Abstract
1979
Objectives 18F-FDG PET-CT has demonstrated that brown adipose tissue (BAT) activity can be stimulated in humans. This has enabled the study of the role of this tissue in diseases such as obesity. PET quantification of BAT activity in all these regions together, to arrive to the total BAT activity, is cumbersome and operator dependent. We studied different processing techniques to develop an automatic method by which activity in BAT regions can be objectively and operator-independently determined.
Methods Out of our prospectively collected cold-induced BAT database we selected patients with a different body mass index (BMI) and degree of BAT activity. PET-active BAT regions of interest (ROI’s) were selected by a 3D cubic ROI template model, as well as tissue anatomy characteristics in Hounsfield Units on low-dose-CT (PMOD 3.0 View and Fusion). Different thresholds were studies for region deliniation by PET (SUVmax 0.8 till 1.5), CT (HU -200 till 0), and the combination of PET with that of CT thresholds. CT-characteristics were based on earlier rodent studies and a pilot analysis of our own database.
Results With lower PET thresholds up to 0.8, too much aspecific activity such as in muscle was collected. With higher PET thresholds up to 1.5, too much BAT regions with low activity were discarded. With high CT thresholds up to -200, too much low density tissue, especially lung with paravertebral regions, was included. With too low CT thresholds up to 0, too much high density tissue, especially muscle, was included. An optimal combination was found for a pre-selection of regions on the basis of SUV>1.0, and corrected for fat tissue by HU on low-dose-CT between -30 and -185. However, expert visual inspection of results and recognition of possible artefacts remained important.
Conclusions Our study demonstrates the feasibility of performing a semi-automated analysis of total BAT activity in humans to standardize the PET-CT technique in the determination of human BAT physiology