System and method for artifact reduction of computed tomography reconstruction leveraging artificial intelligence and a priori known model for the object of interest

Inventors

Ziabari, Amir • Venkatakrishnan, Singanallur • Bingham, Philip R. • Kirka, Michael M. • Paquit, Vincent C. • Dehoff, Ryan R. • Dubey, Abhishek

Assignees

UT Battelle LLC

Abstract

Nondestructive evaluation (NDE) of objects can elucidate impacts of various process parameters and qualification of the object. Computed tomography (CT) enables rapid NDE and characterization of objects. However, CT presents challenges because of artifacts produced by standard reconstruction algorithms. Beam-hardening artifacts especially complicate and adversely impact the process of detecting defects. By leveraging computer-aided design (CAD) models, CT simulations, and a deep-neutral network high-quality CT reconstructions that are affected by noise and beam-hardening can be simulated and used to improve reconstructions. The systems and methods of the present disclosure can significantly improve the reconstruction quality, thereby enabling better detection of defects compared with the state of the art.

Core Innovation

The invention provides an artifact reduction artificial intelligence training system for computed tomography (CT) of an object of interest. The system includes a computer-aided design (CAD) model representing the object of interest stored in memory, and an artifact characterization stored in memory. The artifact characterization provides a set of beam-hardening parameters and detector noise parameters for simulating artifacts caused by beam hardening and detector noise during an actual CT scan.

A CT simulator generates a plurality of CT simulated projections based on the CAD model, using the beam-hardening parameters and the detector noise parameters. A subset of the plurality of CT simulated projections includes simulated artifacts, while a subset lacks simulated artifacts. The generated artifact CT simulated projections are realistically noisy, and the simulation is configured to produce corresponding artifact CT simulated projections based on the CAD model and the artifact characterization.

A deep learning component trains a deep learning artifact reduction model based on the plurality of CT simulated projections and generates a set of deep learning artifact reduction model parameters. The deep learning training uses both artifact CT simulated projections and non-artifact CT simulated projections, so the deep learning artifact reduction model parameters are generated for artifact reduction based on the simulated noisy artifact data.

Claims Coverage

The document includes two independent claims: an artifact reduction artificial intelligence training system for CT and an artifact reduction artificial intelligence training method for CT. Both independent claims center on generating realistically noisy artifact and non-artifact CT simulated projections from a CAD model using artifact characterization, then training a deep learning artifact reduction model to generate model parameters.

Across the independent claims, the inventive content is directed to a CT artifact reduction AI training approach that uses a CAD model plus artifact characterization with beam-hardening parameters and detector noise parameters to generate realistically noisy artifact and non-artifact CT simulated projections, and trains a deep learning artifact reduction model on the simulated artifact and non-artifact projection subsets to generate deep learning artifact reduction model parameters.

Stated Advantages

Documented Applications