Fetal Standard Plane Detection in Freehand Ultrasound Using Multi Layered Extreme Learning Machine
Identifying the standard planes for measurement from ultrasound imaging is a tedious task since the quality of the ultrasound view of the fetus depends on a number of factors: artifacts in the ultrasound images; body mass index of the mother; position and motion of the fetus. Manual detection of standard planes of fetal parts and locating the specified fetal structure is time-consuming and expert dependent, requiring many years of experience for accurate results. This chapter proposes a method based on Multi Layer Extreme Learning Machine (ML-ELM) for automatic detection of three fetal standard views for measurement of Bi-parietal Diameter (BPD), Femur Length (FL), and Abdominal Circumference (AC) in freehand 2D ultrasound midpregnancy examinations. The ML-ELM uses ELM-based Auto encoders (ELM-AE) as basic building blocks for feature representation using singular values.