Abstract:<section class="article-section__content"><h3 class="article-section__sub-title section1"> Purpose</h3>Positron Emission Tomography (PET) imaging with various tracers is increasingly used in Alzheimer's disease (AD) studies. However, access to PET scans using new or less-available tracers with sophisticated synthesis and short half-life isotopes may be very limited. Therefore, it is of great significance and interest in AD research to assess the feasibility of generating synthetic PET images of less-available tracers from the PET image of another common tracer, in particular 18F-FDG. </section><section class="article-section__content"><h3 class="article-section__sub-title section1"> Methods</h3>We implemented advanced deep learning methods using the U-Net model to predict 11C-UCB-J PET images of synaptic vesicle protein 2A (SV2A), a surrogate of synaptic density, from 18F-FDG PET data. Dynamic 18F-FDG and 11C-UCB-J scans were performed in 21 participants with normal cognition (CN) and 33 participants with Alzheimer's disease (AD). Cerebellum was used as the reference region for both tracers. For 11C-UCB-J image prediction, four network models were trained and tested, which included 1) 18F-FDG SUV ratio (SUVR) to 11C-UCB-J SUVR, 2) 18F-FDG Ki ratio to 11C-UCB-J SUVR, 3) 18F-FDG SUVR to 11C-UCB-J distribution volume ratio (DVR), and 4) 18F-FDG Ki ratio to 11C-UCB-J DVR. The normalized Root Mean Square Error (NRMSE), Structure Similarity Index (SSIM), and Pearson's correlation coefficient were calculated for evaluating the overall image prediction accuracy. Mean bias of various ROIs in the brain and correlation plots between predicted images and true images were calculated for ROI-based prediction accuracy. Following a similar training and evaluation strategy, 18F-FDG SUVR to 11C-PiB SUVR network was also trained and tested for 11C-PiB static image prediction. </section><section class="article-section__content"><h3 class="article-section__sub-title section1"> Results</h3>The results showed that all four network models obtained satisfactory 11C-UCB-J static and parametric images. For 11C-UCB-J SUVR prediction, the mean ROI bias was -0.3%±7.4% for the AD group and -0.5%±7.3% for the CN group with 18F-FDG SUVR as the input, -0.7%±8.1% for the AD group, and -1.3%±7.0% for the CN group with 18F-FDG Ki ratio as the input. For 11C-UCB-J DVR prediction, the mean ROI bias was -1.3%±7.5% for the AD group and -2.0%±6.9% for the CN group with 18F-FDG SUVR as the input, -0.7%±9.0% for the AD group, and -1.7%±7.8% for the CN group with 18F-FDG Ki ratio as the input. For 11C-PiB SUVR image prediction, which appears to be a more challenging task, the incorporation of additional diagnostic information into the network is needed to control the bias below 5% for most ROIs. </section><section class="article-section__content"><h3 class="article-section__sub-title section1"> Conclusions</h3>It is feasible to use 3D U-Net based methods to generate synthetic 11C-UCB-J PET images from 18F-FDG images with reasonable prediction accuracy. It is also possible to predict 11C-PiB SUVR images from 18F-FDG images, though the incorporation of additional non-imaging information is needed. </section>

Cyclic 2.5D Perceptual Loss for Cross-Modal 3D Image Synthesis: T1 MRI to Tau-PET

A 3D multi-scale CycleGAN framework for generating synthetic PETs from MRIs for Alzheimer's disease diagnosis

Cross-Modality Generation of Amyloid PET from FDG PET for Alzheimer’s Disease Diagnosis

Synthesizing Missing PET from MRI with Cycle-consistent Generative Adversarial Networks for Alzheimer's Disease Diagnosis

Synthesizing PET Scans Using Attention GAN Based on MRI Scans and Alzheimer's Disease Information.

Ultra-low-dose PET reconstruction using generative adversarial network with feature matching and task-specific perceptual loss

Perception-Enhanced Generative Adversarial Network for Synthesizing Tau Positron Emission Tomography images from Structural Magnetic Resonance Images: a cross-center and cross-tracer study

MRI to PET Cross-Modality Translation using Globally and Locally Aware GAN (GLA-GAN) for Multi-Modal Diagnosis of Alzheimer's Disease

Cross-Modality Image Translation From Brain 18 F-FDG PET/CT Images to Fluid-Attenuated Inversion Recovery Images Using the CypixGAN Framework

Synthesizing Multi-Tracer PET Images for Alzheimer's Disease Patients using a 3D Unified Anatomy-aware Cyclic Adversarial Network

A novel locally-constrained GAN-based ensemble to synthesize arterial spin labeling images

GANDALF: Generative Adversarial Networks with Discriminator-Adaptive Loss Fine-tuning for Alzheimer's Disease Diagnosis from MRI

BPGAN: Brain PET synthesis from MRI using generative adversarial network for multi-modal Alzheimer's disease diagnosis

Synthesizing beta-amyloid PET images from T1-weighted Structural MRI: A Preliminary Study

Cross-Modal PET Synthesis Method Based on Improved Edge-Aware Generative Adversarial Network

Cross-modal Medical Image Generation Based on Pyramid Convolutional Attention Network

Generation of synthetic PET images of synaptic density and amyloid from 18 F‐FDG images using deep learning

Generation of Synthetic PET Images of Synaptic Density and Amyloid from 18F-FDG Images Using Deep Learning

Visual and Quantitative Evaluation of Amyloid Brain PET Image Synthesis with Generative Adversarial Network

An Unsupervised Region of Interest Extraction Model for Tau PET Images and Its Application in the Diagnosis of Alzheimer's Disease.

GANBERT: Generative Adversarial Networks with Bidirectional Encoder Representations from Transformers for MRI to PET synthesis