Enhanced Image Processing Techniques Based On Technical Challenges Of Mammogram Image Characteristics
Loading...
Date
2010-04
Authors
Mohd Nordin, Zailani
Journal Title
Journal ISSN
Volume Title
Publisher
Universiti Sains Malaysia
Abstract
Breast cancer is the leading type of cancer in Malaysia. More than 30% of total
cancer case reported among women in Malaysia is made up of breast cancer. At the
moment, one of the best known techniques used for breast cancer detection is
mammography. Unfortunately, the image produced through mammography is normally
noisy and low in contrast making detection of early signs of breast cancer (i.e.
microcalcification and mass) difficult. Therefore, a lot of studies have been conducted to
develop image processing techniques which would help in the detection of these early
signs. However, most of these techniques were developed without a thorough study
upon the mammogram image technical characteristics. Hence, an image which has been
enhanced from one aspect may end up worst from another aspect. In this study, new
image processing techniques have been developed based on findings which have been
gathered from characterization of mammogram images. The characterization process
covered the analysis of grey level distribution, noise, edge and texture. Detailed
understanding of these characteristics, provide a solid basis for the development of new
image processing techniques in this study. The new techniques brought forth through
this study include contrast enhancement (i.e. Moving Contrast Sweep), noise
suppression (i.e. Mean Approximation Adaptive Wiener Filter) and segmentation
algorithm (i.e. Mean Median Crossing Segmentation). In addition to that, a new edge
detection algorithm (i.e. Delta Variance Edge Detection) has also been developed to
assist in microcalcification detection. Furthermore, a new classification system which is
based on texture characteristic has also been developed for mass detection. Based on the
analysis, it has been found that the Mean Approximation Adaptive Wiener Filter and
Delta Variance Edge Detection techniques perform better than their predecessor. The
performance of Mean Median Crossing Segmentation and Moving Contrast Sweep
techniques is equivalent to existing techniques but they can be implemented through a
more practical approach. The mass detection technique through statistical texture
analysis demonstrates a lot of potential but still has room for improvement. It is hoped
that, the new image processing techniques developed through solid understanding of
mammogram image characteristic in this study could provide a strong foundation for the
development of medical imaging applications in the future.
Description
Keywords
Image processing techniques based on technical Challenges , of mammogram image characteristics