Objectives: Simultaneous acquisition of MR and PET data has great potential both for clinical diagnosis and research. However, contrary to PET/CT scanners, MR images often do not show the complete body of the patient due to field-of-view restrictions imposed either by acquisition time or by magnetic field inhomogeneities. This poses a problem for the correction of attenuation necessary in PET reconstruction. The influence of this source of error and its possible impact on system design must therefore be evaluated. Methods: Twenty pulmonary and mediastinal tumors were simulated with GATE, using both the NCAT phantom and MR patient data. The resulting sinograms were reconstructed with an OSEM OSL algorithm, first with a complete attenuation map and then with a clipped version. In order to determine the feasibility of restoring the attenuation coefficients by means of image-processing, a third reconstruction was performed with a clipped attenuation map for which the missing parts had been replaced by phantom data. Results: With simulated lesions of 1 to 2 cm diameter and activities of 5 kBq/ml and 30 KBq/ml, the reduced field-of-view introduces an error of 14.4% ±3.8%. The recovery of missing coefficients by means of phantom data reduced this error to 1.9% ±0.4%, even for deviations of up to 38% in the positioning of the phantom and 0.04 cm-1 in the attenuation factor. Conclusions: The results show that MR field-of-view restrictions can make reconstructed PET data unacceptable for diagnostic purposes. However, this can be compensated by means of simple preprocessing, the correction being highly tolerant to inaccuracies in patient positioning.     «

Objectives: Simultaneous acquisition of MR and PET data has great potential both for clinical diagnosis and research. However, contrary to PET/CT scanners, MR images often do not show the complete body of the patient due to field-of-view restrictions imposed either by acquisition time or by magnetic field inhomogeneities. This poses a problem for the correction of attenuation necessary in PET reconstruction. The influence of this source of error and its possible impact on system design must ther...     »