Co-registration method of wearable magnetocardiography system and CT

Wearable magnetocardiography (MCG) systems offer advantages such as flexible sensor distribution and higher signal strength than that of traditional rigid cardiac systems. However, the flexible nature of wearable cardiac magnetic systems presents challenges in acquiring accurate optically pumped magnetometer sensor positions and orientations. In this study, we developed a sensor-positioning indicator cap (SPIC) system. The SPIC used point-cloud data acquired through a structured light scanner to extract the three-axis orientations and positions of the sensors. Additionally, we proposed a novel combined branch and bound iterative closest point algorithm, which eliminates the need for initial alignment. Without requiring additional fiducial markers on the subject's body, the sensors were registered in the computed tomography (CT) coordinate system based on the clavicular region. We designed and fabricated a reference phantom with known sensor positions and orientations to evaluate the accuracy of our method. Experimental results demonstrated that after registration, the average position error for all sensors <1mm, with an average orientation error < 1°. In addition, we conducted simulations to assess the impact of registration errors on MCG source localization. The results indicated that errors resulting from our proposed registration method would introduce an approximately 1 mm error in source localization. This study introduces a registration method facilitating the reusability of CT data in wearable MCG systems, thereby benefiting research in MCG source localization.