Clinical issues regarding misclassification by Dixon based PET/MR attenuation correction

• Published in: Hellenic journal of nuclear medicine Vol. 18; no. 1; pp. 42 - 47
• Main Authors: Kong, Eunjung, Cho, Ihnho
• Format: Journal Article
• Language: English
• Published: Greece 01.01.2015
• Subjects: Aged; Algorithms; Body Mass Index; Decision Support Systems, Clinical; Female; Fluorodeoxyglucose F18; Humans; Image Processing, Computer-Assisted; Imaging, Three-Dimensional; Magnetic Resonance Imaging - methods; Male; Middle Aged; Multimodal Imaging - methods; Muscle, Skeletal - drug effects; Neoplasms - diagnosis; Neoplasms - diagnostic imaging; Positron-Emission Tomography - methods; Reference Values; Tomography, X-Ray Computed - methods
• ISSN: 1790-5427

The Dixon sequence is acquired for attenuation correction (AC) of positron emission tomography (PET) data in integrated PET/magnetic resonance (MR). However it sometimes misclassifies soft tissue and fat in μ-map. In the present study, we investigated factors related to this misclassification and their clinical impacts. Forty-eight oncological patients (19 males and 29 females, mean age: 59±11 years old) underwent a single fluorine-18 fluorodeoxyglucose ((18)F-FDG) injection/dualimaging protocol on PET/computed tomography (CT) and subsequently PET/MR. Patients were assigned to either of two groups; group A with a misclassification in at least one bed position or group B with a correctly classified μ-map. We compared body mass index (BMI), lean body mass, fasting duration, volume of hydration and age between group A and group B. In addition, we analyzed the impact of PET quantification using standard uptake ratio (SUR) defined as uptake in volume of interest/uptake in thigh muscle. The Dixon-AC SUR was compared with CT-AC SUR in misclassified bed positions and correctly classified bed positions. All patients were scanned in four bed positions by PET/MR. Ten patients were assigned to group A; six showed misclassification in a bed position (5 in head and 1 in abdomen), three patients in 3 bed positions (head-thorax-abdomen), and one patient in partial bed position in neck. Misclassification was observed in 21% of 48 patients. Group A and group B showed no statistically significant differences in BMI, lean body mass, fasting duration or age, however the volume of hydration in group A (245mL) was smaller than in group B (452.6mL) (P=0.027). In group A, we analyzed Dixon-AC SUR/CT-AC SUR ratios in 16 misclassified and 24 correctly classified regions, and ratios in these regions were significant different 0.80 and 0.93, respectively (P=0.046). Because no corrective method has been devised after a scan, we recommend that Dixon images with μ-maps should be checked before interpreting PET/MR images and emphasize the importance of hydration, pre-examination. Misclassification errors do not change the presence of (18)F-FDG uptake but can have significant impacts on PET quantification in affected bed positions.