Fig. 1.

Framework of the proposed convolutional neural network (CNN) system for cervical cancer screening. The convolution network extracts the image information to obtain the feature map, and the proposal network screens the target region to obtain the target position information on the basis of the feature map. The pathological cells and their locations were identified by a convolutional classifier based on the feature map and target location. Pathological cell information was obtained by combining the recognition results of the two networks. The results of the two models are similar. However, the two structural models are trained with images of different magnification levels of 200X and 400X, and the parameters of the two models are different