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

Figure 5

Examples of two classes and preprocessed inputs. The two columns on the left show images that were obtained on lesion edge and related preprocessed ones. The two columns on the right are from the bone of sinus cavity. All sampled points locate in the center of images with pixels. By CLAHE analyzing, inputs of different classes stress their obvious features, assisting the CNN network a better performance of learning and predication.