A-Mode Ultrasound Bone Registration for Computer-Assisted Knee Surgery: Calibration and Robustness Test

Abstract

Computer-assisted orthopedic surgery (CAOS) has been widely used in clinical procedures because of its high accuracy. Conventional methods for bone registration, which is crucial in CAOS, are performed by utilizing mechanical probes. One of the drawbacks of mechanical probe is that it requires extensive incisions to access the bone. We investigated a noninvasive A-mode ultrasound (US) probe instead of conventional mechanical probes for bone registration in computer-assisted orthopedic knee surgery. This paper represents the first stage of the study, focusing on US probe calibration and robustness test. A sphere phantom in water bath was used for the US calibration. Three-dimensional coordinates of mechanical and US probes were tracked by an optical tracking system. The location of the transmitting/sensing element inside the US transducer housing was calibrated based on the Levenberg–Marquardt algorithm. Registration accuracy was evaluated for both the mechanical probe and US probe before and after calibration. After five calibrations, each of which with thirty different collected sampling points, the mean error of the mechanical probe was 0.1969 ± 0.1087 mm. In the same manner, the mean error of the US probe was 0.7496 ± 0.0550 mm before calibration and 0.4046 ± 0.1017 mm after calibration. Our results indicated that the registration error of the US probe significantly reduced after calibration and was within the acceptable range. Our results also suggested that the US probe is comparable to the mechanical probe. With further modifications, noninvasive A-mode US probe, instead of conventional mechanical probes, may be utilized for bone registration in computer-assisted knee surgery.