Hybrid Sparsity Model for Fast Terahertz Imaging

In order to shorten the long-term image acquisition time of the terahertz time domain spectroscopy imaging system while ensuring the imaging quality, a hybrid sparsity model (HSM) is proposed for fast terahertz imaging in this paper, which incorporates both intrinsic sparsity prior and nonlocal self-similarity constraints in a unified statistical model. In HSM, a weighted exponentiation shift-invariant wavelet transform is introduced to enhance the sparsity of the terahertz image. Simultaneously, the nonlocal self-similarity by means of the three-dimensional sparsity in the transform domain is exploited to ensure high-quality terahertz image reconstruction. Finally, a new split Bregman-based iteration algorithm is developed to solve the terahertz imaging model more efficiently. Experiments are presented to verify the effectiveness of the proposed approach.

Deep learning early stopping for non-degenerate ghost imaging

Quantum ghost imaging offers many advantages over classical imaging, including the ability to probe an object with one wavelength and record the image with another (non-degenerate ghost imaging), but suffers from slow image reconstruction due to sparsity and probabilistic arrival positions of photons. Here, we propose a two-step deep learning approach to establish an optimal early stopping point based on object recognition, even for sparsely filled images. In step one we enhance the reconstructed image after every measurement by a deep convolutional auto-encoder, followed by step two in which a classifier is used to recognise the image. We test this approach on a non-degenerate ghost imaging setup while varying physical parameters such as the mask type and resolution. We achieved a fivefold decrease in image acquisition time at a recognition confidence of $$75\%$$ 75 % . The significant reduction in experimental running time is an important step towards real-time ghost imaging, as well as object recognition with few photons, e.g., in the detection of light sensitive structures.

A Review on Medical Image Super Resolution with Application of Deep Learning

Super resolution problems are often discussed in medical imaging. The spatial resolution of medical images is insufficient due to limitations such as image acquisition time, low radiation dose or hardware limitations. Various super-resolution methods have been proposed to solve these problems, such as optimization or learning-based approaches. Recently, deep learning methodologies have become a thriving technology and are evolving at an exponential rate. We believe we need to write a review to illustrate the current state of deep learning in super-resolution medical imaging. In this article, we provide an overview of image resolution and the deep learning introduced in super resolution. This document describes super resolution for single images versus super resolution for multiple images, evaluation metrics and loss functions.

Deep Learning Early Stopping for Non-degenerate Ghost Imaging

Quantum ghost imaging offers many advantages over classical imaging, including the ability to probe an object with one wavelength and record the image with another (non-degenerate ghost imaging), but suffers from slow image reconstruction due to sparsity and probabilistic arrival positions of photons. Here, we propose a two-step deep learning approach to establish an optimal early stopping point based on object recognition, even for sparsely filled images. In step one we enhance the reconstructed image after every measurement by a deep convolutional auto encoder, followed by step two in which a classifier is used to recognise the image. We test this approach on a non degenerate ghost imaging setup while varying physical parameters such as the mask type and resolution. We achieved a 5-fold decrease in image acquisition time at a recognition confidence of 75%. The significant reduction in experimental running time is an important step towards real-time ghost imaging, as well as object recognition with few photons, e.g., in the detection of light sensitive structures.

Initial Experience with 64Cu-DOTATATE Digital PET of Patients with Neuroendocrine Neoplasms: Comparison with Analog PET

The recent introduction of solid-state detectors in clinical positron emission tomography (PET) scanners has significantly improved image quality and spatial resolution and shortened acquisition time compared to conventional analog PET scanners. In an initial evaluation of the performance of our newly acquired Siemens Biograph Vision 600 PET/CT (digital PET/CT) scanner for 64Cu-DOTATATE imaging, we compared PET/CT acquisitions from patients with neuroendocrine neoplasms (NENs) grades 1 and 2 and stable disease on CT who were scanned on both our Siemens Biograph 128 mCT PET/CT (analog PET/CT) and digital PET/CT within 6 months as part of their routine clinical management. Five patients fulfilled the criteria and were included in the analysis. The digital PET acquisition time was less than 1/3 of the analog PET acquisition time (digital PET, mean (min:s): 08:20 (range, 07:59–09:45); analog PET, 25:28 (24:39–28:44), p < 0.001). All 44 lesions detected on the analog PET with corresponding structural correlates on the CT were also found on the digital PET performed 137 (107–176) days later. Our initial findings suggest that digital 64Cu-DOTATATE PET can successfully be performed in patients with NENs using an image acquisition time of only 1/3 of what is used for an analog 64Cu-DOTATATE PET.

Single-cell radioluminescence microscopy with two-fold higher sensitivity using dual scintillator configuration

Radioluminescence microscopy (RLM) is an imaging technique that allows quantitative analysis of clinical radiolabeled drugs and probes in single cells. However, the modality suffers from slow data acquisition (10 – 15 minutes), thus critically affecting experiments with short-lived radioactive drugs. To overcome this issue, we suggest an approach that significantly accelerates data collection. Instead of using a single scintillator to image the decay of radioactive molecules, we sandwiched the radiolabeled cells between two scintillators. As proof of concept, we imaged cells labeled with [18F]FDG, a radioactive glucose popularly used in oncology to image tumors. Results show that the double scintillator configuration increases the microscope sensitivity by two-fold, thus reducing the image acquisition time by half to achieve the same result as the single scintillator approach. The experimental results were also compared with Geant4 Monte Carlo simulation to confirm the two-fold increase in sensitivity with only minor degradation in spatial resolution. Overall, these findings suggest that the double scintillator configuration can be used to perform time-sensitive studies such as cell pharmacokinetics or cell uptake of short-lived radiotracers.