Accurate quantification of brown adipose tissue through PET-guided CT image segmentation

Objectives To develop a faster, more accurate method of finding brown fat tissue (BAT) boundaries in CT images using a computer-aided quantification system.

Methods With IRB approval, we analyzed 51 PET-CTs of cancer patients, consisting of 31 women (35.6±14.5 years) and 20 men (26.3±10.3 years). We present a novel method for the detection and delineation of significant uptake regions on PET, and corresponding anatomy on CT. For BAT delineation, we adopted a random walk graph-based technique. For total fat tissue segmentation, an HU based thresholding technique was integrated with the Fuzzy Connectedness image segmentation algorithm (Fig 1b). The proposed tool outputs the standardized uptake value (SUV) of BAT regions from PET images, BAT metabolic and anatomical volumes, and total fat volume from the CT images as well as all volume ratios. Several textural parameters were also extracted from segmented regions in both PET and CT images for texture based correlations if desired by physicians for quantification purposes. Uptake regions from PET were located and propagated into the corresponding CT image as seeds to guide the delineation process (Fig 1a).

Results We compared true positive (TP) and false positive (FP) volume fractions of segmented tissues with the ground truth, provided by two experts blinded to their evaluations. Sensitivity and specificity of the proposed method were reported as TP: 92.3±10.1%, 1-FP: 82.2±18.9%. Volumes derived by the proposed segmentation were correlated with expert derived volumes, and the resulting correlation coefficient, after linear regression, was found to be R2=0.97, p<0.01. The time to segment BAT was ~2 seconds per slice, and at most 2 minutes per study, whereas manual identification took on average 40 minutes per study.

Conclusions PET-guided BAT segmentation in CT images is a novel technique that adds robustness and accuracy to the quantification of BAT and provides a comparative volume percentage, with respect to total fat tissue.