Detection and monitoring of microwave ablation by ultrasound imaging based on convolutional neural network

Microwave ablation (MWA) is widely used in clinical treatment, but conventional ultrasound B-mode monitoring imaging method cannot provide real-time changes in the ablation area accurately during thermal ablation. With the further maturity of classification technology of convolutional neural network (CNN), the CNN will certainly play an important role in the processing and analysis of medical images. In this study, we proposed and evaluated an US image based on CNN architecture for the detection and monitoring of thermal lesions induced by MWA in porcine liver. The values of the ablation area under the receiver operating characteristic curve for US image based on CNN and B-mode image were 0.8728 and 0.6904, respectively. The results show that it is feasible to use convolutional neural network to monitor the changes of ablation area during MWA.