Program Director/Principal Investigator (Last, First, Middle): Durkin, Anthony J. Abstract The central aim of this 3 year competing R01 renewal is to characterize and apply a new, compact, clinic- friendly Spatial Frequency Domain Imaging (SFDI) device to objectively and non-invasively classify burn severity (burn grade) over a large areas of skin. Delays in determining burn severity directly impacts patient treatment plans (including decisions whether to graft), rates of infection and scarring, duration of hospitalization and ultimately cost of care. Currently, the primary method of determining burn severity continues to be clinical assessment, which is highly subjective. While both superficial thickness and full-thickness burns are typically readily diagnosed based on visual clinical impression, partial thickness burns are difficult to classify and carry with them considerable potential for complications. Burn severity classification accuracy, even by experts, is only 60–80%. Our research in animal models demonstrates that SFDI data can successfully be used to classify different regions of burn severities. Typically, these differences are not apparent to the unaided eye and a great deal of training and experience is required in order for clinicians to accurately differentiate them Our work using a research grade, hybrid-SFDI device suggests that objective parameters provided by SFDI can be used within 24 hours after injury, to accurately classify burn severity. Specifically, we have demonstrated in a porcine burn model that the research grade SFDI outperforms laser speckle imaging and thermal imaging at 24 hours post-burn, in terms of predicting whether a burn will require a graft or not. However, translating these results to the clinic has been difficult due to several device limitations. The research grade SFDI device has slow acquisition times that can result in motion artifacts. It is also sensitive to ambient light which is often an issue in a clinical setting. Additionally, the SFDI device generates so much diverse data (oxygenated and deoxygenated hemoglobin, water fraction, reduced scattering coefficients at multiple wavelengths), there is no obvious way to present it to a clinical user to make a quick decision. To this end, we propose to methodically investigate an improved next generation SFDI device that addresses these issues by using brighter LEDs and fewer wavelengths to rapidly collect data in a way that reduces motion artifacts and is independent of clinical lighting conditions. In addition, we will develop a machine learning based classification framework that will provide the clinical with actionalble diagnostic information. The central aim of this 3 year competing R01 renewal is to characterize and then modify a new clinic-friendly SFDI device (Clarifi) to objectively classify in- vivo regions of different burn severity over large areas. The proposed research seeks to investigate this via the following Specific Aims: 1) Test ...