Justin Lee

Massachusetts Institute of Technology

Justin Lee earned his biology bachelor’s degree at the Massachusetts Institute of Technology in 2008 and enrolled in medical school in Texas, where he’d graduated from high school.

Lee had been interested in medicine since he was 13, when his father died of lung cancer. But in the Baylor University M.D.-Ph.D. program, he liked the research more than the medical training.

“I feel like my personality is more of an engineer’s personality,” says Lee, a Department of Energy Computational Science Graduate Fellowship (DOE CSGF) recipient. “I like to build stuff, like to figure out why and how things work. But just knowledge for the sake of it might not necessarily be good enough.”

Lee found his niche in a joint MIT-Harvard University Medical School graduate program. The Health Science and Technology curriculum includes helping treat hospital patients. That lets Lee pursue his computational approaches to medical imaging while keeping the ultimate goal in sight. “It really lets you see end-to-end the technology you’re developing,” from the “cutting edge all the way to the other end” – the patient who benefits.

In his research, Lee devises imaging algorithms: mathematical approaches that extract information from what microscopes collect, however hazily. Or as he puts it, “I’m using computational imaging to help better diagnose disease or to better treat disease.”

For example, one of Lee’s algorithms reconstructs images of things like cells and their parts by interpreting information about the phase of light the microscope or other instrument gathers. Light is a wave, with an amplitude and phase, but in the visible spectrum it oscillates so quickly that the phase can only be averaged over time. That loses the details, Lee says. “All you see is the intensity of the light, but the phase information actually has a lot of information about the cells” being imaged.

With his doctoral advisor, MIT’s George Barbastathis, Lee is developing techniques to computationally retrieve phase information from disorganized, multidirectional light entering microscopes.

Lee: Breast Cancer Cells

Take, for example, a digital picture from a tissue biopsy. In a standard, two-dimensional image, each pixel has information only about the light’s brightness. But with that information and at least one other picture, taken from a slightly greater distance, Lee applies a technique called transport of intensity to computationally retrieve the light’s amplitude and phase. By taking more pictures from different distances or illuminating the sample from different angles, he can retrieve the light field’s entire phase space. “If you know the phase space, you not only know how bright each pixel is, but you know from what direction the light is pointing at that pixel,” Lee says.

In standard biopsies, doctors can adjust a microscope’s focus to get a sense of depth when viewing a sample, but they lose that capability with 2-D digital images. Capturing phase-space information to create 3-D images would give doctors more confidence in what they see in digital pictures, which are becoming standard in medicine.

In another aspect of his studies, Lee develops machine-learning approaches to predict cancer recurrence rates. His method, combined with techniques like genetically testing cells, could help doctors choose the best treatment options. His algorithms analyze images of a patient’s cells and determine what treatment is most likely to work for him or her.

The work appeals to Lee’s hands-on engineering nature. It’s not a surprising course for someone who, with his brother, attached plants to a ceiling fan and spun them to see if they would grow diagonally in response to artificial gravity created through centrifugal force. (They did.)

Lee also made headlines in 2009 when he and a friend, Oliver Yeh, sent a weather balloon carrying a digital camera into the stratosphere, capturing pictures from the edge of outer space. They posted instructions on line and people around the world have since launched dozens of balloons with equally spectacular results.

Lee continues to satisfy his tinkering urge by creating demonstrations, like building a homemade laser wiretap, at the Boston Museum of Science and the MIT Museum. Most of these devices are unique, designed by Lee, and he delights in creating them. “I have a dream of one day retiring and selling science toys to kids,” Lee adds.

But first he must graduate, probably around December 2016. Lee hasn’t decided what to do then, other than move to California, where his wife lives.

Image caption: A microscope image of cancerous breast tissue, with cancer cells stained blue. In his research, Justin Lee devises algorithms that extract information from medical images. Credit: National Cancer Institute via Wikimedia Commons.