Harmonizing FDG PET quantification while maintaining optimal lesion detection: prospective multicentre validation in...

Area of Science:

• Nuclear Medicine
• Medical Imaging
• Radiochemistry

Background:

• Advanced positron emission tomography (PET) reconstruction algorithms like point-spread function (PSF) and time-of-flight (TOF) improve lesion detection in oncology.
• However, these methods can introduce variability in quantitative measurements, specifically standardized uptake values (SUVs), across different PET systems.
• This inconsistency hinders direct comparison of quantitative data obtained from various PET scanners.

Purpose of the Study:

• To validate a proprietary software tool, EQ.PET, designed to harmonize SUVs across diverse PET systems.
• To ensure quantitative comparability of PET data regardless of the reconstruction algorithm used, including advanced PSF or PSF+TOF methods.
• To maintain the benefits of improved lesion detection offered by newer reconstruction techniques.

Main Methods:

• NEMA NU2 phantom data were used to determine optimal filters for PSF or PSF+TOF reconstructions on three PET systems, ensuring European Association of Nuclear Medicine (EANM) compliant recovery coefficients.
• PET data from 517 oncology patients were reconstructed using both PSF/PSF+TOF (for optimal lesion detection) and OSEM3D (EANM compliant).
• The EQ.PET software applied a proprietary filter to PSF/PSF+TOF reconstructed data (PSFEQ/PSF+TOFEQ), and SUVs were compared against OSEM3D reconstructions. Confounding factors and adherence to imaging guidelines were assessed.

Main Results:

• Bland-Altman analysis initially showed significant mean SUV ratios between standard PSF/PSF+TOF and OSEM3D reconstructions (SUVmax: 1.46, SUVpeak: 1.23).
• Application of the EQ.PET proprietary filter significantly improved these ratios, bringing them closer to 1 (SUVmax: 1.02, SUVpeak: 1.04), indicating successful harmonization.
• The impact of confounding factors like lesion size, location, radial offset, and patient BMI on the harmonized SUVs was minimal (less than 5%). Adherence to EANM guidelines was good.

Conclusions:

• The EQ.PET software effectively harmonizes SUVs across different PET systems and reconstruction algorithms.
• It is possible to leverage advanced reconstruction techniques like PSF/PSF+TOF for superior lesion detection without compromising quantitative accuracy and cross-system comparability.
• This approach facilitates more reliable and consistent quantitative analysis in oncologic PET imaging.