LOS ALAMOS – A Manzano High School senior who taught a computer how to identify plaques in brain scans of mice for Alzheimer’s disease research won this year’s Supercomputing Challenge in Los Alamos.
The 22nd annual competition, sponsored mainly by Los Alamos National Laboratory and Sandia National Laboratories, had more than 200 New Mexico elementary through high school students, who competed in teams of up to five, according to a lab statement.
These groups worked through the school year using supercomputers to program and model answers to complex real-world problems such as the environmental impact of polystyrene used in restaurants, a model of how water moves through the Santa Fe watershed, a simulation of ant colony behavior and how airborne diseases infect passengers on an airplane. The award ceremony took place Tuesday at the Church of Christ in Los Alamos.
Jordan Medlock took the first-place, $1,000 cash prize for his project, “Detection of Alzheimer’s Disease Plaques …” in a magnetic resonance image analysis of a mouse brain. Medlock, who hopes to attend the University of New Mexico for computer science and possibly biology, said he started the project as a solution for UNM researchers who are studying Alzheimer’s disease. One symptom of the disease is the buildup of protein deposits in the brain, called “plaques.”
Through brain imaging, these plaques can be seen but, according to Medlock, researchers now count them manually, taking up to one month to complete one scan. This makes research, which often involves many such scans, very time consuming and difficult, he said.
But Medlock said his project can identify and count the plaques on such an image in just 20 seconds.
“One of the researchers gave me a hug,” Medlock said of the time he discussed his project with UNM scientists.
In his conclusion, Medlock states his processing algorithm found 1,723 plaques in his scan of the mouse brain. There were 668 plaques in the scan when the plaques were counted manually. Medlock concludes the differences could be due to the computer recognizing and counting regions in the brain that look similar to plaques. He states the program would be more useful for measuring plaque density than the previous method of hand-counting.
Medlock’s mentor on the project, Laurel Sillerud of UNM’s Department of Biochemistry and Molecular Biology, could not be reached for comment Tuesday. In addition to the cash prize, Medlock was selected for a $4,000 scholarship, he said.
Los Alamos Middle School student Cole Kendrick, last year’s first-place winner, came in second with a model that studies the role dark matter could play in the collision of two galaxies. Kendrick said he used the Milky Way and Andromeda galaxies as his subjects because they’re expected to collide with one another in about 4 billion years.
“No one really knows what dark matter is,” Kendrick said.
But its existence can be inferred, Kendrick said, and it plays a role in how galaxies rotate or – in the case of his project this year – crash. He said his model showed the presence of this mystery substance makes a universe of difference in the collisions. If both galaxies have dark matter “halos” that are similar in mass, the collision will cause the galaxies to merge into one larger elliptical galaxy. If one galaxy has more dark matter, the smaller galaxy will be torn apart and absorbed into the larger galaxy.
Kendrick, who said the project combines his interest in computers and astronomy, had his father Brian Kendrick, a theoretical physicist at LANL, as a mentor.
Cole Kendrick won $500 for placing second, as well as a $100 crowd favorite award chosen by teachers and students and a best-presentation award.
Third-place finalists were a team from La Cueva High School in Albuquerque. Alexandra Porter, Stephanie Djidjev and Lauren Li each took home $250 for their project “Optimizing Community Detection.” The three students also received the Women in Science and Engineering Award.
More detail on the projects can be found in the participants’ final reports. These are posted to the competition’s website at www.supercomputingchallenge.com.