Khan M Iftekharuddin
Old Dominion University, USAFor research leadership and contributions to research and development of 3D medical imaging analysis, computational modeling, pattern recognition, and human-system interaction.
Khan Iftekharuddin never imagined the type of interdisciplinary research he does today as a child in Bangladesh. Without easy access to computers until college, he learned about technology more abstractly. He recalls that while away at boarding school, he received issues of The Reader’s Digest, a periodical that features special interest stories on a variety of topics. One issue included a story about artificial intelligence. This sparked his interest in science instantly, and he now gets to work at the intersection of engineering and medicine developing AI technologies.
Following this interest in AI, he decided to pursue degrees in electronic engineering. Like many of his peers, Khan wanted to move to the US and learned of a strong electro-optics program at the University of Dayton in Ohio. A trusted friend helped him connect with the school, and he made the journey, first to New York and finally to Ohio. Khan remembers arriving in the fall when the days were already short, but feeling an immediate sense of space. It wasn’t like the movies he’d seen growing up, which usually depicted America’s big cities and dazzling lights. Khan quickly adapted to the new setting, always cognizant of his greater goal of pursuing his research.
Khan’s early research was based on optical information processing and pattern recognition. After graduating, he worked for about five years in industry, first in a software company and later in a corporate Research and Development unit, doing sensor signal processing research. When an opportunity came for an academic position at North Dakota State University (NDSU), Khan was ready to make that transition. Many of his family members were professors, and he was eager to follow in their footsteps.
At NDSU, he began a research track that he still pursues today: computational modeling for brain tumors. This topic started to gain momentum when he moved to the University of Memphis and began collaborating with clinical colleagues at St. Jude’s Children’s Hospital. Since then, he’s expanded to include brain tumor recognition, detection, classification, and grading in his current position at Old Dominion University. Specifically, Khan and his team are developing machine learning and artificial intelligence methods to help triage the high number of brain MRI scans radiologists are required to read daily for disease diagnosis. The aim is never to replace the clinicians, but to help radiologists by discarding unnecessary imaging slides as they make potentially life-and-death diagnosis for patients.
While it all started with trying to help doctors lessen their workload, the focus has widened as their technology has advanced. One of the aims has been to answer the question: “Is it possible to predict the patient survivability given the aggressiveness of a tumor from analyzing the imaging scan?” The current focus of this research is to improve the tumor grading ability. There are over 100 different types of tumors, and Khan and his group are developing non-invassive imaging analysis and AI methods to detect recurrence of a particularly aggressive form of brain tumor, known as glioblastoma. These tumors invariably recur; when they do, they are more aggressive. Khan and his team are working to track and diagnose these specific tumor recurrences to help develop patient-specific treatment plans. To do this, they’re expanding from MRI scans to include histopathology, Proteomics, genomics, clinical, and other types of data to get a better understand of patient-specific diagnosis and prognosis.
Another interdisciplinary research topic relates to developing computer vision and AI methods to help children with Autism. Children with autism spectrum disorder (ASD) are often unable to effectively communicate and interact socially, and some are even non-verbal. However, many patinets with ASD have strong interests and skills in gaming. Using 3D high-resolution optical imaging and IR devices paired with other types of sensors (e.g., eye tracker, mouse), Khan and his team are trying to determine what attracts these children's attention, aiming to identify biomarkers that set them apart non-invasively. . Ultimately, they hope to develop a better way to engage these children so they can more easily interact with other people.
Initially, it was difficult for Khan to navigate these intersectional interdisciplinary spaces because he was coming from the optical information processing side. He had to work with the clinicians directly to learn the basics and slowly expand so he could better meet their needs. He recalls, “I attended these tumor clinics early in the morning, at 7 o’clock, with radiologists, surgeons, radio physicists, and more, to learn about their most difficult cases and challenges. I used to take my students to learn, so we could all learn how they did their analysis.” These types of projects take a large team with diverse areas of expertise in order to succeed, and Khan has been fortunate to have great support from his collaborators.
Photo courtesy of Khan Iftekharuddin
Profile written by Samantha Hornback
