DOI: 10.1186/s40658-017-0177-4Pages: 1-23

MLAA-based attenuation correction of flexible hardware components in hybrid PET/MR imaging

1. Medical Physics in Radiology, German Cancer Research Center (DKFZ)

2. Department of Radiology, University of Pennsylvania

3. RWTH Aachen University, Physics of Molecular Imaging Systems

4. German Cancer Research Center (DKFZ), Department of Radiology

Correspondence to:
Thorsten Heußer
Email: thorsten.heusser@dkfz.de

Close

Abstract

Background

Accurate PET quantification demands attenuation correction (AC) for both patient and hardware attenuation of the 511 keV annihilation photons. In hybrid PET/MR imaging, AC for stationary hardware components such as patient table and MR head coil is straightforward, employing CT-derived attenuation templates. AC for flexible hardware components such as MR-safe headphones and MR radiofrequency (RF) surface coils is more challenging. Registration-based approaches, aligning CT-based attenuation templates with the current patient position, have been proposed but are not used in clinical routine. Ignoring headphone or RF coil attenuation has been shown to result in regional activity underestimation values of up to 18%.

We propose to employ the maximum-likelihood reconstruction of attenuation and activity (MLAA) algorithm to estimate the attenuation of flexible hardware components. Starting with an initial attenuation map not including flexible hardware components, the attenuation update of MLAA is applied outside the body outline only, allowing to estimate hardware attenuation without modifying the patient attenuation map. Appropriate prior expectations on the attenuation coefficients are incorporated into MLAA. The proposed method is investigated for non-TOF PET phantom and 18F-FDG patient data acquired with a clinical PET/MR device, using headphones or RF surface coils as flexible hardware components.

Results

Although MLAA cannot recover the exact physical shape of the hardware attenuation maps, the overall attenuation of the hardware components is accurately estimated. Therefore, the proposed algorithm significantly improves PET quantification. Using the phantom data, local activity underestimation when neglecting hardware attenuation was reduced from up to 25% to less than 3% under- or overestimation as compared to reference scans without hardware present or to CT-derived AC. For the patient data, we found an average activity underestimation of 7.9% evaluated in the full brain and of 6.1% for the abdominal region comparing the uncorrected case with MLAA.

Conclusions

MLAA is able to provide accurate estimations of the attenuation of flexible hardware components and can therefore be used to significantly improve PET quantification. The proposed approach can be readily incorporated into clinical workflow.

This article is freely available, click here to access the full text/PDF

  • Accepted: Feb 1, 2017
  • Online: Mar 1, 2017

Article Tools

eanm
EJNMMI Ad