Abstract
1477
Objectives Respiratory motion blurs lung lesions and deteriorates precision on activity concentration and size estimates. Simultaneous gated PET-MRI can compensate for this effect as MRI provides additional location information for PET quantitation. We determine the theoretical improvement that can be achieved with PET-MRI compared to gated or ungated PET.
Methods We use the hybrid Cramer-Rao lower bound (HCRLB) to allow for inclusion of prior information and to determine estimation performance for lung lesions of two sizes (r=0.3, 1cm). Realistic NCAT Monte-Carlo simulations (SimSET) are used to generate PET data with respiratory motion. Our performance metric is the estimation SNR, defined as the ratio of the parameter value to the sqrt of HCRLB. Five imaging conditions are considered: 1. Ideal PET– gold standard data with no location uncertainty; 2. Ungated PET; 3-4. Gated PET– data binned into 8 gates using a belt system with no prior lesion location information. Performance is calculated both from 1 (end-inspiration) or 8 gates; 5. PET/MRI– information on lesion location is provided from MRI with 4 levels of precision, at each of the 8 gates.
Results Size estimation SNR were 7.87 (ideal), 3.17 (ungated), 2.75 (1 gate), 6.82 (8 gates), 7.62 (PET/MRI- 1mm precision), yielding an 8% (12%) SNR improvement in activity (size) estimation performance for a 0.3cm lesion (equivalent to 17% (25%) improvement in sensitivity). Improvement is greater for size than activity estimation (~50%).
Conclusions PET/MRI yields promising improvement in quantitation of lesion activity and size by providing information about lesion location at no additional cost in terms of scanning time or radiation dose.
Research Support RO1-EB5876
- © 2009 by Society of Nuclear Medicine