Active Learning for Segmentation by Optimizing Content Information for Maximal Entropy