From Ultrasound Image Classification to Ultrasound Video Classification Approaches

The aim of this chapter is to present how we can move forward from differentiating between ultrasound image using Deep Learning state-of-the-art algorithms, to classifying ultrasound videos using a combo between Deep Learning and Statistical Learning. Different convolutional neural networks are trained on image scans to recognize distinctive view planes of the fetal abdomen. After this step a statistical analysis is performed to choose the appropriate method that is going to be applied on the ultrasound movie. Since during an ultrasound a fetus breathes and moves it is possible that different view planes overlap, hence it is necessary to find means to establish the correct view plane for further processing. In this chapter, we are going to present two different methods that give good results in determining the correct view plane in a video: a probabilistic approach and a peer pressure approach.