Correlation between structural and color changes in 3D facial images of head and neck cancer patients following reconstructive surgery

Purpose Treatment for head and neck cancer can impair speech and swallowing, and can disfigure the face. Reconstructive surgery can restore some facial function and ameliorate disfigurement, thereby helping patients to psychosocially adjust. However, current surgical reconstruction techniques may not fully restore a patient's pre-cancerous appearance. Therefore, it is important to understand the aspects of disfigurement that persist after surgery, since this could be used to improve the reconstruction process. Efforts towards developing improved reconstruction approaches would be greatly augmented by suitable disfigurement models. For example, by simulating surgically plausible facial disfigurements that may arise from reconstructive surgery, surgeons might be better able to estimate how the reconstructive process affects the patient's appearance. This would assist the physician in determining which reconstruction options are most appropriate for the patient. Recently, 3D imaging techniques have been deployed for planning and evaluating reconstructive surgery [1]. Unlike traditional 2D clinical photographs, 3D images preserve the 3D structure of the face in addition to color information. However, the use of 3D imaging for surgical planning has been limited to the analysis of the bony structure of the face. The development of methods for analyzing changes of the soft tissue (an important aspect of facial disfigurement) that would be consequent to different reconstructive options (such as the choice of autologous tissue) remains in a nascent stage of research. Modeling facial morphology using 3D images has been widely studied in computer graphics (for rendering faces) and in computer vision (for recognizing faces). Beyond this, we believe that 3D face modeling has the potential to be used for simulating surgically plausible facial disfigurements. However, there are many new challenges to be faced in this realm that are not addressed by face rendering or recognition algorithms, such as analyzing and quantifying the changes in the face that can occur as a result of surgery or radiation therapy. These changes can be manifested in both the bony and soft tissues of the face. To develop a facial disfigurement model that is surgically plausible, it is necessary to understand the relationship between the structural and color components of faces that are disfigured by reconstructive surgery. Thus, we evaluated the relationship between change in the structural and color components of the face of patients suffering from head and neck cancer following reconstructive surgery. Methods A total of 24 3D facial images of 12 patients currently undergoing reconstructive surgery on their faces …