Objective facial paralysis grading based on Pface and eigenflow.

To provide physicians with an objective and quantitative measurement of single-sided facial paralysis, the paper presents a computer-based approach that is different from the nine existing, subjective and hand-performed international scales, such as House-Brackman. For voluntary expressions of a patient, this approach used Pface, which stems from Dface, to measure the asymmetry between two sides of the face and used eigenflow to measure the expression variations between the patient and normal subjects. The results from Pface and eigenflow were then combined by the support vector machine produce to Pdegree. A study of 25 subjects revealed that Pdegree could differentiate paralysis states (Pdegree > or = 0) and normal states (Pdegree < 0), with the ability to grade facial paralysis automatically. Moreover, the Pface of specific facial areas can be used in the supervision of the rehabilitation process.