WPI Researcher’s Team Receives +$2.4 Million to Develop Smartphone App to Detect Wound Infections
A team led by Worcester Polytechnic Institute (WPI) researcher Emmanuel Agu has been awarded $2,458,174 by the National Institutes of Health (NIH) to develop a smartphone app that will use photographs, heat images, and algorithms to detect infections in the open wounds of patients at home.
The Deep Infected Wound Detector (DIWD) will enable visiting nurses and other health workers in the field to rapidly identify patients who need specialized care for diabetic ulcers, pressure sores, incisions, trauma-related injuries, and other wounds. The app, deployed on smartphones that are equipped with thermal cameras like those used by plumbers, could help health workers reduce unnecessary medical referrals, delays in care, and wrong care decisions that lead to amputation, said Agu, who is Harold L. Jurist ’61 and Heather E. Jurist Dean’s Professor in the (WPI) Department of Computer Science.
“It can be difficult to visually evaluate a wound when visiting a patient at home or in a remote location, because wounds are often obscured by dead skin and debris that can only be removed in a doctor’s office,” said Agu, who is a principal investigator (PI) and leader of the four-year project. “By combining photos with thermal images, which can detect temperature changes in tissue underneath the skin, we will develop a tool that will help health workers better evaluate wounds and determine whether to refer patients to specialists for medical care.”
Agu and a team of researchers will train deep learning computer models to detect infections from about 1,500 photographs combined with thermal images of wounds. Some photographs and thermal images will be taken from existing data sets. New data will be collected by researchers at UMass Chan Medical School, including photographs and thermal images of wounds. The team will then program a smartphone app and validate the technology in a study of about 100 patients.
Dr. Giorgio Giatsidis, assistant professor of surgery at UMass Chan, also is a PI on the project. Other researchers involved are co-investigators Bengisu Tulu and Diane Strong, professors in The Business School at WPI; and Clifford Lindsay, assistant professor of radiology at UMass Chan. In addition, the team will include three WPI graduate students and consultants Peder Pedersen, WPI emeritus professor of electrical and computer engineering; Dr. Raymond Dunn '78, professor of surgery at UMass Chan; and wound nurse Lorraine Loretz.
The same research team previously developed an app that analyzed photos of patients’ wounds to recommend treatment options. The new project expands on the work by adding thermal imagery of wounds, an important addition because changes in wound temperatures have been linked to the skin ulceration, insufficient oxygen flow to a site, inflammation, and infection.
Chronic wounds that do not heal well represent a significant health issue, accounting for millions of emergency visits to U.S. hospitals every year and costing Medicare an estimated $31.7 billion a year. Patients whose wounds lead to amputation also have a heightened risk of death, Agu said.
“The cost of failing to rapidly diagnose and treat an infected wound is huge,” Agu said. “Infections can lead to amputations, and amputations can lead to a loss in mobility, which ultimately results in obesity and other conditions. Organs may begin to deteriorate, and complications can lead to death. Many of the patients whose wound photographs were used to train our machine learning models have since passed away. It’s sobering, and we want to address this problem.”
Read More About Research from Emmanuel Agu, Peder Pedersen, Diane Strong, and Bengisu Tulu
Experts at WPI
WPI Expert Database
Faculty members and senior administrators are available to offer ideas, opinions, analysis, and commentary on issues ranging from higher education to current events, trending topics, and breaking news.
Keep Up With WPI Research News
Want to learn more about Research at WPI? Subscribe to the monthly newsletter.