On the generation of synthetic CT for a MRI-only radiation therapy workflow for the abdomen

MRI is an important imaging modality for contouring in radiotherapy due to its superior soft tissue contrast over CT, but it is not often used alone in a radiation therapy workflow because of the lack of electron density information which is required for dose calculations. This study investigates a method to generate synthetic CT from MRI to demonstrate the potential of using MRI alone in the radiation therapy workflow for abdominal targets. This method consists of (1) acquiring multiple MRI volumes, (2) applying intensity nonuniformity corrections to MR images, (3) identifying abdominal walls to distinguish air from bone and vertebral bodies to improve classifications between bone marrow and muscle tissues, (4) classifying tissues in each section using fuzzy c-means clustering with a spatial constraint and (5) assigning attenuation properties to individual tissue classes to generate synthetic CT. The quality of the synthetic CT was comparable to clinical CT. The proposed method allows automatic segmentation of tissue and generation of synthetic CT for the abdomen. Our results demonstrate the potential of using MRI alone to support treatment planning and image guidance for abdominal radiation therapy.