Claudia M. Martinez Rios Arellano MD, Presenter: Research Grant, Koninklijke Philips NV

Andrew Sher MD, Abstract Co-Author: Research Grant, Koninklijke Philips NV

Li Fan, Abstract Co-Author: Nothing to Disclose

Karin Anna Herrmann MD, Abstract Co-Author: Consultant, Koninklijke Philips NV

Lingzhi Hu PhD, Abstract Co-Author: Employee, Koninklijke Philips NV

Peter F. Faulhaber MD, Abstract Co-Author: Speaker, Koninklijke Philips NV Grant, Koninklijke Philips NV Medical Advisor, MIM Software Inc

Barbara Ann Bangert MD, Abstract Co-Author: Investigator, Koninklijke Philips NV

To demonstrate the clinical feasibility and quantitative accuracy of Positron Emission Tomography/Magnetic Resonance Imaging (PET/MRI) in a pediatric patient population using a dedicated pediatric PET/MRI acquisition and reconstruction protocol.

Twenty patients (12.4±5.1 yr; 12 girls,8 boys) were evaluated. Eight patients underwent a whole-body (WB) 3D T1-weighted spoiled gradient echo sequence (3DT1w) at two fields of view (FOV) 600 and 400mm to evaluate MR-based Attenuation Correction (MRAC) segmentation algorithm. Lung boundaries from automatic versus gold standard manual segmentation were compared. Twelve oncologic patients underwent PET/CT followed by PET/MRI after a single injection of 18F-FDG. PET time was 65±5 and 96±6 minutes respectively after tracer injection. 3-segment MRAC was achieved utilizing two automatic segmentation algorithms, one developed for adults and one for pediatric patients. Non-enhanced WB 3D modified Dixon was performed for anatomical reference. Image quality assessment and regions of interest for quantitative analysis of standardized uptake values (SUVmax/mean) were performed on PET/CT and PET/MR. Statistical analysis included DICE coefficient, Pearson’s correlation, and t-test.

The correlation factor of lung volumes across groups was r>0.9 (p<0.001) with a similarity coefficient of 90±2% and 88±4% between automatic and manual segmentation. Accuracy of MR based transmission map was appreciable with the pediatric MRAC compared to the adult method (0/12 vs 5/12 suboptimal, respectively). Correlation coefficients between SUV(max)/SUV(mean) of PET/MRI and PET/CT are 0.28/0.36, 0.52/0.58 and 0.39/0.52 for liver, spleen and lungs respectively with adult MRAC method. Pediatric MRAC lung segmentation increased correlation factors to 0.58/0.61, 0.77/0.62 and 0.44/0.61. Lung SUV(max) and SUV(mean) (0.48±0.15 and 0.33±0.11) on PET/MRI reconstructed with the pediatric protocol are significantly lower than PET/CT (0.59±0.22 and 0.45±0.16), p<.0017.

Improved quantitative accuracy of MRAC is seen with a dedicated pediatric PET/MRI reconstruction method, yielding superior image quality and increased SUVmax/mean correlation values in comparison to an adult reconstruction method.

MRAC in children is uniquely challenging due to lung size variation and MR motion artifact. A dedicated pediatric PET MRAC allows superior quantitative accuracy for pediatric PET/MRI.

Martinez Rios Arellano, C, Sher, A, Fan, L, Herrmann, K, Hu, L, Faulhaber, P, Bangert, B, Improving Quantitative Accuracy of PET/MRI in a Pediatric Patient Population Using a Dedicated Pediatric PET/MRI Reconstruction Paradigm.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14016857.html