DOI: 10.1007/s00259-017-3864-4Pages: 642-653

Impact of improved attenuation correction featuring a bone atlas and truncation correction on PET quantification in whole-body PET/MR

1. University Hospital Essen, High Field and Hybrid MR Imaging

2. University Duisburg-Essen, Erwin L. Hahn Institute for MR Imaging

3. University Dusseldorf, Department of Diagnostic and Interventional Radiology, Medical Faculty

4. University Hospital Essen, Department of Nuclear Medicine

5. University Hospital Essen, Department of Diagnostic and Interventional Radiology and Neuroradiology

6. Siemens Healthcare GmbH

Correspondence to:
Mark Oehmigen
Tel: +49 201/723-84534




Recent studies have shown an excellent correlation between PET/MR and PET/CT hybrid imaging in detecting lesions. However, a systematic underestimation of PET quantification in PET/MR has been observed. This is attributable to two methodological challenges of MR-based attenuation correction (AC): (1) lack of bone information, and (2) truncation of the MR-based AC maps (μmaps) along the patient arms. The aim of this study was to evaluate the impact of improved AC featuring a bone atlas and truncation correction on PET quantification in whole-body PET/MR.


The MR-based Dixon method provides four-compartment μmaps (background air, lungs, fat, soft tissue) which served as a reference for PET/MR AC in this study. A model-based bone atlas provided bone tissue as a fifth compartment, while the HUGE method provided truncation correction. The study population comprised 51 patients with oncological diseases, all of whom underwent a whole-body PET/MR examination. Each whole-body PET dataset was reconstructed four times using standard four-compartment μmaps, five-compartment μmaps, four-compartment μmaps + HUGE, and five-compartment μmaps + HUGE. The SUVmax for each lesion was measured to assess the impact of each μmap on PET quantification.


All four μmaps in each patient provided robust results for reconstruction of the AC PET data. Overall, SUVmax was quantified in 99 tumours and lesions. Compared to the reference four-compartment μmap, the mean SUVmax of all 99 lesions increased by 1.4 ± 2.5% when bone was added, by 2.1 ± 3.5% when HUGE was added, and by 4.4 ± 5.7% when bone + HUGE was added. Larger quantification bias of up to 35% was found for single lesions when bone and truncation correction were added to the μmaps, depending on their individual location in the body.


The novel AC method, featuring a bone model and truncation correction, improved PET quantification in whole-body PET/MR imaging. Short reconstruction times, straightforward reconstruction workflow, and robust AC quality justify further routine clinical application of this method.

To access the full text, please Sign in

If you have institutional access, please click here

  • Accepted: Oct 19, 2017
  • Online: Nov 9, 2017

Article Tools