Click here to

Session: Advanced Imaging Applications for Radiotheranostics [Return to Session]

Comparison of Physiologic FDG PET Uptake in CT-Based Vs PET-Corrected Organ Contours in Non-Small Cell Lung Cancer Patients

O Lokre*, A Weisman, T Perk, AIQ Solutions, Madison, WI


SU-F-202-6 (Sunday, 7/10/2022) 2:00 PM - 3:00 PM [Eastern Time (GMT-4)]

Room 202

Purpose: Quantifying functional activity of organs on PET/CT images may provide clinically meaningful information for patient management but is complex to automate due to spillover artifacts and patient motion. Here, we assess the differences in quantification between organ contours generated on the CT alone compared to those corrected for spillover on PET.

Methods: FDG PET/CT images of 94 patients with non-small cell lung cancer were gathered. Ten structures (liver, spleen, thyroid, kidneys, pancreas, bladder, aorta, bowel, stomach and heart) were manually contoured on the CT images, after which the structure contours were manually corrected for spillover using the PET images. The maximum and average SUV (SUVmax, SUVmean) and organ volume were extracted from the images using both sets of contours. CT-based and PET-corrected contours were compared using Pearson’s correlation coefficient (R) and Bland-Altman’s analysis of 95% limits of agreement (LOA) on log-transformed data.

Results: Volume differences of CT based contours and PET-corrected contours ranged from a mean of 0.19% (thyroid) to 37.04% (bladder). SUVmax showed minimal changes (LOAs within +/-10%). in heart, bladder, pancreas, stomach and thyroid. The largest impact was found in liver (R=0.85, LOA=[-29%, 49%]) and bowel (R=0.56, LOA=[-57.9%, 569.4%]). Other structures showed LOAs within +/-40% with R>/=0.95. SUVmean showed minimal changes (R>/=0.98, LOAs within +/-11%) in all structures except the bladder (R=0.92, LOA=[-22.2%, 101.0%]) and bowel (R=0.97, LOA=[-7.3%, 16.3%]).

Conclusion: Most organs exhibited minimal to moderate change in quantification, with SUVmean being the least sensitive and SUVmax the most sensitive. Thus, for CT based contours, SUVmean may be a more reliable measure compared to SUVmax.

Funding Support, Disclosures, and Conflict of Interest: All authors on this abstract are employed by AIQ Solutions


Quantitative Imaging, PET, CT


IM/TH- Image Analysis (Single Modality or Multi-Modality): Quantitative imaging

Contact Email