Detection of Microcalcifications in Digital Mammogram Using Curvelet Fractal Texture Features
European Journal of Molecular & Clinical Medicine,
2020, Volume 7, Issue 2, Pages 251-256
AbstractIn this work, an attempt is made to Segment and find features of the segmented
mammogram Images and finally mammogram images are classified as normal and
abnormal. The mammogram images used for this work are considered from MIAS
Database. The database includes 322 digitized films and all the images are of size
1024x1024. It consists of 322 images (208 normal images and 114 abnormal images).
Initially, mammogram images are subjected to pre-processing using Discrete Cosine
Transform to enhance the edges of the mammograms. Then, sharpened images are cl
using the Fuzzy C-means Clustering algorithm. After segmentation, Curvelet coefficient
and fractal Dimension values are obtained using Discrete Curvelet Transformand Fractal
textures respectively. The average values of obtained curvelet coefficient and fractal
dimension values for both normal and abnormal mammogram images are compared.
Finally, The mammogram images are classified using an Ensemble Fully Complex-
Valued Relaxation Network Classifier. The Classifier is used foe the classification of the
mammogram images as normal and abnormal.
- Article View: 129
- PDF Download: 226