Current Research and Scholarly Interests
Our laboratory is an ultrasound engineering laboratory that develops and implements ultrasonic beamforming methods, ultrasonic imaging modalities, and ultrasonic devices for diagnostic imaging applications. Our current focus is on beamforming methods that are capable of generating high-quality images in difficult-to-image patients and imaging conditions. These methods include general B-mode and Doppler imaging techniques that utilize additional information from the ultrasonic wavefields to improve image quality. We attempt to build these imaging methods into real-time imaging systems in order to apply them to clinical applications, such as cardiac, liver, and fetal imaging. In addition, our laboratory develops ultrasonic imaging devices, such as small, intravascular ultrasound (IVUS) arrays that are capable of generating high acoustic output. These arrays are capable of generating radiation force in order to push on tissue to elucidate the mechanical properties and structure of vascular plaques, but can be utilized for therapeutic applications of ultrasound as well.
Current projects in our laboratory involve the simulation of nonlinear, acoustic wave propagation under complex models of human anatomy and the impact of anatomy and acoustic parameters on the resulting images. Often, the anatomy and acoustic parameters are the source of aberration and diffuse reverberation of the wavefronts, both of which contribute to image clutter. In addition to modeling and understanding these sources of clutter, we have developed imaging methods that utilize the spatial coherence of the ultrasonic wavefields in order to mitigate the impact of reverberation noise (called short-lag spatial coherence [SLSC], short-lag angular coherence [SLAC], and coherent flow power Doppler [CFPD] imaging) and the estimation of local sound speed in order to mitigate wavefront distortion. These methods demonstrate significant improvement in image quality and the ability to detect slow flow under difficult-to-image scenarios. We have developed a prototype imaging system capable of implementing some of these techniques at up to 30-35 frames per second. We are currently developing methods and approximations to the spatial coherence functions in order to increase the real-time display and image quality. This system will be utilized in clinical studies of cardiac function and placental imaging.
We have recently integrated machine learning techniques to construct neural-network beamformers for a variety of imaging tasks. For example, we recently constructed a neural-network beamformer to output ultrasound images with speckle reduction. These images maintained the resolution of conventional ultrasound images while improving the visualization of tissue structures in human imaging. We have also demonstrated that this neural-network beamformer can be implemented in real-time imaging.
Other projects in our laboratory include molecular imaging techniques and B7-H3 targeted microbubbles, passive cavitation mapping, and therapeutic ultrasound systems for drug delivery.
