ZTE-based clinical bone imaging for PET/MR

Objectives One of the main limitations of hybrid PET/MR is the difficulty of imaging cortical bone with MR. Several methods have been published [1-4] to overcome this limitation. Among the most popular is the combination of short-echo-time MR sequences with dedicated post-processing to segment cortical bone. The goal of these methods is generally the creation of attenuation maps for PET reconstruction, rather than clinically usable images. In this study, we evaluate a method based on zero-echo-time (ZTE) acquisitions for the purpose of clinical imaging of the skeletal system.

Methods Twenty examinations of joints were performed on healthy volunteers using a Discovery 750w MR (GE Healthcare). Based on a recently published bone attenuation correction method [5], a customized version of the SilentMR sequence was acquired in each case. The geometry and coil varied with the imaged anatomy, but acquisition times were kept around 3min, with resolution ≈1.2mm³, FA 1deg, BW ±62.5kHz. Data were automatically equalized and rescaled for bone visualization. The resulting images were evaluated by two experienced radiologists.

Results A typical PDw ZTE acquisition is depicted in fig. 1. Clinically, musculoskeletal structures such as cortical bone and tendons are clearly depicted, with thinner epiphysis layers visible in 2D views but requiring some interaction in rendering views. The main limitation of the method is local coil sensitivity variation, due to the PDw acquisition, which confines target bone structures to a narrow intensity range. Adequate intensity equalization is therefore mandatory. This can be successfully implemented for regions delimited by clear air-tissue interfaces, but remains challenging in unbounded acquisitions (e.g. hip or shoulder)

Conclusions Our results prove that clinically usable images of the skeletal system can be obtained with ZTE MR. Future work with tri-modality data will validate these results versus CT. The technique will then be adapted for the identification PET/MR bone lesions.