Review on Mammogram Mass Detection by Machine Learning Techniques

Abstract

Abstract-Breast cancer continues to be a significant public health problem in the world and number one cause for death rate in Malaysia. Early detection is the key for improving breast cancer prognosis. Mammography is the most effective tool now available for an early diagnosis of breast cancer. However, the detection of cancer signs in mammograms is a difficult task due to irregular pathological structures and noise which are present in the image. It has been shown that in current breast cancer screenings 8%-20% of the tumors are missed by the radiologists. For this reason, a lot of research is currently being done to develop systems for computer aided detection to improve the accuracy. In this paper, review of mammogram mass detection and segmentation is focused. The main aim of the paper is to summarize and compares the method of mass detection in mammogram images. In specific, preprocessing, segmentation, feature extraction and classifications are discussed, Receiver operating curve and free-response receiver operating curve of each method is highlighted to show the sensitivity and specificity of the tumor detection.

IMCAD: Computer Aided System for Breast Masses Detection based on Immune Recognition

“…They are standard methodologies for measurement of performance of detection and diagnosis algorithms in CAD systems. Raman et al present more details about ROC in [31].…”

Optimized Artificial Neural Network based Digital Mammogram Analysis for Breast Cancer Diagnosis

“…The GLCM (Gray Level Co-occurrence Matrix) is method in which the spatial relationship between pixels of different gray levels is considered [25]. Second-order texture features such as Inertia Autocorrelation, Contrast, Correlation, Dissimilarity Energy and Entropy may be computed using GLCM which gives the complete details of the image.…”

A semi-supervised fuzzy GrowCut algorithm to segment and classify regions of interest of mammographic images

“…In edge-based segmentation, it is difficult to determine the boundary of the tumor due to some ill-defined edges lesions. Region-based segmentation are more suitable for mass detection, since regions of tumor are usually brighter than their surrounding tissue, have an almost uniform density and a fuzzy boundary (Raman et al, 2011).…”

“…However, it can be very difficult to detect the contour of the tumor accurately depending on several factors, such as shape of the tumor, density, size, location and image quality. Some challenges in tumor segmentation include low contrast images, intensity levels which vary greatly across different regions, poor illumination and high noise levels, non-defined contours, and masses which are not always obviously detected (Raman et al, 2011).…”