Abstract:Purpose 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 F-18-FDG. Methods We implemented advanced deep learning methods using the U-Net model to predict C-11-UCB-J PET images of synaptic vesicle protein 2A (SV2A), a surrogate of synaptic density, from F-18-FDG PET data. Dynamic F-18-FDG and C-11-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 C-11-UCB-J image prediction, four network models were trained and tested, which included 1) F-18-FDG SUV ratio (SUVR) to C-11-UCB-J SUVR, 2) F-18-FDG K-i ratio to C-11-UCB-J SUVR, 3) F-18-FDG SUVR to C-11-UCB-J distribution volume ratio (DVR), and 4) F-18-FDG K-i ratio to C-11-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, F-18-FDG SUVR to C-11-PiB SUVR network was also trained and tested for C-11-PiB static image prediction. Results The results showed that all four network models obtained satisfactory C-11-UCB-J static and parametric images. For C-11-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 F-18-FDG SUVR as the input, -0.7% +/- 8.1% for the AD group, and -1.3% +/- 7.0% for the CN group with F-18-FDG K-i ratio as the input. For C-11-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 F-18-FDG SUVR as the input, -0.7% +/- 9.0% for the AD group, and -1.7% +/- 7.8% for the CN group with F-18-FDG K-i ratio as the input. For C-11-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. Conclusions It is feasible to use 3D U-Net-based methods to generate synthetic C-11-UCB-J PET images from F-18-FDG images with reasonable prediction accuracy. It is also possible to predict C-11-PiB SUVR images from F-18-FDG images, though the incorporation of additional non-imaging information is needed.

Automated Neural Architecture Search for Cardiac Amyloidosis Classification from [18F]-Florbetaben PET Images

Convolutional Neural Networks for Fully Automated Diagnosis of Cardiac Amyloidosis by Cardiac Magnetic Resonance Imaging

MedNAS: Multi-Scale Training-Free Neural Architecture Search for Medical Image Analysis

18F-Florbetaben PET/CT for the Diagnosis and Subtyping of Cardiac Amyloidosis: A Case Series and Review of the Literature

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

Distinguishing Benign and Malignant Lesions on Contrast-Enhanced Breast Cone-Beam CT with Deep Learning Neural Architecture Search

AmyloidPETNet: Classification of Amyloid Positivity in Brain PET Imaging Using End-to-End Deep Learning

Improved amyloid burden quantification with nonspecific estimates using deep learning

Automatic Attention Learning Using Neural Architecture Search for Detection of Cardiac Abnormality in 12-Lead ECG

Deep learning to diagnose cardiac amyloidosis from cardiovascular magnetic resonance

Deep learning approach for automated segmentation of myocardium using bone scintigraphy single-photon emission computed tomography/computed tomography in patients with suspected cardiac amyloidosis

A robust radiomic-based machine learning approach to detect cardiac amyloidosis using cardiac computed tomography

EAMNet: an Alzheimer's disease prediction model based on representation learning

Detection and risk stratification of cardiac amyloidosis patients by integration of imaging and non-imaging data using a machine learning approach

Deep learning-based binary classification of beta-amyloid plaques using 18 F florapronol PET

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

Automated deep learning segmentation of cardiac inflammatory FDG PET

Classification of Alzheimer’s Disease by Combination of Convolutional and Recurrent Neural Networks Using FDG-PET Images

GC-CNNnet: Diagnosis of Alzheimer's Disease with PET Images Using Genetic and Convolutional Neural Network

MRI-based Deep Learning Assessment of Amyloid, Tau, and Neurodegeneration Biomarker Status across the Alzheimer Disease Spectrum

Automated lesion detection of breast cancer in [18F] FDG PET/CT using a novel AI-Based workflow