AI Insight
Researchers developed and validated a deep learning model based on a conditional generative adversarial network (cGAN) that generates delayed-phase dopamine transporter (DAT) images from early-phase [18F]-FP-CIT PET scans, trained on 477 dual-phase acquisitions. The generated images showed strong quantitative agreement with real delayed-phase images, with striatal binding ratio correlations of R=0.93 and R=0.90 for internal and independent validation sets respectively. Diagnostic performance for detecting degenerative parkinsonism was comparable between generated and real images across both validation sets, though real images showed significantly higher accuracy for general abnormality detection in the independent cohort.
Why it matters
This approach could reduce total PET scan duration and radiation exposure by eliminating the need for a separate delayed acquisition, potentially improving patient comfort and increasing clinical throughput in the evaluation of parkinsonian disorders.
Understand the Science
by Changhwan Sung, Jungsu S. Oh, Sun Young Chae, Dong Yun Lee, Minyoung Oh, Sang Ju Lee, Seung Jun Oh, Sungyang Jo, Sun Ju Chung, Jae Seung Kim
Objectives
To develop and validate a deep learning-based model capable of generating dopamine transporter (DAT) images from early-phase [18F]-FP-CIT positron emission tomography (PET) imaging.
Materials and methods
Conditional generative adversarial network was trained using 477 dual-phase [¹⁸F]-FP-CIT PET scans acquired with a conventional PET system. The model generated delayed-phase images from early-phase dynamic scans (30–40 min post-injection), using five adjacent axial slices to predict the central delayed-phase slice. The model was evaluated using an internal validation set from the same scanner and an independent prospective validation set from a digital PET system. Striatal binding ratios (SNBR) and inter-subregional ratios were compared using Pearson’s correlation and receiver operating characteristic-area under the curve (AUC) analyses.
Results
Generated images showed high similarity to real delayed images. SNBRs for the whole striatum correlated strongly between real and generated images (R = 0.93 and 0.90 for internal and independent sets, respectively). Diagnostic performance was comparable with the highest AUC observed in the posterior putamen (real vs generated: 1.00 vs 0.98, p > 0.2). Visual assessments revealed no uninterpretable images, and interpretability did not differ significantly. Diagnostic accuracy of generated images was comparable to that of real images in the internal validation set for detection of abnormality (p = 0.453) and degenerative parkinsonism (DP) (p = 1.000). In the independent validation set, DP detection remained comparable (p = 0.25), whereas real images demonstrated significantly higher accuracy for abnormality detection (p < 0.001).
Conclusion
Deep learning-based model generated DAT images achieve satisfactory performance in quantitative and visual assessments across internal and independent validation sets.