Reliability and Agreement of 3D Anthropometric Measurements in Facial Palsy Patients Using a Low-Cost 4D Imaging System

The reliability (precision) and agreement (accuracy) of anthropometric measurements based on manually placed 3D landmarks using the RealSense D415 were investigated in this paper. Thirty facial palsy patients, with their face in neutral (resting) position, were recorded simultaneously with the RealSense and a professional 3dMD imaging system. First the RealSense depth accuracy was determined. Subsequently, two observers placed 14 facial landmarks on the 3dMD and RealSense image, assessing the distance between landmark placement. The respective intra- and inter-rater Euclidean distance between the landmark placements was 0.84 mm (±0.58) and 1.00 mm (±0.70) for the 3dMD landmarks and 1.32 mm (±1.27) and 1.62 mm (±1.42) for the RealSense landmarks. From these landmarks 14 anthropometric measurements were derived. The intra- and inter-rater measurements had an overall reliability of 0.95 (0.87 - 0.98) and 0.93 (0.85 - 0.97) for the 3dMD measurements, and 0.83 (0.70 - 0.91) and 0.80 (0.64 - 0.89) for the RealSense measurements, respectively, expressed as the intra-class correlation coefficient. Determined by the Bland-Altman analysis, the agreement between the RealSense measurements and 3dMD measurements was on average -0.90 mm (-4.04 - 2.24) and -0.89 mm (-4.65 - 2.86) for intra- and inter-rater agreement, respectively. Based on the reported reliability and agreement of the RealSense measurements, the RealSense D415 can be considered as a viable option to perform objective 3D anthropomorphic measurements on the face in a neutral position, where a low-cost and portable camera is required.