A data-driven approach for spatial alignment of PET add-on detectors with sub-millimeter accuracy.

The integration of high-resolution add-on detectors with clinical PET scanners has the potential to significantly enhance imaging resolution and sensitivity. However, precise geometric alignment of these auxiliary detectors is critical to fully realizing their benefits. This study presents a novel data-driven methodology for the precise geometric alignment of high-resolution auxiliary detectors with clinical PET scanners, relying solely on point source measurements. Method: The proposed approach calculates transformation parameters that minimize angular discrepancies in lines of response (LORs) by assuming collinearity between the point source, the add-on detector crystal, and the scanner crystal. An effective depth of interaction (DOI) model was incorporated to improve interaction point estimation. The method was first validated through Monte Carlo (MI) simulations, comparing the predefined add-on detector geometry with the estimated geometry. Experimental validation was performed by comparing images reconstructed using the estimated geometry with those from the native scanner, using 54 point source positions arranged in a structured grid and multiple line source configurations, including parallel and arbitrary angle orientations. Centroid deviations and angular differences were analyzed to assess alignment accuracy. Result: The discrepancy between the ground truth and estimated geometry was measured at 0.23 ± 0.10 mm and 0.20 ± 0.07 mm for the two add-on detectors, respectively. In experimental validation, the centroid deviation of reconstructed point source images using the estimated geometry remained consistently below 1 mm across the field of view deviation was consistently less than 1 degree, and the centroid deviation remained below 1 mm across all slices. Conclusion: The results demonstrate that our alignment method achieves sub-millimeter accuracy. The accuracy and simplicity of this approach, combined with its non-reliance on physical phantoms, hold significant promise for improving the quality and diagnostic accuracy of the PET systems with auxiliary devices. .