Defense Event

Architectural Support for Medical Imaging

Richard Sampson

Thursday, March 30, 2017
2:00pm - 4:00pm
3725 Beyster Bldg.

Add to Google Calendar

About the Event

Advancements in medical imaging research are continuously providing doctors with better diagnostic information, removing the need for unnecessary surgeries and increasing accuracy in predicting life-threatening conditions. However, many of these techniques are currently limited by the capabilities of existing computer hardware, restricting them to expensive, custom-designed machines that only the largest hospital systems can afford or even worse, precluding them entirely. Many of these issues are due to limitations in existing hardware which is ill-suited for these types of algorithms. In this thesis we discuss our efforts to improve architectural suppport for advanced medical imaging methods. We begin with Sonic Millip3De, a low-power, highly parallel hardware for 3D ultrasound. Using this design, we illustrate the orders-of-magnitude performance per watt improvement that better medical imaging hardware can provide, especially with a hardware-software co-design. We also show validation of the design using a scaled-down FPGA proof-of-concept and discuss our further refinement of the hardware to support a wider range of applications and produce higher frame rates. We then introduces MIRAQLE, an advanced medical imaging benchmarking suite to help bridge the gap between computer architects and medical imaging. We discuss background for iterative X-ray CT and dynamic MRI algorithms, which with 3D ultrasound, constitute the three key imaging problems in MIRAQLE. In addition to providing a benchmark suite for these workloads, we also develop a unique analysis tool that provides image quality feedback for each algorithm. This will allow hardware designers to explore acceptable image quality trade-offs in algorithm-hardware co-design, potentially allowing for even more efficient solutions. The overall goal of this work is to illustrate and enable hardware support for these critical medical imaging tasks to make them practical for wide clinical use.

Additional Information

Sponsor(s): Thomas F. Wenisch

Open to: Public