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

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

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

Abdus Sattar PhD, Abstract Co-Author: Research collaboration, Koninklijke Philips NV

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

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

Evaluation for potentially increased diagnostic confidence in distinguishing brown fat from lesions using mDixon during PET/MRI in pediatric oncologic patients.

Twelve patients (mean age 13.8±3.37; 6 girls, 6 boys) underwent a PET/CT and PET/MRI for evaluation of their oncologic disease following a single dose of 18F-FDG. MR imaging included whole-body 3D T1-weighted (wb3DT1) for attenuation correction and non-enhanced whole body 3D modified Dixon (mDixon) for anatomical reference. Automated three-segment model of MR-based Attenuation Correction (MRAC) was used for PET reconstruction. Two observers (senior and junior level) evaluated data for tracer-avid foci in PETCT and PETMR in a blinded randomized fashion and determined their etiology as either related or unrelated to activated brown fat using a diagnostic confidence scale. Maximum standardized uptake values (SUVmax) of tracer-avid lesions were measured. Statistical analysis included descriptive statistics, sign test and Kappa analysis.  

Twenty four studies were reviewed and 30 FDG-avid lesions were seen. Readers’ agreement was for nodes (62.5%), neoplasia (91.6%), activated brown fat (100%) and normal (66.67%) [Kappa (k) = 0.14, P=0.23; k=0.62, P <0.0005; k=1 and k=0.33, P=0.042], with high to very high diagnostic confidence. Moderate agreement for lesion detection was seen (k= 0.42). 18 (60%) lesions were identified by both readers, 12 lesions were seen by a reader but not the other. No lesion was seen by both readers in 7 patients. Perfect agreement for activated brown fat visualization and grading as “no activated brown fat” (54.17%), “moderate” (54%) or “significant” (8.3%) was seen. Substantial agreement and very high diagnostic confidence for absence of lesions using activated brown fat and WB 3DT1 in 87.5% (k=0.72, P<0.005). Although diagnostic confidence using mDixon was altered for either reader (P<0.05), the junior reader considered mDixon helpful (100%) for tissue characterization. Both readers agreed that mDixon water and fat images (k=0.75 and 0.62; P <0.005) were diagnostically useful.

mDixon yields equivalent diagnostic confidence for defining activated brown fat versus lesions as compared to PET/CT.

Using the mDixon sequence in PET/MRI allows for distinction of brown fat from pathologic lesions in pediatric patients by virtue of its novel, fat delineating images.

Martinez Rios Arellano, C, Sher, A, Faulhaber, P, Sattar, A, Herrmann, K, Bangert, B, Value of mDixon to Distinguish Brown Fat in a Pediatric Oncologic Patient Population.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14018827.html