Abstract: Embodiments include obtaining an input image depicting a body part of a person and processing the input image against a set of semantic landmarks representing landmarks of the body part; obtaining a mesh model for a set of images; generating, from the mesh model and the set of semantic landmarks, a body part mesh of the person, wherein the body part mesh is an approximation of a 3D model for the body part depicted in the input image; obtaining a target body part mesh data structure, distinct from the body part mesh; and generating a modified view image of the body part, modified to reflect differences between the target body part mesh data structure and the body part mesh while retaining at least some texture of the body part from the input image.

Abstract: A medical image might be generated for a patient image using a convolutional neural network trained on prior patient pre-procedure and post-procedure 2D images. A method might generate 3D models from pre-procedure 2D images and from post-procedure 2 images, and train the convolutional neural network with training images being the 3D models, to generate at least one 3D model of the present patient from the 2D image of the present patient, apply the 3D model of the present patient to the convolutional neural network in an inference stage, and apply patient-specific parameters derived from the proposed surgical procedure as a second input to the convolutional neural network in the inference stage to generate an inferred post-surgery 3D model of the present patient given the patient-specific parameters.

Abstract: A medical image might be generated for a patient image using a convolutional neural network trained on prior patient pre-procedure and post-procedure 2D images. A method might generate 3D models from pre-procedure 2D images and from post-procedure 2 images, and train the convolutional neural network with training images being the 3D models, to generate at least one 3D model of the present patient from the 2D image of the present patient, apply the 3D model of the present patient to the convolutional neural network in an inference stage, and apply patient-specific parameters derived from the proposed surgical procedure as a second input to the convolutional neural network in the inference stage to generate an inferred post-surgery 3D model of the present patient given the patient-specific parameters.