Please share a little about your background and your research experiences.
Dan Clymer is a recently graduated PhD student from the Mechanical Engineering Department at CMU. His research focused on machine learning, computer vision, and pattern recognition. Jonathan Cagan is the George Tallman and Florence Ladd Professor in Mechanical Engineering at CMU with expertise in engineering design automation and methods merging AI, machine learning (ML), and optimization methods with cognitive science problem-solving. Philip LeDuc is the William J. Brown Professor in Mechanical Engineering at CMU who has been working on the intersection of engineering with medicine and biology for over two decades. The team also included collaborators at UPMC and the University of Pittsburgh. Liron Pantanowitz was the Vice-Chairman. For Pathology Informatics at UPMC and a Professor of Pathology and Biomedical Informatics. Janet Catov is an Associate Professor, Department of Obstetrics, Gynecology & Reproductive Sciences, and the Department of Epidemiology, University of Pittsburgh.
What led you to the PHDA?
As machine learning (ML) researchers, two factors that caused us to be interested in healthcare-related research are: (1) there are many unsolved problems left in healthcare that ML has the potential to help with, and (2) healthcare problems tend to be extremely important ones that urgently need to be solved. However, to have a successful ML project, one needs a pipeline to gather, process, annotate and move data – all of which can be difficult in the healthcare setting. The PHDA helped make available the medical expertise necessary for this pipeline to work, and this is one of the primary reasons why we were led to the PHDA.
Walk us through your project.
After a baby is born, doctors sometimes examine the placenta – the organ that links the mother to the baby – for features that indicate health risks in any future pregnancies. Unfortunately, this is a time-consuming process that must be done by a specialist, so most placentas are discarded without examination. In this project, we developed an ML approach to examine placenta slides so more women can be informed of their health risks. One reason placentas are examined is to look for a type of blood vessel lesions called decidual vasculopathy (DV). These indicate the mother is at risk for pre-eclampsia – a complication that can be fatal to the mother and the baby in any future pregnancies. Once detected, pre-eclampsia can be treated, so there is considerable benefit from identifying at-risk mothers before symptoms appear. However, while there are hundreds of blood vessels in a single slide, only one diseased vessel is needed to indicate risk. It is quite difficult for a computer to simply look at a large picture and classify it, so the team introduced a novel approach where the computer follows a series of steps to make the task more manageable. First, the computer detects all blood vessels in an image. Each blood vessel can then be considered individually, creating smaller data packets for analysis. The computer will then access each blood vessel and determine if it should be deemed diseased or healthy. At this stage, the algorithm also considers features of the pregnancy, such as gestational age, birth weight, and any conditions the mother might have. If there are any diseased blood vessels, then the picture, and therefore the placenta, is marked as diseased. This technology can decrease healthcare costs, allowing a majority of mothers and infants to have access to a microscopic placenta examination.
How do you and your project partners’ strengths complement each other?
The researchers from CMU developed the algorithms to analyze whole slide placental images. However, this would have been more challenging without the UPMC team of Drs. Stefan Kostadinov, Janet Catov, Lauren Skvarca, and Liron Pantanowitz who helped with making the de-identified data available and provided the medical background to understand, annotate, label, etc. the data. This was a beautiful collaboration between engineering and medicine as each brings expertise to the table that, when combined, creates novel findings that can help many people. This is a wonderful example of the sum of the parts being greater than the individual parts.
In what ways has UPMC played a role lending clinical expertise and sharing data?
The UPMC team has been collecting data from childbirths for years to build a database for research purposes that includes these placenta images along with outcome data for the mother and infant and has been de-identified to maintain privacy. Researchers from UPMC also collaborated with us to help understand and annotate these images to mark diseased blood vessels, which were used to train the algorithm. It would have been difficult to conduct this research without them.