Localization Bar Detection by Deep Learning in Magnetic Resonance Guided Focused Ultrasound

Magnetic resonance guided focused ultrasound (MRgFUS) can focus the ultrasonic energy on a targeted tissue to achieve ablation. In this study, we propose a convolutional neural network (based on Tensorflow) based upon deep learning to detect the position bars to registry the coordinate of MR and HIFU systems, instead of the traditional detection algorithm. The results show that the model achieves the average precision (AP) of 95.6% with the precision of 99.0% and the recall of 99.0% at a score threshold of 0.5.