Informatics Publishing Limited

S. Meenalosini ¹, J. Janet ², E. Kannan ³

Affiliations
1 Veltech Dr.RR. & Dr.SR Technical University, Chennai, IN
2 Dean of Computing, Veltech Dr.RR. & Dr.SR Technical University, Chennai, IN
3 Dean & Professor, Veltech Dr.RR. & Dr.SR Technical University, Chennai, IN

Abstract

Breast cancer is one of the most dangerous diseases that cause innumerable fatal in the female society. Early detection is the only way to reduce the mortality. Due to variety of factors sometimes manual reading of mammogram results in misdiagnosis. So that the diagnosis rate varies from 65% to 85%. Various Computer Aided Detection techniques have been proposed for the past 20 years. Even then the detection rate is still not high. The proposed method consists of the following steps Preprocessing, Segmentation, Feature extraction and Classification. Noise, Artifact and Pectoral region are removed in preprocessing step. Contrast enhancement, alarm region generation and Region growing method is used to segment the mass region. Segmented features are extracted using Gray Level Co-occurrence Matrix. Extracted features are classified using Support Vector Machine. Performance of the proposed system is evaluated using partest method. Proposed algorithm shows 95.2% sensitivity and 94.4% Specificity. The proposed algorithm is fully automatic and will be helpful in assisting the radiologists to detect the malignancy efficiently.

Keywords

Mammogram, Computer Aided Detection, Adaptive Histogram, Segmentation, Feature Extraction, Support Vector Machine.

Full Text

     

S. Meenalosini ¹, J. Janet ¹

Affiliations
1 Veltech Dr. RR. & Dr. SR Technical University, Chennai, IN

Abstract

This paper presents a hybrid technique which aims to assist radiologist in identifying breast cancer at its earlier stages through mammograms. It is difficult to identify masses in raw mammogram. Hence, in this paper an intelligent system is designed to diagnose breast cancer in mammogram using intelligent techniques such as Gabor filter and gray level co-occurrence matrices. Preprocessing, Segmentation and mass extraction are the three major steps involved in the proposed method. In preprocessing, down sampling and quantization is applied on input mammogram, following it noise removal is efficiently performed using median filter and finally Region of Interest is extracted using histogram matching. In segmentation, a band-pass filter is formed by rotating a 1-D Gaussian filter(off center) in frequency space, termed as-Circular Gaussian Filter (CGF). A CGF can be uniquely characterized by specifying a central frequency and a frequency band. Usually mass appears as a brighter region on a mammogram. Mass region can be segmented out using a threshold that is adaptively decided upon the histogram analysis of the CGF-filtered mammogram. Finally extraction of masses is performed using gray level co-occurrence matrices (GLCM) features. GLCM Features like entropy, contrast and homogeneity is analyzed in order to detect whether extracted region contains masses or normal tissue. Efficiency of the proposed method is calculated by analyzing true positive, true negative and false positive, false negative results. Receiver Operating Characteristics curve method is used to analyze efficiency of the proposed method. Thus, the proposed approach would be helpful for automated real time breast cancer diagnosis.

Keywords

Gabour Filter, Gaussian Filter, Gray Level Co-Occurrence Matrices, Histogram Matching, Mammogram, Masses, Segmentation.

Full Text