3D Reconstruction of Large-scale Head Surface with a Binocular Camera in Transcranial Magnetic Stimulation Navigation

In image-guided TMS navigation, markerless registration methods use the head surface to match the tracking system space with the image space. Large-scale head surface is beneficial for improving registration accuracy. Currently, most 3D cameras have a limited view field and cannot reconstruct large-scale head surface with a single shot. Extra tracking devices are needed to stitch several single-shot 3D surfaces into a large-scale one. To simplify the above point cloud stitching process, an automatic 3D reconstruction method for awake TMS patients is proposed to obtain head surfaces without extra tracking devices. Experimental results show the accuracy of the reconstruction can reach 1.39 mm and registration is quicker and simpler in repeated TMS treatment sessions.