Meet Professor Gadget
|Beyond Video Karaoke
Imagine replacing Michael Jackson’s head with your own in the Thriller video, and hearing the King of Pop’s falsetto come out of your mouth. Thanks to computer scientist Gerard Medioni, such hyper-video karaoke stunts are now feasible, and thanks to the USC Stevens Institute for Innovation, they are already making waves on YouTube. Earlier this year, thousands of music fans uploaded their own versions of Thriller, starring themselves, in a promotion involving media company Big Stage Entertainment, the startup born of Medioni’s technology. (Most videos were taken down after the promotion, which was timed to the 25th anniversary of Thriller’s release.)
Medioni, who is a professor in the USC Viterbi School of Engineering, has been studying how humans infer three-dimensional reality from photographs for 25 years. His first commercial discovery was a system for creating a 3-D model of a face out of photos taken simultaneously from two camera angles. The system is still in use in some prisons, where it helps authorities confirm that they are booking or releasing the right inmate.
Medioni’s next step was to use video from three slightly different images, all taken by one camera, to render a face in 3-D. At the time, he was thinking mainly of security applications, such as suspect identification from security camera images.
But a group of multimedia experts immediately saw the technology’s potential for entertainment and social media. Last year, they negotiated a license deal with Medioni and the institute, and startup Big Stage Entertainment was born.
USC Viterbi alum Doug Fidaleo PhD ’03, who worked as a postdoctoral researcher in Medioni’s lab, joined the company as chief scientist and brought along three other USC alums. Medioni is still involved as a consultant.
Big Stage is fine-tuning its technology for a planned commercial launch later this year. Fidaleo and company developed a suite of key technologies to enable insertion of a complete 3-D head into any “Big Stage-enabled” video. This breakthrough allowed the company to launch the Thriller promotion that prompted the proliferation of Michael Jackson impersonators on YouTube.
Currently, the company has 25 employees crammed into a two-story loft in South Pasadena, working madly toward a commercial launch this summer.
The team sees limitless applications: from video karaoke to user-generated spoofs, from avatars in virtual worlds to big-budget television commercials that put viewers behind the wheel of the newest Lexus. (The technology also has decidedly evil potential: Imagine the possibilities for bitter spouses to embarrass their exes, or for disgruntled employees to ridicule their bosses).
Big Stage’s next goals are more user-generated content tools, including the ability to add your own voice to any scene.
YouTube (and entertainment) will never be the same.Faking Out MS
Drugs against multiple sclerosis cost society about $6 billion annually, even though they do not work: Fewer than one in three patients respond, and most experience significant side effects, says Keck School neurologist Leslie Weiner.
That observation led Weiner to develop a radical new therapy: a chemical decoy that works something like a nicotine patch.
In experiments with mice, the patch – actually a wafer-thin pouch implanted under the skin – reduced MS symptoms by 90 percent.
The treatment has none of the downsides of existing drugs, Weiner says, because it is not a drug in the traditional sense. He calls it a “cell-based therapy” because that’s what the patch contains: healthy human cells.
The cells in the patch, which lasts around six months, are engineered to release protein-like snippets called peptides. These snippets fool specific immune cells into latching onto them. In patients with MS, those same immune cells, if not fooled by the decoys, would enter the brain and consume the myelin sheath that protects the nervous system.
It is the consumption of the myelin sheath, and subsequent degradation of the nerves inside, that causes the loss of muscle function associated with MS, which afflicts 2.5 million people worldwide.
With help from the USC Stevens Institute, Weiner created a start-up company, Anergix, to find the capital necessary for clinical trials.
“They [the institute staff members] have arranged for us to make presentations to a variety of investors,” Weiner says. “It’s difficult to sell something new. But we have great hope for this. This is a really innovative approach.”Custom Chemo
Doctors have long known that not all chemotherapies are equal. Certain cocktails work better in certain people with certain tumors. Yet the standard treatment for all first-time cancer patients is pretty much the same.
That never made sense to oncologist Heinz-Josef Lenz, who chairs the gastrointestinal oncology program at the USC/Norris Cancer Center. Years ago, he started working with biochemist and molecular biologists Peter and Kathy Danenberg at the Keck School of Medicine of USC to see if they could decipher the genetic signatures of human tumors. Lenz was hoping to find markers – specific genes or gene mutations that would correlate strongly with a tumor’s response, or failure to respond, to a particular cancer drug.
It was a lonely search. Though Lenz and the Danenbergs identified the first markers in 1993, “nobody believed in us in the beginning,” he says. “It took us two years to publish a paper.”
But even the most dogmatic researchers could not ignore the growing evidence from the Lenz/Danenberg lab. After identifying several more markers for drug response in colorectal, gastrointestinal and lung cancer, Lenz now states: “I can predict in all cases who will not respond, and I can predict with over 95 percent probability which patients with colon cancer will respond to a specific treatment.”
Lenz’s markers predict response not only to traditional chemo cocktails, such as Folfox and Folfiri, but also to newer drugs like Avastin and Erbitux. Lenz’s laboratory was one of the first to identify molecular markers associated with outcomes to these two novel targeted therapies.
The ability to optimize response to a first course of treatment is especially important, Lenz says, because the second course tends to be far less effective. In cancer care, you don’t often get more than one or two shots.
Last year, in a deal negotiated through USC Stevens, the drug company Abraxis licensed the rights to all of Lenz’s markers. The company plans to develop mass-market test kits for use in doctors’ offices.Parboiled, Not Frozen
Kathy Danenberg followed Heinz-Josef Lenz’s work with cancer markers from a privileged vantage point – as one of the researchers in her husband Peter Danenberg’s collaborative lab. In theory, Lenz’s work was revolutionary. But pragmatically, it had hit a wall, because to identify the markers in a particular tumor, Lenz depended on a sample of frozen tissue.
The problem is that no one in clinical oncology collects and saves frozen tumor tissue. The standard practice after a biopsy is to preserve the sample in paraffin – a waxy chemical preservative.
“You just can’t run around getting frozen samples from patients,” Kathy Danenberg explains. You have to stand in the operating room with a canister of liquid nitrogen. Not practical.
And, she adds, “the frozen biopsies don’t even show the cancer cells very accurately.”
In the meantime, hospitals and oncology labs have literally millions of biopsy samples sitting in paraffin, just waiting to be analyzed.
Danenberg’s solution flouted the conventional wisdom that genetic material in tissue needs to be cooled for proper analysis. Instead, she heated it near to the boiling point – and discovered she could extract more than enough genetic information from the paraffin-dipped tissue to perform accurate tests.
Since 1999, she says, “we’ve been analyzing 15,000 samples a year.”
But that’s not enough to bring customized cancer care to the millions of patients a year who need it. So with the help of USC Stevens, Danenberg formed Response Genetics, a testing company that went public in 2007. She is its president and CEO.
Response Genetics is currently the only company offering a single genetic test for four markers crucial in predicting chemotherapy response. By using the test and customizing care appropriately, oncologists have been able to double patients’ response rates to the common chemo cocktails.
Response Genetics is now working with the pharmaceutical giant Roche to develop a mass-market version of its genetic test for biopsy samples.
– Carl Marziali