Project MATCH: Multi-spectral AI/ML powered Burn and Wound Evaluation Technology using a Cost-Effective Handheld Device.

ABSTRACT
Annually, around 11 million individuals suffer burn injuries, with approximately 180,000 fatalities. In the U.S.
over 450,000 individuals seek medical treatment for burns each year, resulting in approx. 3,275 deaths. Early
and accurate assessment of burn severity is crucial for patient care, especially when under surgical consideration.
Inaccurate burn assessments, particularly in depth and area, can have severe consequences. Overestimation may
lead to unnecessary surgeries and prolonged recovery, while underestimation can result in premature discharge,
increasing the risk of infection and complications like hypertrophic scarring, which impairs mobility. Visual
clinical assessment by experienced burn surgeons is the current gold standard, but studies show varying accuracy
levels.
Laser Doppler Imaging (LDI), the most widely accepted technology for assessment is costly, has low usability,
and lacks widespread adoption. Efforts to create alternatives with hyperspectral, and laser speckle imaging
technologies, have also faced adoption issues as these devices remain large and expensive. The device proposed
by this project addresses current market gaps by being smaller, more affordable, and user-friendly compared to
existing LDI devices. Its portability brings advanced imaging directly to the point-of-care, facilitating immediate
and precise burn evaluations.
To achieve this goal Perfusio Corp. will utilize their Multi-Spectral Physiologic Visualization (MSPV) technology,
a platform that delivers real-time dynamic physiologic data to medical providers. This technology has been
implemented via the FDA approved clinical form factor, CERTES™, designed for open surgical applications. This
device provides OR surgeons with blood flow distribution information as an indicator of tissue integrity. MSPV
offers three main feature sets: perfusion information, 2D peripheral oxygen saturation, and physiologic status
parameters derived from imaging metadata. While these features are valuable in burn cases, the current
CERTES™ device is too large and costly, falling into the same issues as current LDI devices.
Building upon the MSPV platform, Perfusio will utilize Machine Learning (ML) to produce burn depth
classification, artifact segmentation, tissue viability and the total body surface area (TBSA). These model outputs
will deliver intuitive interpretation of information and insights to facilitate better clinical decision making.
Perfusio has analyzed the burn patient journey to ensure this project meets critical needs. The initial focus is on
enhancing the accuracy and efficiency of the initial triage process, where misdiagnosis is most prevalent. By
providing objective data on TBSA and burn depth, this device empowers healthcare providers to initiate optimal
treatment plans early on. Through this project Perfusio will develop a proof-of-concept prototype solution for
burn wound triage and assessment, that provides actionable insights, improving burn care standards and
techniques in all environments.

Grant Number: 1R43GM156148-01A1
NIH Institute/Center: NIH
Principal Investigator: Cheng Chen