Lutetium Background

Utilizing the Lutetium Background Radiation for Quantitative PET Data Corrections

Most state of the art clinical PET scanners use lutetium-based scintillation crystals (LSO/LYSO) for photon detection. A naturally occurring isotope, ¹⁷⁶Lu, present in these crystals emits low-level background radiation continuously. Traditionally, this background radiation has been treated as noise. We view it as a signal that can be used for a variety of data corrections.

Why This Matters

In long axial-field-of-view (LAFOV) PET systems—such as the uEXPLORER and the NeuroEXPLORER—the large volume of detector material creates a substantially increased flux of lutetium background radiation. The high sensitivity of these systems enables us to harness this intrinsic radiation as a built-in transmission source for attenuation correction (AC), scatter correction (SC), and motion correction (MC), which are among major factors that limit the quantitative accuracy in PET. 

Attenuation and Scatter Correction

Comparison of attenuation maps obtained with lutetium transmission data enhanced with deep learning to CT-based attenuation maps, used for PET attenuation and scatter correction.

AC and SC are typically dependent on additional CT scans which introduce additional radiation dose and are susceptible to attenuation-emission mismatches. As long-AFOV PET scanners enable ultralow-dose and ultrafast imaging protocols, the relative contribution of CT dose becomes dominant. Reducing or replacing CT-based corrections becomes increasingly important—especially in pediatric imaging, longitudinal studies, and healthy volunteer research. 

The 307 keV photons emitted during ¹⁷⁶Lu decay can be detected across the field of view and used as a transmission source to reconstruct attenuation maps (μ-maps) [Rothfuss et al. Phys Med Biol. (2014)]. Our studies demonstrate that ¹⁷⁶Lu transmission data in long-AFOV PET systems can provide significantly improved μ-maps compared to prior conventional systems and can be combined with deep learning based methods to provide CT-quality μ-maps. We have developed a deep learning framework trained on simulated digital phantoms generated from clinical total-body PET/CT datasets acquired on the uEXPLORER scanner. Using this approach, we can generate CT-quality attenuation maps from 10-minute lutetium transmission scans that can be used for PET attenuation and scatter correction. These developments move lutetium-based attenuation correction closer to clinical feasibility.

Motion Correction

Lutetium background based motion tracking methods validated against external motion tracking data.

Patient motion remains a major source of quantification error in PET imaging. Data-driven motion correction methods based on emission data can fail in early dynamic frames, low-count regions, or ultralow-dose scans. We have developed data-driven motion tracking approaches using the lutetium transmission events as a tracer-independent signal. Because the lutetium background is always present, is generally independent of tracer distribution, and is acquired simultaneously with PET, it provides a stable reference for motion estimation, particularly in challenging low-count scenarios. This enables generation of motion-compensated μ-maps, and improved attenuation-emission alignment. Proof-of-concept validation data has been acquired on the NeuroEXPLORER scanner against United Imaging's external motion tracking device (UMT).

Broader Impact

This project represents one of the first comprehensive efforts to study the quantitative applications of lutetium background radiation in total-body PET. Our long-term goal is to establish lutetium-based correction methods as an integrated component of next-generation PET imaging.

Logo of the National Institute of Biomedical Imaging and Bioengineering, featuring "NIH" and arrow.

Funding:

R03 EB032457, PI: N Omidvari