2017 IEEE International Conference on Wireless Communications, Signal Processing and Networking (WiSPNET 2017)

171 / 2016-12-12 14:13:59

SURF features based classifiers for Mammogram Classification

Mammogram, SURF, Feature descriptor, PreARM, Apriori algorithm, ESAR.

Breast cancer is the second leading cause of cancer-related death among women. Computer-aided diagnosis of mammogram images is essential for early detection of breast cancer. Authors use SURF (Speeded-Up Robust Features) local descriptors to generate feature vector and use different classifier for mammogram classification. SURF is a scale and rotation invariant detector and descriptor. SURF features extracted from region of interest (ROI) of mammogram images are of high dimension, and a large number of features are extracted. So, the features are optimized using PreARM algorithm [1] to get most discriminating features. Optimized SURF feature vectors and the class of training mammograms are used to form transaction database. It is submitted to Apriori algorithm to generate Association rules. Authors use ESAR (Extraction of Strong Association Rules) algorithm [2] to get strong and optimize association rules. Filtered, strong rules are used for classification of mammograms. Proposed scheme is validated on standard MIAS and DDSM medical image data set. Algorithm’s performance is measured in terms of area under receiver operating characteristic (ROC) curve value and classification accuracy. Results of associative classifier are compared with classification using SURF descriptor and distance measure and random forest method. Experimental results revels that SURF outperforms other schemes in-terms of distinctiveness, repeatability, and robustness. SURF is computed and compared much faster by maintaining its performance. For SURF based associative classifier, accuracy values for MIAS and DDSM database are 92.3076% and 96.875% respectively, and area under ROC curve values for MIAS and DDSM database are 0.9535 and 0.9221.