Synthetic computed tomography: An artificial contrast for magnetic resonance imaging

Abstract

The musculoskeletal (MSK) system provides support, stability and movement to the body. Disorders to the MSK system cause pain and stiffness, and can debilitate anatomical regions, hinder movement, or result in disability. As such, they are among the leading cause of activity limitation and absence from work and result in a major medical burden and economic cost. Treatment and recommendations for diseases affecting the MSK system are very dependent on the type of disorder, the involvement of the affected soft tissues and bones, and the progression of the disease. In this context, medical imaging is crucial to localize a disorder, assess the stage of disease progression, support a diagnosis, or plan a treatment. For a comprehensive assessment and treatment planning of MSK pathologies, acquisition of both magnetic resonance imaging (MRI) and computed tomography (CT) is warranted, which is logistically complex and constitutes a high patient burden. This work discusses the use of synthetic CT (sCT), a deep learning-enabled CT-like image generated from MR images, as a CT surrogate. Synthetic CT has been thoroughly investigated, from the acquisition and processing of the original MRI data to its clinical application, with a focus on orthopaedic care. Synthetic CT has been found acceptable for bone visualization, simultaneously providing valuable information on soft tissues via the original MR acquisition. This paves the way to a purely MRI-based workflow with simplified logistics and radiation-free acquisition, and opens up new perspectives for diagnosis, treatment planning and disease understanding.