Differences among [¹⁸F]FDG PET-derived parameters in lung cancer produced by three software packages

Area of Science:

• Nuclear Medicine
• Oncology
• Medical Imaging Analysis

Background:

• ¹⁸F-fluorodeoxyglucose Positron Emission Tomography (FDG PET) is crucial for lung cancer staging and treatment response assessment.
• Accurate quantification of metabolic and volumetric parameters from FDG PET images is essential for clinical decision-making.
• Variability in image analysis software can potentially introduce discrepancies in these quantitative metrics.

Purpose of the Study:

• To investigate and compare derived metabolic and volumetric parameters from [¹⁸F]FDG PET scans across three different software programs in lung cancer patients.
• To evaluate the impact of software choice on quantitative metrics such as Standardized Uptake Value (SUV), Total Lesion Glycolysis (TLG), and Metabolic Tumor Volume (MTV).
• To identify potential discrepancies in parameter estimation that could influence diagnostic and therapeutic procedures.

Main Methods:

• Retrospective analysis of baseline [¹⁸F]FDG PET/CT studies from 98 lung cancer patients.
• NEMA phantom studies were conducted using Philips EBW, MIM Software, and Rover to establish optimal tumor delineation methods based on tumor size.
• Semiquantitative parameters (SUV_max, SUV_mean, TLG, MTV) were calculated using body weight (BW), lean body mass (LBM), and Bq/mL normalization and compared across the software platforms.

Main Results:

• Statistically significant differences were observed in SUV_mean (LBM) between MIM Software and Rover (p < 0.005).
• Significant differences were found in SUV_mean (Bq/mL) between Rover and Philips EBW (p < 0.005), and Rover and MIM Software (p < 0.005).
• Metabolic Tumor Volume (MTV) showed a statistically significant difference between MIM Software and Philips EBW (p = 0.0489).

Conclusions:

• The choice of image analysis software significantly impacts the estimation of semiquantitative [¹⁸F]FDG PET metabolic and volumetric parameters in lung cancer.
• Observed discrepancies in SUV_mean and MTV highlight the need for standardization or careful consideration of software used in multicenter studies.
• These findings underscore the importance of acknowledging software-specific variations for consistent diagnostic interpretation and treatment planning in lung cancer management.