Magnetic resonance imaging (MRI) provides excellent soft-tissue contrast for clinical diagnoses and research which underpin many recent breakthroughs in medicine and biology. The post-processing of reconstructed MR images is often automated for incorporation into MRI scanners by the manufacturers and increasingly plays a critical role in the final image quality for clinical reporting and interpretation. For image enhancement and correction, the post-processing steps include noise reduction, image artefact correction, and image resolution improvements. With the recent success of deep learning in many research fields, there is great potential to apply deep learning for MR image enhancement, and recent publications have demonstrated promising results. Motivated by the rapidly growing literature in this area, in this review paper, we provide a comprehensive overview of deep learning-based methods for post-processing MR images to enhance image quality and correct image artefacts. We aim to provide researchers in MRI or other research fields, including computer vision and image processing, a literature survey of deep learning approaches for MR image enhancement. We discuss the current limitations of the application of artificial intelligence in MRI and highlight possible directions for future developments. In the era of deep learning, we highlight the importance of a critical appraisal of the explanatory information provided and the generalizability of deep learning algorithms in medical imaging.

Image processing plays a crucial role in maximising diagnostic quality of positron emission tomography (PET) images. Recently, deep learning methods developed across many fields have shown tremendous potential when applied to medical image enhancement, resulting in a rich and rapidly advancing literature surrounding this subject. This review encapsulates methods for integrating deep learning into PET image reconstruction and post-processing for low-dose imaging and resolution enhancement. A brief introduction to conventional image processing techniques in PET is firstly presented. We then review methods which integrate deep learning into the image reconstruction framework as either deep learning-based regularisation or as a fully data-driven mapping from measured signal to images. Deep learning-based post-processing methods for low-dose imaging, temporal resolution enhancement and spatial resolution enhancement are also reviewed. Finally, the challenges associated with applying deep learning to enhance PET images in the clinical setting are discussed and future research directions to address these challenges are presented.

Background 4-[18F] fluorobenzyl dexetimide (F-DEX) is the first non-subtype selective fluorine-18 labelled tracer for muscarinic receptors (mAChR) used in humans. A recent first-in-human study found high regional brain uptake with low variation in normal subjects. Disturbance of mAChR has been reported in Alzheimer’s and Parkinson’s disease, schizophrenia and depression and various cardiac diseases. The following work assesses the biodistribution, organ tracer kinetics and radiation dose associated with F-DEX. Method Dose calculations were based on activity uptake derived from multiple time point whole body PET CT imaging and the organ-specific dosimetric S-factors derived from the ICRP 133 standard man and woman mathematical phantoms. Effective doses were calculated using the latest ICRP tissue weighting factors. Results Serial images and time activity curves demonstrate high brain and left ventricular myocardial uptake (5% and 0.65% of injected activity, respectively) with greater retention in brain than myocardium. The mean effective dose was in concordance with other 18F labelled tracers at 19.70 ± 2.27 μSv/MBq. The largest absorbed doses were in the liver (52.91 ± 1.46 μGy/MBq) and heart wall (43.94 ± 12.88 μGy/MBq) for standard man and the liver (61.66 ± 13.61 μGy/MBq) and lungs (40.93 ± 3.11 μGy/MBq) for standard woman. The absorbed dose to all organs, most notably, the red bone marrow (20.03 ± 2.89 μGy/MBq) was sufficiently low to ensure no toxicity after numerous follow-up procedures. Conclusions The radiation dose associated with an administration of F-DEX is comparable to that of other 18F labelled tracers such as FDG (19.0 μSv/MBq) and lower than tracers used for SPECT imaging of muscarinic receptors (I-DEX 28.5 μSv/MBq). Clinical use would likely result in an effective dose less than 4 mSv for the ICRP 133 standard phantoms after dose optimisation allowing justification for numerous follow-up procedures. Recent results from first in-human studies and a comparatively low radiation dose make F-DEX an attractive option for future applications of imaging muscarinic receptors in the brain. Further investigation of the potential of F-DEX for imaging parasympathetic innervation of the heart may be warranted.

Purpose A method for calculating nuclear medicine ionization chamber (NMIC) calibration settings with a Monte Carlo model is presented and validated against physical measurements. This work provides Monte Carlo–calculated calibration settings for select isotopes with no current manufacturer recommendations and a method by which NMIC manufacturers or standards laboratories can utilize highly detailed specifications to calculate comprehensive lists of calibration settings for general isotopes. Methods A Monte Carlo model of a Capintec PET series NMIC was developed and used to calculate the chamber response to relevant radioactive decay products over an energy range relevant to nuclear medicine. The photon detection efficiency (PDE) of a high purity germanium (HPGe) detector was modeled and physically validated to facilitate measurements of NMIC calibration settings with HPGe detector spectroscopy. Modeled NMIC response to various isotopes was compared against spectroscopic measurements and National Institute of Standards and Technology (NIST)‐validated calibration settings to validate the Monte Carlo–calculated NMIC calibration settings. Results HPGe detector PDE was validated against the physical measurements to within 3.3%$3.3\%$ at 95%$95\%$ confidence and used to measure calibration settings, which produced activity readings 0.7%$0.7\%$, 1.6%$1.6\%$, 0.8%$0.8\%$, and 1.0%$1.0\%$ different than those validated by NIST for 11$^{11}$C, 18$^{18}$F, 68$^{68}$Ga, and 64$^{64}$Cu respectively. The Monte Carlo model of the NMIC reproduced measured calibration settings to within 7%$7\%$ at 95%$95\%$ confidence for isotopes with a sufficiently small yield of low energy photons. Conclusions A method of calculating NMIC calibration settings with Monte Carlo modeling has been developed and validated against HPGe detector spectroscopy. NMIC manufacturers or standards laboratories can use more detailed specifications of the chamber geometries to extend the applicability of this method to a wider range of isotopes.

No abstract