Applications of Machine Learning and Deep Learning

Ultrasound (US) imaging (sonography) is the most frequently performed cross-sectional diagnostic imaging modality in the field of medicine. It is non-ionizing, portable, cost-effective, and capable of real-time image acquisition and display. US is a rapidly evolving technology with substantial opportunities and challenges. Challenges include limited image quality control and high inter- and intra-operator variability. As US devices become smaller, due to progressive miniaturization of US devices in the last decade, increased computational capability can contribute significantly to decreasing variability through advanced image processing. In this paper, leading Machine Learning (ML) and Deep Learning (DL) approaches and research directions in US, with an emphasis on recent ML and DL advances is discussed. An outlook on future opportunities for ML and DL techniques to further improve clinical workflow and US-based disease diagnosis and characterization is also presented.