Ultrasound is a relatively inexpensive, portable, and versatile imaging modality that has a broad range of clinical uses. However, ultrasound images suffer from the disadvantage of being user dependent and noisy which makes the interpretation of ultrasound images is sometimes difficult. Recently, algorithms based on artificial intelligence have been significantly improved in medical imaging. To alleviate the difficulty of processing ultrasound images, deep learning techniques are gradually introduced to improve imaging quality, tissue characterization and lesion localization for better diagnosis and therapy.
Contrast-enhanced ultrasound imaging using gas-filled microbubbles (1-10 Î¼m in diameter) as vascular tracers is well established. Contrast imaging modes such as harmonic and subharmonic imaging are used worldwide to create parametric data based on time intensity curves of the microbubble kinetics. Moreover, subharmonic signals can be utilized for noninvasive pressure measurements, and examples from cardiology, portal hypertension and cancer treatment monitoring will be presented.
In this session several new technologies will be discussed including (a) ultrasound image quality improvement using deep learning, (b) deep-learning-based applications in ultrasound-aided diagnosis and radiotherapy, (c) quantitative time intensity curves derived from contrast-enhanced ultrasound imaging, and (d) the use of contrast-enhanced ultrasound imaging for noninvasive pressure estimation.
1.Â Understand the limitations of current ultrasound imaging techniques.
2. Understand where several gaps for advancement exist in the field of ultrasound imaging.
3.Â Understand technologies that can improve the accuracy and efficiency of diagnosis and therapy in ultrasound imaging.