Adaptive Localizing Region-Based Level Set for Segmentation of Maxillary Sinus Based on Convolutional Neural Networks

Figure 9

Examples of CRF-FCN methods. For explicit expression, masks had been introduced. The left column represents original CT images with lesions of different cases. Images of the middle column are ground truth of manual segmentation. The right column is the result processed by CRF-FCN.

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)