Validating an imaging and analysis system for assessing torso deformities

We present the results of the numeric and functional validation of an imaging and analysis system used for assessing human torsos for deformities such as scoliosis. The system comprises of image acquisition, image reconstruction, and shape analysis components. The numeric validation procedure consists of assessing the accuracy of reconstruction of the system using inanimate models (a calibration box and a mannequin). The functional validation involves determining the system's response to variations in shape caused by sway and breathing, and evaluating the variability of a clinically relevant index, the Cosmetic Score, from multiple scans of scoliosis and non-scoliosis volunteers. Results show that the reconstruction accuracy of the system is 1.16+/-1.04 mm. This is better than the required accuracy for monitoring scoliosis of 2 mm. The system is robust to shape variations caused by sway and breathing and shows limited variability to the Cosmetic Score.