Related Experiment Video
Updated: Jan 10, 2026
Author Spotlight: A Multi-Depth Porcine Model for Comprehensive Study of Burn Injuries and Healing Processes
Published on: February 23, 2024
Advancements in burn size assessment: A systematic review of emerging technologies
Amy Woods1, Poh Tan1, Tamara Mertz1
1Northern Regional Burn Centre, Royal Victoria Infirmary, Newcastle upon Tyne NE1 4LP, UK.
Background:
Accurate and reproducible assessment of total body surface area (TBSA) in burns is vital for guiding fluid resuscitation, triage, referral decisions, and overall clinical management. Traditional methods are widely used but prone to significant inter-rater variability and diagnostic inaccuracies.
Objective:
This systematic review evaluates emerging technologies developed to improve TBSA estimation, focusing on diagnostic accuracy and inter-rater reliability.
Methods:
A comprehensive search of EMBASE, MEDLINE (OVID), Web of Science, Scopus, PubMed and Cochrane Library databases was conducted. Studies assessing technological tools for TBSA estimation in both human and artificial burns were included. Primary outcomes were accuracy and reliability, while secondary outcomes included speed and user feedback. Study quality was appraised using the QUADAS-2 tool.
Results:
Thirty-six studies were included: 3D programs (n = 7), mobile applications (n = 11), 3D stereophotogrammetry (n = 8), and machine learning models (n = 10). 3D stereophotogrammetry showed the highest accuracy (mean ICC = 0.988) and excellent inter-rater reliability (ICC = 0.989). 3D programs demonstrated good diagnostic performance and reduced variability. Mobile applications improved accuracy and consistency, particularly among non-specialists, and offered practical benefits in prehospital settings. Machine learning, though largely in experimental phases, demonstrated promising accuracy, with some models outperforming clinician estimates.
Conclusion:
3D stereophotogrammetry provided more consistent and reliable estimates of TBSA, while mobile applications offered practical and scalable solutions. Early machine learning approaches showed potential but remain largely experimental. Overall, most studies were of low quality and methodologically heterogeneous; therefore, definitive conclusions cannot yet be drawn. Further high-quality, real-world research is required to validate accuracy and establish clinical utility.
