Alexander Lu – Mulford Endowment Scholar
1st Year Scholar, MD/PhD candidate, Biomedical Engineering
Johns Hopkins University

Research: 
Surgeons use a variety of imaging systems to guide diagnostics and interventions; patient movement during imaging is often unavoidable and can substantially distort the images. This current research combines machine learning with physics-driven methods to improve the quality of intra-operative imaging technologies with the intent of improving diagnostics and subsequent treatment.

How Will Your Research Benefit Society?
The advent of intra-operative three-dimensional imaging, often achieved using cone-beam computed tomography (CBCT) scans, provides surgeons and interventional specialists with broad anatomical context and the ability to locate tumors and other procedural targets with greater precision than ever before. Despite its advantages, the moderately long acquisition time of CBCT (6-60 seconds) increases the likelihood of intra-scan patient motion. Small motions result in blur, also altering the shape and position of key anatomical structures; larger motions can completely obliterate the appearance of these structures. As such, motion can render CBCT images misleading or unusable for therapeutic guidance.

My research aims to develop novel strategies to estimate and compensate for motion in intra-operative CBCT images by combining anatomy-aware deep learning strategies with physics-driven image reconstruction methods. The major advantage of our approach is that it requires few assumptions about patient motion or appearance. Rather, our proposed method “autofocuses” CBCT images without needing prior imaging, external motion tracking, or user interaction, which is well-suited to interventional workflows, and allows this approach to be applied in many clinical settings, and allows surgeries and minimally invasive procedures to be performed with even greater accuracy and with fewer side-effects.

How will an ARCS Award Benefit Your Research?
Accuracy, validation and quality assurance of imaging algorithms is traditionally performed using a combination of simulation studies and experimental measurements obtained from known objects, known as phantoms, before being tested on real patients. It is costly and time-consuming to design and fabricate high-fidelity phantoms for studying patient motion. Support from the ARCS Foundation would help to defray these costs and provide more flexibility in designing and executing novel, rigorous experimental studies.

Career objectives:
I am pursuing a career as an academic physician-scientist, bridging cutting-edge clinical medicine and innovative biomedical engineering. I am passionate about the power of interdisciplinary research in medicine but at the same time, my interactions with patients remind me that each clinical image carries a unique story. Working with patients keeps me grounded with patience that complements the fervor I gain in the laboratory. To translate engineering solutions into improved patient experiences and outcomes, and make these innovations accessible to all patients, I strive to be a leader in the fields of radiology and biomedical engineering.