Abstract: Metal artifacts are reduced in x-ray computed tomography (“CT”) images using a suitably trained neural network, such as a convolutional neural network (“CNN”). Virtual metal DATA objects are inserted to either the raw projection data or CT image data (e.g., from pre-procedural CT scans) to generate sets of matching artifact-corrupted and artifact-uncorrupted images, and a CNN, or other neural network, is trained to separate the contribution to each image pixel due to patient anatomy, metal object, or metal object-induced artifact. The contributions from metal object-induced artifacts can then be removed to generate a final, artifact-reduced image.

Abstract: Images are reconstructed from data acquired using an ultra-fast-pitch acquisition with a CT system. As an example, an ultra-fast-pitch acquisition mode in single-source helical CT (p?1.5) can be used to acquire data. A trained machine learning algorithm, such as a neural network, is used to reconstruct images in which artifacts associated with insufficient data acquired in the ultra-fast-pitch mode are reduced. An example neural network can include customized functional modules using both local and non-local operators, as well as the z-coordinate of each image, to effectively suppress the location- and structure-dependent artifacts induced by the ultra-fast-pitch mode. The machine learning algorithm can be trained using a customized loss function that involves image-gradient-correlation loss and feature reconstruction loss.