Formal clinical reading study for synthetic PET diagnostic utility

Quantitatively evaluate, through a formal clinical reading study, the diagnostic utility of synthetic PET images generated from MRI using the PASTA pathology-aware conditional diffusion framework.

Background

PASTA is a conditional diffusion model framework designed to translate brain MRI to PET while preserving both structural and pathological information. The study demonstrates strong quantitative performance and pathology awareness, and shows improvements for Alzheimer’s diagnosis compared to MRI.

Despite these findings, the work acknowledges the lack of a formal clinical reading study to rigorously and quantitatively assess the diagnostic utility of the synthesized PET images in a clinical setting. Addressing this gap is necessary to establish clinical validity and reliability.

References

A limitation of this study is the absence of a formal clinical reading study to quantitatively evaluate the diagnostic utility of the synthetic PET, which needs to be addressed in future work.

— Translating MRI to PET through Conditional Diffusion Models with Enhanced Pathology Awareness  (2603.18896 - Li et al., 19 Mar 2026) in Section 6: Conclusion