| abstract |
A method includes accessing ultrasound data encoding 3D surface information of one or more body structures in a region, extracting a first 3D surface representation of the one or more body structures based on the ultrasound data, accessing non-ultrasound imaging data obtained from performing non-ultrasound imaging of the region; extracting a second 3D surface representation of the one or more body structures based on the non- ultrasound imaging data; identifying, using a deep learning algorithm, one or more regions of interest (ROI)s from the first 3D surface representation such that when aligning the first 3D surface representation and the second 3D surface representation over the one or more regions of interest (ROI)s, a spatial transformation is determined; and based on applying the spatial transformation, generating an overlay of the 3D surface anatomical information from the non-ultrasound imaging data on the 3D surface information from the ultrasound data. |