Georgetown College. Georgetown University.

Dr. Sellers' current work focuses on data imaging and analysis of proteins. (Photo: Roland Dimaya)
Dr. Sellers discusses some of the data behind the images she works with. (Photo: Roland Dimaya)
Dr. Sellers is one of the faculty members who are part of the new Statistics program. (Photo: Roland Dimaya)

"I always tell my students, statistics is everywhere. Everywhere! It's a fundamental component of any type of research."
--Dr. Kimberly Sellers

By Katherine Morrissey

When Dr. Kimberly Sellers looks at a photograph of proteins taken in a laboratory, she doesn’t just see dark spots on a light background. Instead, Dr. Sellers, a professor in Georgetown’s Department of Mathematics, sees numbers. Thousands of them, combined by imaging technology and software to create the pictures that scientists analyze daily in labs around the world.

For Dr. Sellers and her colleagues, numbers build the image. Statistics are then used to break the image down again and make them easier to read. Scientists studying proteins typically contend with blurred and cluttered raw images, filled with hard-to-analyze shapes and shadows. In analyzing these images, minimizing human error and differences of interpretation becomes an important concern. Dr. Sellers’ work focuses on developing statistical techniques to process scientific data and images, focusing in particular on background correction methods, normalization, alignment, and feature detection. She looks for ways to improve imaging software and lab methods, seeking to automate analytical procedures where possible.

Standardization is of particular importance, but so is accuracy. When creating modeling tools and trying to screen background noise from an image, Dr. Sellers pays just as much attention to what is being taken out of the image as she does to the information that remains, carefully checking to be sure that important small details in the data are not being removed by accident.

Much of her current work involves proteomic data analysis—the imaging and analysis of proteins. With the information gained from the Human Genome Project, scientists have begun to focus with greater intensity on the arrangement and appearance of individual proteins, but the images that show individual proteins are incredibly challenging to interpret. Some proteins are large and easy to detect, while others are small, barely visible blurs within the image.

“Interest in image analysis relating to proteomics deals primarily with differential expression—an object ‘appearing’ and ‘disappearing’ from view, like a gene turning proteins on and off,” explains Dr. Sellers. “We can also ask, however, about differential modification—an object moving slightly or changing shape. This implies that the protein is present in both healthy and diseased states, but its makeup is somehow altered. Thus, the positioning of the imaging technology, how the image is taken, and the overall accuracy of the photograph, are incredibly important.”

The different modeling tools Dr. Sellers has created help with this process, but she feels that it is also important to increase awareness about the complexity of visual analysis and the importance of thorough, careful lab procedures for image production and processing. Researchers need to have a sense of the types of image modeling that are available to them, which are best suited to their research questions, and the ways that they can consult with statisticians and use processing tools to eliminate human error.

Assisting other scientists with their research allows her to apply her skills across disciplines and the interdisciplinary nature of this work appeals greatly to Dr. Sellers.

“I always tell my students, statistics is everywhere,” she says. “Everywhere! It’s a fundamental component of any type of research.”

Dr. Sellers’ enthusiasm for mathematics and statistics has been with her since she was a young girl. She decided to be a mathematician in the third grade and never lost sight of her goal.

“My parents were both educators who had master’s degrees, so I interpreted a master’s degree as a ‘standard’ with regard to education among their generation,” she explains. “That being said, I thought that a Ph.D. would be expected by the time I was of age to get a Ph.D. I was in the third grade and I knew I liked math, so I chose that or some math-related field as my goal.”

As a woman of color, Dr. Sellers also grew up acutely aware that the field of mathematics lacked diversity. But her parents, a college professor and public school teacher, encouraged and supported her in her goals.

“When I was growing up, my father received a publication called Black Issues in Higher Education,” says Dr. Sellers. “We’d look together, regularly, at the listings of African Americans graduating with Ph.D.s in Mathematics. Sometimes there would be one person listed, sometimes three, and sometimes none, but that wasn’t stopping me. I just knew I was going to be one of those names.”

Given her love for mathematics and her awareness of the challenges some potential mathematicians may face, it’s no surprise that Dr. Sellers works hard to promote diversity in the field and maintain a welcoming and supportive environment. She’s actively involved in diversity initiatives and serves on the steering committee for the Infinite Possibilities Conference, an event that celebrates and promotes diversity in mathematics. In addition to this work, Dr. Sellers also teaches two sections of Introduction to Probability and Statistics for undergraduates and the Mathematical Statistics course for graduate students in the new Master’s in Applied Mathematics and Statistics Program.

See article Georgetown Launches Mathematics and Statistics Master's Degree for more information about the new Master’s program in Applied Mathematics and Statistics.

Print Article