Liver Tumor Volume Determination by PET Using Ventilator Suspended Respiration
G Li1*, I Burger2, C Schmidtlein3, C Ridge4, S Solomon5, J Humm6, (1) Memorial Sloan-Kettering Cancer Center, New York, NY, (2) University Hospital Zurich, Yyy, ,(3) Mem. Sloan-Kettering Cancer Ctr, New York, NY, (4) Memorial Sloan-Kettering Cancer Center, New York, New York, (5) Memorial Sloan-Kettering Cancer Center, New York, NY, (6) Memorial Sloan-Kettering Cancer Center, New York, NYSU-D-500-2 Sunday 2:05PM - 3:00PM Room: 500 Ballroom
Purpose: To determine the effect of respiratory motion on liver tumor volume by FDG PET.
Methods: PET/CT was performed on patients during free breathing (FB) and then breath hold (BH) by suspending respiration on a ventilator, allowing 2-3 minutes of near-motionless PET data to be acquired. A reference region on normal liver was selected as a control for background FDG uptake quantification. A dataset of 11 patients with 13 liver lesions were studied under both BH and FB conditions before tumor ablation. A threshold-based segmentation method (GE Advantage) and a background subtraction classification method (in-house) were used to delineate FDG tumor volumes under FB and BH conditions. For tumors with an SUVmax>2xSUV(liver-bkg), a threshold of 50% of the maximum uptake was used for segmentation via the GE Advantage PET VCAR software module. The background-subtracted lesion (BSL) method was implemented by a Gaussian fit of the background from the SUV-volume histogram (SVH). A cutoff threshold was defined as twice sigma of the Gaussian fit plus the mean. Subtraction of this Gaussian fit from the SVH beyond this cutoff determines the signal above background. Custom software was developed for BSL classification. Pixels with PET intensity above the fitted background were calculated as the tumor volume.
Results: In BH PET images, there was an increase in SUVmax [16%(σ=11%)] and SUVmean [20%(σ=9%)], but a reduction in delineated volume [57%(σ=55%)], in comparison with FB PET images. The background uptake in normal liver showed no bias, with an average of -1%(σ=10%). The BSL values are similar between BH and FB for high-intensity lesions (SUVmax>9 g/ml).
Conclusion: This is the first time that a motion-free PET image of the liver was acquired in patients and compared with free-breathing results. Motion blurring caused an apparent volume increase of 50% on average with 16%-20% reductions in SUVmax and SUVmean.