Simulation of a MR-PET protocol for staging of head-and-neck cancer including Dixon MR for attenuation correction

Purpose: To simulate and optimize a MR protocol for squamous cell cancer of the head and neck (HNSCC) patients for potential future use in an integrated whole-body MR-PET scanner. Materials and methods: On a clinical 3T scanner, which is the basis for a recently introduced fully integrated whole-body MR-PET, 20 patients with untreated HNSCC routinely staged with 18F-FDG PET/CT underwent a dedicated MR protocol for the neck. Moreover, a whole-body Dixon MR-sequence was applied, which is used for attenuation correction on a recently introduced hybrid MR-PET scanner. In a subset of patients volume-interpolated-breathhold (VIBE) T1w-sequences for lungs and liver were added. Total imaging time was analyzed for both groups. The quality of the delineation of the primary tumor (scale 0-3) and the presence or absence of lymph node metastases (scale 1-5) was evaluated for CT, MR, PET/CT and a combination of MR and PET to ensure that the MR-PET fusion does not cause a loss of diagnostic capability. PET was used to identify distant metastases. The PET dataset for simulated MR/PET was based on a segmentation of the CT data into 4 classes according to the approach of the Dixon MR-sequence for MR-PET. Standard of reference was histopathology in 19 cases. In one case no histopathological confirmation of a primary tumor could be achieved. Results: Mean imaging time was 35:17 min (range: 31:08-42:42 min) for the protocol including sequences for local staging and attenuation correction and 44:17 min (range: 35:44-54:58) for the extended protocol. Although not statistically significant a combination of MR and PET performed better in the delineation of the primary tumor (mean 2.20) compared to CT (mean 1.40), MR (1.95) and PET/CT (2.15) especially in patients with dental implants. PET/CT and combining MR and PET performed slightly better than CT and MR for the assessment of lymph node metastases. Two patients with distant metastases were only identified by PET. Conclusion: We established a potential MR-protocol to be used for HNSCC patients in a recently introduced MR-PET scanner. The proposed protocol can be performed in an acceptable time frame and did not lead to a loss of diagnostic capability compared to PET/CT.