Monocular depth level estimation for breast self-examination (BSE) using RGBD BSE dataset
John Anthony C. Jose, Melvin K. Cabatuan, Robert Kerwin C. Billones, Elmer P. Dadios, Laurence A. Gan Lim · 2015
Up until now, there had been no existing literature in depth level estimation algorithm for BSE using a simple camera that provides quantitative accuracy. They can only show their effectiveness thru graphs. In this paper, we present the RGBD BSE dataset and a depth level quantization scheme that provides an avenue for training a Machine learning model and calculating its hit rate. We were able to show that the previous study's accuracy is 30.33%. Moreover, adding a simple shadow area as feature and changing the Machine Learning prediction model to Support Vector Machine boosts the algorithm's accuracy to 58.83%.