Figure 6. Comparison of the output from the proposed method with that of a previously described specialised PCM image segmentation algorithm. Images shown are from the mESCs data-set. Trainable segmentation based on five scales local BIFs histograms. For the specialised algorithm, images were processed using optimal segmentation parameters determined using the same mESC data-set as reported in the original paper (Jaccard et al. 2014). TP is true positives, FP is false positives, TN is true negatives and FN is false negatives.