Digital Thermography and Machine Learning Team Up to Improve Burn Wound Care

A team at McGill University in Canada and the Universidad Autonoma de San Luis Potosi in Mexico have developed a system for analyzing thermographic scans of burn wounds to improve how they are analyzed and how patients are treated.

The team used digital infrared thermography, a non-invasive imaging technique, to study wounds when they were presented and for the following few days afterward. Infrared thermography shows the heat signature of the tissues being observed, which can indicate a variety of underlying processes taking place, particularly the amount of blood perfusion below the surface. “Digital infrared thermography allows us to visualize the heat emitted by objects,” said Dr. José Luis Ramírez García Luna, PhD candidate at McGill and the paper’s corresponding author. “In the human body, heat emission depends on blood flow to the tissues, thus it is an indirect measurement of blood perfusion to the skin or a wound bed. Previous research has shown that the temperature of the skin correlates well with blood flow to it and that there are differences in the temperature pattern of healing vs. non-healing wounds.”

As part of their study, the researchers developed a clinically useful prediction rule that can be used to decide what treatment plans to put into effect. This was based on performing thermographic scans of wounds for the first few days and letting conventional therapeutic decisions to take place. The outcomes were then studied and compared against the thermographic scans to see which treatments fared best for the different wound types. This machine learning method has led to a nearly 90% accuracy rate at predicting how to best treat different wounds, according to the researchers. Of course more study will be required of this new method to further validate it, but it’s certainly encouraging to see the effects of burn wounds being minimized with advanced computational technologies.

Study in PLOS ONE: Development and validation of an algorithm to predict the treatment modality of burn wounds using thermographic scans: Prospective cohort study…

Via: McGill University…