An Adaptive Thresholding Method for BTV Estimation Incorporating PET Reconstruction Parameters: A Multicenter Study...

Area of Science:

• Nuclear Medicine
• Medical Imaging
• Radiotherapy Planning

Background:

• Accurate biological target volume (BTV) estimation is crucial for effective radiotherapy.
• Adaptive thresholding algorithms offer potential for improved BTV delineation in PET/CT imaging.
• Reconstruction parameters significantly influence image quality and subsequent volume calculations.

Purpose of the Study:

• To evaluate the robustness and reliability of an adaptive thresholding algorithm for BTV estimation.
• To assess the impact of image reconstruction parameters on algorithm performance.
• To determine if the algorithm can be applied universally across different PET/CT scanner models.

Main Methods:

• A multicenter study utilized a phantom with varying sphere sizes and (18)F-FDG concentrations (target-to-background ratios).
• Scans were performed with different acquisition times, and image reconstruction parameters (iterations, smoothing) were varied.
• Optimal thresholds (TS) were determined, and multiple regression analysis identified predictors of TS.

Main Results:

• The predictive model demonstrated a high goodness of fit (R²: 0.74-0.92).
• Target-to-background ratio was the primary predictor of TS; FWHM (Full Width at Half Maximum) was also significant for most scanners.
• Reliable model estimation was confirmed through small cross-validation shrinkage, indicating excellent reproducibility.

Conclusions:

• Integrating post-reconstruction filtering FWHM into adaptive thresholding provides a robust BTV estimation method.
• This approach ensures reliable BTV measurements across various PET/CT scanners.
• The validated algorithm eliminates the need for scanner-specific calibration, simplifying clinical implementation.