Similarity Measurement of Breast Cancer Mammographic Images Using Combination of Mesh Distance Fourier Transform and Global Features

<p>Similarity measurement in breast cancer is an important aspect of determining the vulnerability of detected masses based on the previous cases. It is used to retrieve the most similar image for a given mammographic query image from a collection of previously archived images. By analyzing these results, doctors and radiologists can more accurately diagnose early-stage breast cancer and determine the best treatment. The direct result is better prognoses for breast cancer patients. Similarity measurement in images has always been a challenging task in the field of pattern recognition. A widely-adopted strategy in Content-Based Image Retrieval (CBIR) is comparison of local shape-based features of images. Contours summarize the orientations and sizes images, allowing for heuristic approach in measuring similarity between images. Similarly, global features of an image have the ability to generalize the entire object with a single vector which is also an important aspect of CBIR. The main objective of this paper is to enhance the similarity measurement between query images and database images so that the best match is chosen from the database for a particular query image, thus decreasing the chance of false positives. In this paper, a method has been proposed which compares both local and global features of images to determine their similarity. Three image filters are applied to make this comparison. First, we filter using the mesh distance Fourier descriptor (MDFD), which is based on the calculation of local features of the mammographic image. After this filter is applied, we retrieve the five most similar images from the database. Two additional filters are applied to the resulting image set to determine the best match. Experiments show that this proposed method overcomes shortcomings of existing methods, increasing accuracy of matches from 68% to 88%.</p>

Name thesis:degree_name

Master of Science (MS)

Level thesis:degree_level

Thesis - Open Access

Discipline thesis:degree_discipline

Electrical Engineering and Computer Science

Year dc:date.available

2016

Author dc:creator

• Kasaudhan, Ravi

Contributors dc:contributor

• Sung Y. Shin

• mammograms
• similarity measurment
• Computer Sciences

Licence

• http://rightsstatements.org/vocab/InC/1.0/

Language dc:language

en

Repository record dc:identifier

https://openprairie.sdstate.edu/etd/995

OAI identifier oai:identifier

oai:openprairie.sdstate.edu:etd-1995