Vision-based guidance system for breast awareness application using RGB-D sensor

Breast cancer is one of the most dominant cancers detected among middle-aged women worldwide. In Southeast Asia, Philippines have the highest incidence rate of breast cancer in fact it ranks at number one in incidence and mortality rates among the common cancer in women. Become aware of cancer early considerably affects its curability. Breast self-examination (BSE) is an inexpensive, non-invasive and non-hazardous procedure which can be performed by women to observe their own breasts and detect any abnormalities thus can resolve the issue of the presence of the lumps which can be cancerous. Without a medical supervision and guidance from a health professional, BSE is complicated to master. In response to this issue, this research intentions to develop computer vision- based guidance system which can monitor and guide women while they perform BSE. The proposed system requires a Microsoft Kinect and a personal computer. The proposed system is comprised of three categories. The first algorithm is concerned with the identification of breast area using extensive use of object detection using feature cascaded classifier. The second algorithm is concerned on finger presence for every palpated areas, using image annotation and training algorithm and feature classifier. The third algorithm is concerned with depth classification and hand tracking in every palpated areas using the Support Vector Machine and Consensus Matching and Tracking. These algorithms were integrated together to create a complete BSE guidance system with a custom graphical user interface with additional features such as visual and tactile inspection and visual feedback. This research aims to develop the ability of women in performing BSE through the use of the proposed system in order to educate breast awareness among women and eventually reduce the breast cancer mortality rate