Can Burn AI Assess the Head, Hands, and Feet? DeepView Launches UK Study to Expand Its Use

A key question in burn care is which wounds can heal on their own and which may require further intervention. When wounds involve the head and face, hands, or feet, the anatomy is more complex, making early assessment of healing potential especially important. DeepView, which combines multispectral imaging with artificial intelligence, is testing through a UK study whether its assessment capabilities can extend to these areas.

Spectral AI announced on October 1 that it had completed the first study site initiation visit at Pinderfields Hospital in Wakefield, UK, preparing the site for enrollment. The medical trade publication Plastic Surgery Practice reported the same development on October 5. This prospective study plans to enroll adults and children with thermal burns at multiple UK centers; the announcement marks a study launch milestone, with no information yet provided on enrollment of the first participant or diagnostic performance results.

DeepView uses multispectral imaging to capture wound information, which an AI model then assesses to identify areas that may not heal within 21 days, supporting physicians’ judgment. According to the company announcement, the system received De Novo classification authorization from the US Food and Drug Administration in May 2026. This study aims to build evidence for expanding its use to the head, hands, and feet; the existing authorization cannot be taken to mean that these additional areas have already been approved.

The ICH GCP mirror of the ClinicalTrials.gov record under registration number NCT07677657 lists an expected enrollment of 280 participants. Participants will undergo initial imaging within 72 hours of injury and will be followed through the healing assessment on day 21. New images will be used separately for model training and independent validation, and will also be combined with existing data from other body areas to assess the updated model’s performance.

The images also require a reliable reference standard. According to the registry entry, a panel of burn surgeons will establish a benchmark against which to check the model’s assessments; for study wounds requiring surgical excision, the protocol also includes skin punch biopsy sampling. Enrollment has limits on its scope: for example, chemical burns are not included, and burns confined to the surface of the palms or soles are also excluded. The results therefore cannot be directly generalized to all wounds of the hands and feet.

Both the company announcement and registry information state that DeepView will be used in a blinded manner during the study, collecting observational data only and taking no part in care decisions. This design helps test whether predictions agree with actual wound outcomes, but cannot directly demonstrate that using AI improves treatment outcomes. Sensitivity, specificity, and separate validation results for adults and children for the new areas have also not yet been released.

The company expects the study to be completed in 2027; an earlier version of the registry entry estimated the primary completion date as April 1, 2027, and the overall completion date as May 1, 2027. Both are estimated timelines. The recruitment status in that version predates this launch announcement and cannot be used to determine the current enrollment situation. The questions that now need to be answered are whether independent validation for the additional areas can support expanded use and whether, after authorization is obtained, these predictions can provide measurable benefits in actual care.