Radiodermatitis grade estimation by RGB color imaging

Radiodermatitis is visually evaluated by the Radiation Therapy Oncology Group (RTOG) scoring scheme, using characteristics such as erythema, desquamation, moist and bleeding. However, subjectivity and differences in skin types and melanin content may bias interpretation. This paper describes the use of RGB cameras for radiodermatitis estimation using image processing. We imaged radiodermatitis evolution throughout the treatment of 23 breast cancer radiotherapy patients of all Fitzpatrick skin phototypes. To prevent confounding information from skin fluids and skin reflection, we used cross-polarized imaging. RGB intensity was corrected using white medical tape as a reference. The RGB color as a function of RTOG grade depended strongly on skin phototype. Yet, when patients are grouped into white, brown, and black skin, the normalized RGB colors reveal stable characteristic signatures that uniquely predict RTOG grade for each group. We conclude cross-polarized RGB imaging as proposed is viable to document and estimate radiodermatitis on all skin phototypes.