Evaluation of a 2D UNet-Based Attenuation Correction Methodology for PET/MR Brain Studies

Abstract

Deep learning (DL) strategies applied to magnetic resonance (MR) images in positron emission tomography (PET)/MR can provide synthetic attenuation correction (AC) maps, and consequently PET images, more accurate than segmentation or atlas-registration strategies. As first objective, we aim to investigate the best MR image to be used and the best point of the AC pipeline to insert the synthetic map in. Sixteen patients underwent a 18F-fluorodeoxyglucose (FDG) PET/computed tomography (CT) and a PET/MR brain study in the same day. PET/CT images were reconstructed with attenuation maps obtained: (1) from CT (reference), (2) from MR with an atlas-based and a segmentation-based method and (3) with a 2D UNet trained on MR image/attenuation map pairs. As for MR, T1-weighted and Zero Time Echo (ZTE) images were considered; as for attenuation maps, CTs and 511 keV low-resolution attenuation maps were assessed. As second objective, we assessed the ability of DL strategies to provide proper AC maps in presence of cranial anatomy alterations due to surgery. Three 11C-methionine (METH) PET/MR studies were considered. PET images were reconstructed with attenuation maps obtained: (1) from diagnostic coregistered CT (reference), (2) from MR with an atlas-based and a segmentation-based method and (3) with 2D UNets trained on the sixteen FDG anatomically normal patients. Only UNets taking ZTE images in input were considered. FDG and METH PET images were quantitatively evaluated. As for anatomically normal FDG patients, UNet AC models generally provide an uptake estimate with lower bias than atlas-based or segmentation-based methods. The intersubject average bias on images corrected with UNet AC maps is always smaller than 1.5%, except for AC maps generated on too coarse grids. The intersubject bias variability is the lowest (always lower than 2%) for UNet AC maps coming from ZTE images, larger for other methods. UNet models working on MR ZTE images and generating synthetic CT or 511 keV low-resolution attenuation maps therefore provide the best results in terms of both accuracy and variability. As for METH anatomically altered patients, DL properly reconstructs anatomical alterations. Quantitative results on PET images confirm those found on anatomically normal FDG patients.

Direct attenuation correction for 99mTc-3PRGD2 chest SPECT lung cancer images using deep learning

“…For the choice of neural network, the Unet framework in 3D rather than 2D was used in this study. This is mainly due to the fact that the chest SPECT image is a 3D structure after reconstruction, and the 2D approach tends to lead to interlayer errors (12). Torkaman et al used Conditional Generative Adversarial Networks (CGAN) corrected for myocardial attenuation, and the preliminary results showed no significant advantage (13).…”

Utilizing deep learning techniques to improve image quality and noise reduction in preclinical low‐dose PET images in the sinogram domain

“…In recent years, many deep learning techniques have been developed to handle PET image problems, providing a new dimension of precision and efficiency in healthcare. For example, in PET/MR brain imaging, a 2D‐UNet‐based attenuation correction method 14 and a modified 3D‐UNET were utilized to resolve problems in low‐dose brain PET imaging over the sinogram and image domain 15 . These innovative approaches leverage the full potential of DL to enhance the quality of imaging data.…”

Classical and Learned MR to Pseudo-CT Mappings for Accurate Transcranial Ultrasound Simulation

“…A variety of models and techniques have been used, including generative neural networks (GANs) [12], [13], supervised learning [14], contrastive learning [15], [16] and denoising diffusion probabilistic models [17]. Often, the network architecture accounts for features at multiple scales using a U-Net [11], [18]. The inputs to the neural network can be the full 3D image volumes [19], 2D slices along one or multiple planes [20], or small 3D patches [21].…”