Feasibility Study of Deep Learning-Driven Image Restoration for Fast, Low-Count [<sup>18</sup>F]FP-CIT Digital PET/CT.

Research overview

The text below is the source article’s original abstract under a verified reuse license. cc by

Background/Objectives: N-3-[18F]fluoropropyl-2β-carbomethoxy-3β-4-iodophenyl nortropane ([18F]FP-CIT) positron emission tomography/computed tomography (PET/CT) is an effective imaging tool for diagnosing parkinsonism. This study evaluated the feasibility of deep learning (DL)-driven image restoration of short-duration [18F]FP-CIT images. Methods: List-mode data from 202 patients who underwent [18F]FP-CIT PET/CT were reconstructed into 30-s (30sec), 1-min (1min), and reference 10-min (10min) acquisition durations. Patients were divided into training, validation, and test sets in a 6:2:2 ratio. The U-Net model was implemented to generate the DL images from 30sec and 1min data. The visual image quality was assessed using a three-point scale, visual interpretation of striatal dopamine transporter binding patterns, regional standardized uptake value ratios (SUVRs), and quantitative quality metrics including the peak signal-to-noise ratio, root mean squared error, and universal quality index among five series of images: 30sec, 1min, DL-30sec, DL-1min, and 10min. Results: While 30sec and 1min low-count scans showed poor image quality, the DL-driven algorithm showed significant improvement; DL-1min scans achieved excellent ratings in 95% of cases. Concordance with the reference images was 85.0% for visual image quality and 92.5% for visual interpretation in the DL-30sec images, and 97.5% and 92.5%, respectively, in the DL-1min images. Discordant interpretations occurred mainly in patients with atypical parkinsonism. Regional SUVR values and quantitative metrics for the DL-1min images showed good agreement and smaller biases with respect to the reference images, compared with those of DL-30sec images. Conclusions: The DL-driven method generated clinically acceptable [18F]FP-CIT images while reducing acquisition time, with better image quality in the DL-1min than in the DL-30sec images. Concordance in visual interpretation was reduced in the small subgroup of patients with atypical parkinsonism.

Related topics