Image Processing Reference
Andersson, Bjurstam, Frisell, Nordenskjold, and Rutqvist, 2002). Breast screening aims to
detect breast cancers at the very early stage (before lymph node dissemination). Random-
ized trials of mammographic screening have provided strong evidence that early diagnosis
and treatment of breast cancer reduce breast cancer mortality (Nystrom et al., 2002).
Breast cancer usually presents with a simple feature or a combination of the following fea-
tures: a mass, associated calcifications, architectural distortion, asymmetry of architecture,
breast density or duct dilation and skin or nipple changes (Nystrom et al., 2002; Rajendra
et al., 2008).
In fact, a large number of mammogram image analysis systems have been employed for
assisting physicians in the early detection of breast cancers on mammograms (Guo, Shao,
and Ruiz, 2009; Ikedo, Morita, Fukuoka, Hara, Lee, Fujita, Takada, and Endo, 2009). The
earlier a tumor is detected, the better the prognosis. Usually, breast cancer detection sys-
tem starts with preprocessing that includes digitization of the mammograms with different
sampling and quantization rates. Then, the regions of interest selected from the digitized
mammogram are enhanced. The segmentation process is designed to find suspicious areas,
and to separate the suspicious areas from the background that will be used for extracting
features of suspicious regions. In the feature selection process, the features of suspicious
areas will be extracted and selected, and suspicious regions will be classified into two classes:
cancer or non cancer (Aboul Ella, Ali, and Hajime, 2004; Aboul Ella, 2003; Setiono, 2000;
Rajendra et al., 2008; Ikedo et al., 2009; Maglogiannis, Zafiropoulos, and Anagnostopoulos,
Rough set theory (Aboul Ella et al., 2004; Hirano and Tsumoto, 2005; Pawlak, 1982) is
a fairly new intelligent technique that has been applied to the medical domain and is used
for the discovery of data dependencies, evaluates the importance of attributes, discovers
the patterns of data, reduces all redundant objects and attributes, and seeks the minimum
subset of attributes. Moreover, it is being used for the extraction of rules from databases.
One advantage of the rough set is the creation of readable if-then rules. Such rules have
a potential to reveal new patterns in the data material. Other approaches like case based
reasoning and decision trees (Aboul Ella, 2003; Slezak, 2000; Aboul Ella, 2009) are also
widely used to solve data analysis problems. Each one of these techniques has its own
properties and features including their ability of finding important rules and information
that could be useful for data classification. Unlike other intelligent systems, rough set
analysis requires no external parameters and uses only the information presented in the
The combination or integration of more distinct methodologies can be done in any form,
either by a modular integration of two or more intelligent methodologies, which maintains
the identity of each methodology, or by integrating one methodology into another, or by
transforming the knowledge representation in one methodology into another form of rep-
resentation, characteristic to another methodology. Neural networks and rough sets are
widely used for classification and rule generation (Greco, Inuiguchi, and Slowinski, 2006;
Aboul Ella, 2007; Henry and Peters, 1996; Peters, Liting, and Ramanna, 2001; Peters,
Skowron, Liting, and Ramanna, 2000; Sandeep and Rene, 2006; Aboul Ella, 2009).
Instead of solving a problem using a single intelligent technique such as neural networks,
rough sets, or fuzzy image processing alone, the proposed approach in this chapter is to
integrate the three computational intelligence techniques (forming a hybrid classification
method) to reduce their weaknesses and increase their strengths. An application of breast
cancer imaging has been chosen to test the ability and accuracy of a hybrid approach in
classifying breast cancer images into two outcomes: malignant cancer or benign cancer.
This chapter introduces a rough hybrid approach to detecting and classifying breast cancer
images into two outcomes: cancer or non-cancer.