Using Virtual Reality and Machine Learning to Improve the Pancreatic Cancer Prognosis

Part five of our AI Researcher Profile series invites Distinguished Professor Arie Kaufman of the Department of Computer Science, and CEWIT Chief Scientist, to discuss his innovative techniques for 3D virtual pancreatography, its applications to AI, machine learning, and ultimately, human life.

AI Institute: You have developed techniques for 3D virtual pancreatography. Can describe what that is and why it is exciting?

Professor Kaufman: Pancreatic cancer is the most lethal cancer, as it progresses rapidly, with little symptoms to show for it. The prognosis for this disease is typically very poor because of this, and therefore, the five-year survival rate is less than 10 percent. We are interested in trying to curtail this. In 3D virtual pancreatography, we are conducting a CT scan of the patient’s abdomen, segmenting the pancreas and cysts, performing a classification on the pancreatic cysts, and visualizing them. Through this research, we have been able to provide to radiologists the visualization and analysis tools to distinguish between the types of cysts. Ultimately, this work can mean the difference between life and death.

AI: What inspired this research? 

PK: Virtual reality is my area of expertise. The inspiration for this work was started because of my original work with virtual colonoscopy. 3D virtual colonoscopy is more accurate and significantly more cost-effective than the invasive, traditional procedure. Most importantly, we already saved more than 30,000 lives through virtual colonoscopy. Because of this, we were inspired to extend a similar method toward the pancreas and other organs. Our work quickly expanded to areas like the bladder, the brain, the prostate, and now to the pancreas. Due to the innate similarities between the colon and the pancreas, like their respective place in the digestive system and their similar cancer signatures, there was a smooth transition between 3D virtual colonoscopy and 3D virtual pancreatography. The new technology proved to be successful for diagnosing pancreatic cancer, just as it did in the colon.

AI: Your 3D Virtual Pancreatography technology is heavily based in machine learning. What is the future in this work?

PK: Much of the research and work in visualization which my colleagues and I have dedicated ourselves to for decades is being replaced by machine learning. There is a profound impact in this, as machine learning is sweeping the fields of AI and computer science. With respect to 3D virtual pancreatography, the processes of segmentation and classification in this work are done through the use of machine learning because it is a very powerful tool. A promising element of machine learning is that its uses do not have to be employed exclusively for the pancreas, rather, it can extend itself to other organs within the body. In spite of this, I first try to employ conventional techniques before I use machine learning. I believe that machine learning should be used as an additional tool, not as a replacement for other techniques.

AI: What do medical professionals think about this new technology? How will 3D Virtual Pancreatography reshape the face of pancreatic cancer diagnoses?

PK: We collaborated with Johns Hopkins Hospital radiology and pathology departments which provided us with the initial set of fully characterized CT scans. Our work has expanded beyond Johns Hopkins, as we are currently working with a group at Stony Brook University hospital. The group consists of two Stony Brook radiologists, two pathologists, and one surgeon. Collaboratively, we are trying to improve upon the current procedures that exist for pancreatic cancer detection and especially for early detection. One exciting direction we have been working with our team on is augmented reality for 3D virtual pancreatography. This is a three to four year project which will enhance the detection and management of pancreatic cancer. The improvements to 3D virtual pancreatography could revolutionize the face of pancreatic cancer early detection and management, and ultimately, save the people from this lethal disease.

AI: What is your position as CEWIT Chief Scientist, and how does that impact the Stony Brook community?

PK: The Center of Excellence in Wireless and Information Technology (CEWIT), was founded more than a decade ago through a designation of NY State Governor and a 250 million dollar partnership between the university, industry, and NY State. Here at CEWIT we focus on applied research, develop new technologies, and transfer those technologies to industry. With my position, I am in charge of the technology side of this relationship. Because of this, I am in constant discourse with companies, dispersing our work to other institutions for future use. While this does impact the Stony Brook community, by increasing our research prowess, there is an impact that is far beyond the university. Because of CEWIT, we are able to share our findings and work with a larger research community as well as industry, enhancing society as we know it.

AI: What suggestions do you have for somebody looking to get involved with Artificial Intelligence?

PK: You must be careful to not discard the methods which brought us to where we are today, despite the overwhelming benefits of machine learning. These original approaches can enhance current research in ways which the exclusive use of machine learning cannot. Ultimately, be sure to use AI as a supplement to your research projects, not a replacement.

AI: Thank you, Professor Kaufman.