Super-Resolution PET Imaging Using Convolutional Neural Networks

Abstract

Positron emission tomography (PET) suffers from severe resolution limitations which limit its quantitative accuracy. In this paper, we present a super-resolution (SR) imaging technique for PET based on convolutional neural networks (CNNs). To facilitate the resolution recovery process, we incorporate high-resolution (HR) anatomical information based on magnetic resonance (MR) imaging. We introduce the spatial location information of the input image patches as additional CNN inputs to accommodate the spatially-variant nature of the blur kernels in PET. We compared the performance of shallow (3-layer) and very deep (20-layer) CNNs with various combinations of the following inputs: low-resolution (LR) PET, radial locations, axial locations, and HR MR. To validate the CNN architectures, we performed both realistic simulation studies using the BrainWeb digital phantom and clinical neuroimaging data analysis. For both simulation and clinical studies, the LR PET images were based on the Siemens HR+ scanner. Two different scenarios were examined in simulation: one where the target HR image is the ground-truth phantom image and another where the target HR image is based on the Siemens HRRT scanner -a high-resolution dedicated brain PET scanner. The latter scenario was also examined using clinical neuroimaging datasets. A number of factors affected relative performance of the different CNN designs examined, including network depth, target image quality, and the resemblance between the target and anatomical images. In general, however, all CNNs outperformed classical penalized deconvolution techniques by large margins both qualitatively (e.g., edge and contrast recovery) and quantitatively (as indicated by two metrics: peak signal-to-noise-ratio and structural similarity index).

Dopamine and glutamate in schizophrenia: biology, symptoms and treatment

“…In addition to advances in hardware, further progress may be made by employing novel methods, such as super‐resolution techniques, in which multiple low‐resolution images are combined to create a high‐resolution image, or deep learning methods, where anatomical information from an MRI scan is used to help improve the resolution of the PET image. The limited resolution of in vivo imaging techniques means that it is difficult to directly test circuit level hypotheses regarding the nature of disrupted glutamate‐dopamine interactions in schizophrenia.…”

Error propagation analysis of seven partial volume correction algorithms for [18F]THK-5351 brain PET imaging

“…The study of error propagation must continue together with algorithm development. Furthermore, deep learning and machine learning PVC methods are emerging [ 44 ], and it remains to be seen, if these approaches automatically suppress error propagation, as this will have great impact on clinical translation.…”

The promise of artificial intelligence and deep learning in PET and SPECT imaging

“…Therefore, adding anatomical priors into the training procedure would make the model more accurate and robust. For resolution recovery, highresolution anatomical information obtained from MR imaging was employed along with spatially-variant bluring kernels to avoid information loss during image reconstruction [47]. Some groups devised strategies for deep learning-guided denoising models for synthesizing full-dose sinograms from their corresponding low-dose sinograms [48].…”

“…It is gratifying to see in overview the progress that AI has made, from early developments in neural networks to complex deep learning architectures, and more recently towards continuous learning AI in radiology [47]. Challenges remain, particularly in the areas of clinical validation and liability towards wider adoption, ethical and legal aspects and a number of other issues that need to be settled [169].…”