Diagnostic ultrasound is the most widely used medical imaging modality of obstetrics. The corresponding medical image analysis techniques are important for the improvement of diagnostic accuracy and objectivity. This paper first introduces the principles and characteristics of methods in ultrasound medical image analysis, and discusses a number of techniques and corresponding algorithms including image filtering, image segmentation and machine learning. The key points of designing reliable methods for ultrasound medical image analysis are provided. Then, the applications of these techniques, including standard plane selection and biometric measurement, are introduced in the background of obstetrics. Finally, the development direction of intelligent diagnosis in prenatal ultrasound is discussed.