Breast Tumor Detection in Digital Mammogram Based on Efficient Seed Region Growing Segmentation

ABSTRACT

Breast cancer is increasing rapidly in women these days. It is diagnosed after the symptoms appear. The detection at an initial stage and the procedure of the treatment are the most significant strategies curing it effectively. For the determination of breast cancer, breast mammography is required. It can be one of the methods to recognize tumor in women's breasts because it lessens the risk of death by recognizing the tumor at an early level. This paper shows a combination of selection methods for step by step evaluation for Efficient Seeded Region Growing with enhanced performance to spot breast cancer. The suggested method has been divided into the following parts: First, preprocessing of breast image is performed to estimate the automatic extraction of Region of Interest. Second, an automatic threshold is calculated for the binarization process. Third, the numbers of seed points are determined automatically and their positions in the breast image are identified using the density of pixels value. Fourth, a method calculating the threshold value is proposed for the purpose of region creation in seed region growing. The given method is applied and tested on the two publicly available mammogram data sets; Mammographic Image Analysis Society (MIAS) and Digital Database for Screening Mammography (DDSM). It was observed that the proposed algorithm gives 94.6% and 95.3% True Positive Fraction (TPF) in case of MIAS and DDSM images, respectively. The parameters demonstrate that the given algorithm gives better results than the previously available methods.

KEYWORDS:

• Breast tumor
• Image segmentation
• Mammography
• Mammogram images
• Seed region growing
• Seed point
• Threshold

ACKNOWLEDGEMENTS

We would like to thank organizer of Mammographic Image Analysis Society (MIAS) and Digital Database for Screening Mammography (DDSM) for sharing mammograms images with us which helped us to test the performance of our algorithm.

Additional information

Notes on contributors

Neeraj Shrivastava

Neeraj Shrivastava is a PhD research scholar at the Maulana Azad National Institute of Technology, Bhopal. He received the BE in computer science & engineering in 2006, MTech in computer science & engineering in 2009 and currently pursuing PhD. He published more than 30 research papers in international journals/conferences. He also reviews papers in SCI/SCIE Journals. His research interests include image processing, ad-hoc networks and algorithms.

Jyoti Bharti

Jyoti Bharti is a assistant professor at the Maulana Azad National Institute of Technology, Bhopal. She received her PhD degree in image processing in 2012. Her research interests include bioinformatics, image processing, and algorithms. E-mail: [email protected]