easily distinguished
through the surrounding of normal tissues, only the
heterogeneous nature of different breast cancer of
different size poses the real challenges to extract
the features. Thus the techniques that are used to
separate the normal and abnormal region by using.
1) Curvilinear features, 2) Gabor features, 3) Gray
level co-occurrences features.

Table 5 - An Overview of Feature Extraction
Techniques
Methods Remarks
Wavelet decomposition
and artificial neural
networks[14]
To extract the Gabor
feature from the
original image(region
of interest) and shows
the result of 84.3%
Conceptual geometry
projection process[11]
Helps to map to the
intensity plane.
Contrast enhanced
MRI[1]
Capacity to correlate
the feature is
mammograms by
enhancing the regions.
Region of interest
identification. Fuzzy
shell clustering nodular
extraction.[5]
with the micro
and helps to reduce the
iteration by analysis the
detection between the
valid micro
calcification region and
invalid one. Thus
among images 95% of
micro calcification are

detected correctly and
5% were failed to find,
through they are not
nodular in structure.
Therefore concluded
the better way to detect
micro calcification.
Haar wavelet
transform.[28]
Here, fromthe damage
sensitive feature, the
vibration of signal is
derived and shows that
energies of wavelet
coefficients at the
accurate scales can be
referred as damaged
features.
Gray level co
occurrence matrix,
water shed algorithm
kekres median
codebook generation
algorithm. [27]
Used to extract the
needed size of images
that have to classify
and analysis. Hence the
resulted rate is 68.5%.
Genetic algorithm, ant
colony optimization
algorithm, special gray
level dependence
method [26]
Used to select the
feature and perform the
comparison and then
extraction is done.
Contourlet transform
with laplacian
pyramidal filter bank
and steerable Gaussian
filter.[2]
Helps to extract the
images and to identify
the distortion and used
to detect 1502 regions
of interests of 247 true
positives and 1255
false positives. Hence
results are obtained as
0.77 with 26 textural
features.
Haralicks feature with
gray land co
occurrence matrix.[14]
Helps to produce the
changes in the intensity
values of histogram as
the function of distance
and direction.
Gabor and haralick
features along with the
gray level co-
occurrence matrix.[9]
Used to extract the
features for developing
the image signature
over the measurement
of similarity and thus
results in averaging the
precision value
between o.5 and 0.61
using haralick features,
0.49 and 0.57 using
Gabor features and o.51
and 0.78 using
combination of Gabor
and haralick features.
Correlative feature
analysis along with full
field digital
Used to extract the
features from the
segment lesions.
mammography and
dynamic contrast
enhanced magnetic
resonance imaging [22]
Bio-inspired
algorithm[50]
Used to extract the
feature by solving the
optimization problems.
Finite Gaussian
mixture model[36]
Used to get the external
parameter and
threshold valued based
on the segmented
images that yields
better results.
VII. CLASSIFICATION
Classification method is used to separate
the image to get the result of normal or abnormal
regions. After segmentation of the mammogram
images and classify themas benign, malignant or
normal, thereby helps to predict the texture features
which plays a vital role for classification. The size
and stages of the cancer is detected and calculated.
The classification techniques are used as follows.
1) Decision tree, 2) Support vector machine, 3)
Neural Networks, 4) Bayesian learning are
discussed.
Table 6 - An Overview of Classification
Techniques
Methods Remarks
Back propagation
neural network[14]
Used to classify the
features that one extract
as malignant being or
normal and shows the
result of 84.3%
Principles of
component analysis,
support vector
machines[32]
To solve the pattern
recognition problems
K nearest
neighbours[18]
Used in non parametric
methods to find the
specific image fromthe
density model. This
shows the overlap with
the results to each image
contain zero (that is
normal) or more lesions
(that is abnormal)
Contrast enhance
MRI[11]
Used to classify the 3D
distribution of micro
calcification structures
and also correlates to the
enhancement
characteristics of ducal
carcinoma.
Back propagation
network hybrid with
act colony
Selected feature are fed
to the layer of neural to
performclassification

optimization [26]
Clustering
methods[16]
Used to divides the
cluster and helps to
compute the average of
the pixels.
Artificial neural
network with
roughness measures,
haralicks measures,
laws measures[2]
Used to classify the
architectural disortation
AD0 and non AD
images.
Seeded region
growing, split and
merge techniques[34]
Used to split the images
to satisfy the
homogeneity criterion
and postulate the
similarity of pixels
among the regions
Dual three complex
wavelet transform
pyramid structured
wavelet transform,
quad tree
decomposition ,
current transform,
discrete wavelet
transform[3]
Used to subsets
multidimensional data
accuracy and then maps
the best set of clusters
into 2D for visualization

Support vector
machine[49]
To solve the problem of
linearly separable,
binary classification
problem
Computer aided
diagnosis, linear
support vector
machines, area under
the receiver operating
characteristic
curve[29]
Classified unimodal
malignancy to estimate
the multimodal by
averaging and hence
successfully resulted is
classification and
comprising 39% of
original feature sets.
Artificial neural
network, genetic
algorithm, decision
tree, fuzzy c mean[21]
Each algorithmare used
to classify accurately the
positive predictive
values and used to
evaluate the data
analysis. Thus result
with appropriate where
fuzzy c mean was 0.953,
decision tree model were
0.9634, artificial neural
network were 0.96502,
genetic algorithmmodel
were sensivities 0.9878.
Decision tree
algorithmwith
quinlans ID3
C4.5,C5[37]
Used to select the
variable accounting to
the specific criterion
such as information
gain, gini index and chi
squared test. Thus
results with fitness value
of 1, accuracy are train
data as 0.993 for 453
cases and test data as
0.9878 for 246 cases.
Classification and
regression tree
algorithm[40]
Uses the historical data
to construct the decision
tree which helps to
classify the information
as data set to get new
observations
Bayesian artificial
neural network[22]
Depending on the
probability of
correspondence values
the features sets are
divided into subsets.
Modified fuzzy c
mean classification
algorithm[35]
Used to divide the fuzzy
partition
Conventional fuzzy k
mean clustering
algorithm[10]
Helps to classify the
cluster centres as active
and stable groups and
resulting calculating by
38.9% to 86.5% using
the same data sets.
VIII. ANALYSIS
The purpose of analysis stage is used to
create the methods to get the accurate results.
Hence analysis can be computed by building
software for MR image. The techniques that are
used to analysis are, 1) Region of interest - Which
used to compute feature values with the regions
and helps to convert the text format into digital
format and 2)Feature maps Through which are
masked by using non linear filtration.

Table 7 - An Overview of Analysis
Techniques
Methods Remarks
Computer aided
diagnostics[18][51]
Used to process and
analyse the volume of
images to collect the
high quality
information and
whether the disease is
diagnose and
treatment starts for the
patients. Thus shows
the overlap with the
results to each image
certain zero (that is
normal) or more
lesions (that is
abnormal).
Fourier theory[18] To find out the shaped
area of the image.
This shows the
overlap with the
results to each image
contain zero (that is
normal) or more

lesions (that is
abnormal).
Pharmacokinetic
model[11]
Used to chosen the
actual voxel
enhancement
characteristics and
also map the
projection intensities
for each voxel.
Histogrammodified
contrast limited adaptive
histogramequalization
[48]
to determine the
parameter like
enhancement
measures and thus
result to provide the
better contrast
enhancement with
preservation of local
information in the
mammogram images.
Histogram
modification[10]
To control over the
level of enhancement
that deals with the
adjustment of
features. Hence the
work extended to test
the presence of
mammogram images
in the presence of
noise and can be
employed as a pre-
processing step for
analysis of micro
calcification in
mammograms.
Haar wavelet
transform[28]
From the extraction at
the higher scales are
functioned with the
physical parameter to
analysis the image
along with the
vibration of vector.
Back propagation
network hybrid with ant
colony optimization[26]
To evaluate the
performance of
feature selection with
the classified results.
Automated system[42] By extract fromROI
which helps to
analysis the digital
mammography.
Morphological filter
using top hat
transform[2]
Used to analysis
through processing of
geometrical
structures.
Fuzzy c mean
algorithm[33][52]
Used to perform
pattern recognition
and then helps to
analysis the data
between two or more
clusters.
Gabor wavelet with
Gabor filter karhyunen-
loeve transformotsus
method[34]
Used to extract the
elementary function
from the input
features and helps to
analysis the different
scales and orientation.
Genetic algorithm[21] Used for detecting
and diagnosing micro
calcification pattern in
digital
mammography.
Biologically inspired
algorithmwith swarm
filter[19]
Used to focus on the
behaviour collects the
require task with
physical derived or
adhoc-model. Then
start analysis by using
swarmhunting on the
system performing
into gray scale values
and represent the
output signal.
Computer aided
diagnosis (CADx) along
with correlative feature
analysis (CFA)[20]
Used to analysis and
result in improved
diagnostic accuracy.
Moving particle semi
implicit method
(MPS)[41]
Used to analysis the
deformation tissues in
the breast.
Image histogram, co
occurrence and run
length matrices, image
gradient, auto regressive
model and wavelet
transform[48]
Used to calculate the
features and helps to
analysis the image
through textures
Geometric analysis[50] Used to convert 2D to
3D model and then
analyze the images.
Mathematical
morphology[12]
Performance of
analysis takes place
among the pixels
along with the
neighbours
Clustering algorithm[47] Analyzed based on
their clustering
efficiency
CAD approach including
fuzzy logic, neural
network and hybrid
algorithm[44][52]
Used to analyze the
pattern recognition for
the medical images.
Enhanced artificial bee
colony optimization
algorithm.[46]
suspected regions
based on bilateral
subtraction between
the left and right
breast images.


IX. CONCLUSION
In this survey paper, method of detecting
breast cancer by using magnetic resonance imaging
has been studied and discussed. This helps for
focusing the future develops in the field of medical
image processing. Here detailed explanation about
several techniques of medical image processing in
breast cancer detection. This paper mainly deals
with the information about the various stages of
detection such as (1)pre-processing and
enhancement technique, (2) segmentation(3)
feature extraction, (4) classification, (5) Analysis
are studied and explained.
REFERENCES
[1] Ali Qusay Al-Faris, Umi KalthumNgah, Nor Ashidi Mat
Isa, Ibrahim Lutfi Shuaib, Breast MRI Tumor
Segmentation using Modified Automatic Seeded Region
Growing Based on Particle Swarm Optimization Image
Clustering, Imaging and Computational Intelligence
Research Group (ICI), Universities Sains Malaysia, 2012.
[2] Anand S , Aynesh Vijaya Rathna R,Architectural
Distortion Detection in Mammogram using Contourlet
Transformand Texture Features, International Journal of
Computer Applications (0975 8887) Volume 74 No.5,
J uly 2013.
[3] Andrzej Materka, Magnetic Resonance Image analysis
through soft-computing: Visualization Segmentation and
diagnostic tool for Magnetic Resonance images,
Developed Within the Framework of The European COST
(Cooperation in the Field of Scientific and Technical
Research), 2004.
[4] Angayarkanni, Pitchumani S, Kamal, Nadira Banu,
MRI Mammogram Image Classification Using ID3
Algorithm, Published in Image Processing,2012.
[5] Balakumaran T, Vennila, Gowri Shankar C,Detection
of Microcalcification in Mammograms Using Wavelet
Transform and Fuzzy Shell Clustering, International
Journal of Computer Science and Information Security,
Vol. 7, No. 1, 2010.
[6] Bhagwati Charan Patel, Sinha.G R, An Adaptive K-
means Clustering Algorithm for Breast Image
Segmentation, International Journal of Computer
Applications (0975 8887) Volume 10 no.4, November
2010.
[7] Bharati R Jipkate, Gohokar V V, A Comparative
Analysis of Fuzzy C-Means clustering, international
Journal of Advanced Science and Technology Vol. 19, J une,
2008.
[8] Chattopadhyay S, Pratihar D K, De Sarkar S C, A
Comparative Study Of Fuzzy C-Means Algorithm And
Entropy-Based Fuzzy Clustering Algorithms published in
Computing and Informatics, Vol. 30, pp 701720,2011.
[9] Chikamai K, Viriri S, Tapamo J R, Combining Feature
Methods for Content-Based Classification of Mammogram
Images, International journal of computer and
communication, ISSN 1841-9836 8(4):499-513, August
2013.
[10] Chit-tang chang, J imZ C Lai, Mu-derjeng, A Fuzzy
K-means Clustering Algorithm Using Cluster Center
Displacement, journal of information science and
engineering 27, 995-1009, 2011.
[11] Christian P Behrenbruch, Kostas Marias, Paul A
Armitage, Michael Brady J, J ane Clarke, Niall Moore, The
Generation of Simulated Mammograms from Contrast
Enhanced MRI for Surgical Planning and Postoperative
Assessment, Magnetic Resonance Imaging Centre, John
Radcliffe Hospital, Headley Way, Oxford OX3 9DU, UK,
2000.
[12] Christina Olsen, Image Analysis Mathematical
Morphology, Department of Computing Science Umea
University, Feb 2009.
[13] Clifton C, vaidya J , K Means Clustering Algorithms,
International Journal Computational Engineering Research
/ ISSN: 22503005, 2012.
[14] Dheeba J , Wiselin J iji G Detection of
Microcalcification Clusters in Mammograms using Neural
Network, international Journal of Advanced Science and
Technology Vol. 19, J une 2010.
[15] Gandhi K R, Karnan M, Mammogram image
enhancement and Segmentation, Computational
Intelligence and Computing Research (ICCIC), 2010.
[16] Greeshma Gopal, Grace Mary Kanaga E, A Study on
Enhancement Techniques for Mammogram Images,
International Journal of Advanced Research in Electronics
and Communication Engineering (IJARECE) Volume 2,
Issue 1, J anuary 2013.
[17] Hongmei Zhu, Medical Image Processing Overview,
University of Calgary, 2003.
[18] Horia Mihail H Teodorescu and David J Malan, Swarm
Filtering Procedure and Application to MRI
Mammography, Harvard University, USA, Manuscript
accepted for publication, J uly 2010.
[19] J adid Abdulkadir, Suet-Peng Yong, Oi Mean Foong,
Variants of Particle Swarm Optimization in Enhancing
Artificial Neural Networks, Australian Journal of Basic
and Applied Sciences, 7(9): 388-400, ISSN 1991-8178,
2013.
[20] J iyun Park, J ong-Souk Yeo, Nanoplasmonic-based
Colorimetric Detection of MicroRNA miR-21 in Breast
Cancer Cells, Biomarkers, Vol. 13, No.7-8, 2008.
[21] J ohn D Holliday, Sarah L Rodgers, Peter Willett,
Clustering Files of Chemical Structures Using the Fuzzy
k-Means Clustering Method, Krebs Institute for Bio-
molecular Research and Department of Information
Studies, University of Sheffield, Western Bank, Sheffield
S10 2TN, U.K, November 2003.
[22] Kaabouch N, Wen Chen Hu, A Survey Of Medical
Imaging Techniques used For Breast Cancer Detection,
Published In Electro/Information Technology (EIT), 2013.
[23] Karnan M, Siva Kumar R, Almelumangai M,
Selvanayagi K and Logeswari T, Hybrid Particle Swarm
Optimization for Automatically Detect the Breast Border
and Nipple position to Identify the Suspicious Regions on
Digital Mammograms Based on Asymmetries,
International Journal of Soft Computing 3 (3): 220-
223,2008.
[24] Karnan M, Thangavel K , Automatic detection of the
breast border and nipple position on digital mammograms
using genetic algorithm for asymmetry approach to
detection of microcalcifications, Computer methods and
programs in biomedicine 87 (1), 12-20, 2007.
[25] Karnan M, Thangavel K, Siva Kumar R, Geetha K, Ant
colony Optimization for Feature Selection and
Classification of Microcalcifications in Digital
Mammograms, International Conference on Advanced
Computing and Communications,2006.
[26] Karnan M, Thangavel K , Siva Kumar R ,Geetha K ,
Ant colony Optimization for Feature Selection and
Classification of Microcalcifications in Digital
Mammograms, Advanced Computing and
Communications, 2006.
[27] Kekre H B, Tanuja Sarode, SayleeGharge, Kavita Raut,
Detection of Cancer Using Vector Quantization for
Segmentation, International Journal of Computer
Applications (0975 8887) Volume 4 No.9, August 2010.
[28] Krishnan Nair K, J ohn A Blume , Anne S Kiremidjian,
Derivation of a Damage Sensitive Feature Using the Haar
Wavelet Transform, International Workshop on Structural
Health Monitoring, Stanford University, Stanford,
CA/CRC, New York,2010.
[29] Lesniak J M, Van Schie G, Tanner.C, Plate B, Huisman
H, Karssemeijer N, Multimodal Classification of Breast

Masses in Mammography and MRI Using Unimodal
Feature Selection and Decision Fusion, Radboud
University Nijmegen Medical Centre, Department of
Radiology, The Netherlands, 2007.
[30] Manojkumar S, Vilas Thakare, MRI Image Processing
with Intelligence: Step towards Computer Aided Diagnostic
(CAD) in HealthcareSystems International Conference on
Software and Computer Applications (ICSCA) vol. 41 2012.
[31] Nalini Singh, Ambarish G Mohapatra, Gurukalyan
Kanungo, Breast Cancer Mass Detection in Mammograms
using K-means and Fuzzy C-means Clustering,
International Journal Of Computer Applications(0975-
8887),Volume 22 -No.2,May 2011.
[32] Narain Ponraj D, Evangelin J enifer M, Poongodi P,
Samuel Manoharan J , A Survey on the Preprocessing
Techniques of Mammogramfor the Detection of Breast
Cancer, Journal of Emerging Trends in Computing and
Information Sciences , ISSN 2079-8407,Vol. 2, No. 12,
December 2011.
[33] Pradeep Kumar, Rajat Chaudhary, Ambika Aggarwal,
Prem Singh, Ravi Tomar , Improving Medical Image
Segmentation Techniques Using Multiphase Level Set
Approach Via Bias Correction, International Journal of
Engineering and Advanced Technology (IJEAT) ISSN:
2249 8958, Volume-1, Issue-5, J une 2012.
[34] Radhika Siva Ramakrishna Imaging Techniques
Alternative to Mammography for Early Detection of Breast
Cancer Technology in Cancer Research & Treatment ISSN
1533-0346 Volume 4, Number 1, February (2005)
Adenine Press, 2005.
[35] Rashedi E, Nezamabadi-Pour H, Saryazdi S, GSA: A
gravitational search algorithm, Information Science, 179
(13), 2232-2248, 2009.
[36] Roman Timofeev, Wolfgang Hardle, Classification and
Regression Trees (CART) Theory and Its Applications, A
Master Thesis Presented by CASE - Center of Applied
Statistics and Economics,Dec 2004.
[37] Samir Kumar Bandyopadhyay, Pre-processing of
Mammogram I mages, International Journal of
Engineering Science and Technology Vol. 2(11), 2010.
[38] Senthil kumaran N, Rajesh R, Edge Detection
Techniques for Image Segmentation A Survey of Soft
Computing Approaches, International Journal of Recent
Trends in Engineering, Vol. 1, No. 2, May 2009.
[39] Shruti Dalmiya, Avijit Dasgupta, Soumya Kanti Datta,
Application of Wavelet based K-means Algorithm in
Mammogram Segmentation, International Journal of
Computer Applications (0975 8887) Volume 52 No.15,
August 2012.
[40] Siva Kumar R, Marcus Karnan, Diagnose Breast
Cancer through Mammograms Using EABCO Algorithm,
International Conference and workshop on Emerging
Trends in Technology (ICWET), 2011.
[41] Soumi Ghosh, Sanjay Kumar Dubey, Comparative
Analysis of K-Means and Fuzzy C-Means Algorithms,
(IJACSA) International Journal of Advanced Computer
Science and Applications, Vol. 4, No.4, 2013.
[42] Subash Chandra Bose J , Kumar KRS, Karnan M,
Detection of Microcalcification in Mammograms using
Soft Computing Techniques, European Journal of
Scientific Research 86 (1), 103-122, 2012.
[43] Sundarami M, Ramar K, Arumugami N, Prabini G,
HistogramBased Contrast Enhancement For Mammogram
Images, International Conference on Signal Processing,
Communication, Computing and Networking Technologies
(ICSCCN),2011.
[44] Thangavel K, Karnan M, Automatic Detection of
Asymmetries in Mammograms Using Genetic Algorithm,
International Journal on Artificial Intelligence and
Machine Learning 5 (3), 55-62, 2005
[45] Thangavel K, Karnan M, CAD systemfor Preprocessing
and Enhancement of Digital Mammograms,
International Journal on Graphics Vision and Image
Processing 5 (9), 69-74, 2005.
[46] Thangavel K, Karnan M, Computer aided diagnosis in
digital mammograms: detection of microcalcifications by
meta heuristic algorithms, GVIP Journal 5 (7), 41-55,
2005.
[47] Thangavel K, Karnan M, Pethalakshmi A, Performance
analysis of rough reduce algorithms in mammogram,
International Journal on Global Vision and Image
Processing 5 (8), 13-21, 2005.
[48] Thangavel K, Karnan M, Siva Kumar R, KajaMohideen
A, Ant Colony System for Segmentation and
Classification of Microcalcification in Mammograms,
International Journal on Artificial Intelligence and
Machine Learning, Vol 5, Issue 3, Pages 29-40,2005.
[49] Thangavel K, Karnan M, Siva Kumar R, KajaMohideen
A, Automatic Detection of Microcalcification in
Mammograms A Review, ICGST International Journal
on Graphics, Vision and Image Processing, 2005.
[50] Thangavel K, Karnan M, Siva Kumar R, Mohideen AK,
Segmentation and classification of microcalcification in
mammograms using the ant colony system, International
Journal on Artificial Intelligence and Machine Learning, 5
(3), 29-40, 2005.

Computer Aided Detection Algorithm For Digital Mammogram Images - A Survey

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Computer Aided Detection Algorithm For Digital Mammogram Images - A Survey

Transféré par

Droits d'auteur :

Formats disponibles

International Journal of Computer Trends and Technology (IJCTT) volume 8 number 3 Feb 2014

ISSN: 2231-2803 http://www.ijcttjournal.org Page138

Vous aimerez peut-être aussi