Fast Count-Regulated OSEM Reconstruction With Adaptive Resolution Recovery

Unified Noise-aware Network for Low-count PET Denoising with Varying Count Levels

As PET imaging is accompanied by substantial radiation exposure and cancer risk, reducing radiation dose in PET scans is an important topic. However, low-count PET scans often suffer from high image noise, which can negatively impact image quality and diagnostic performance. Recent advances in deep learning have shown great potential for recovering underlying signal from noisy counterparts. However, neural networks trained on a specific noise level cannot be easily generalized to other noise levels due to different noise amplitude and variances. To obtain optimal denoised results, we may need to train multiple networks using data with different noise levels. But this approach may be infeasible in reality due to limited data availability. Denoising dynamic PET images presents additional challenge due to tracer decay and continuously changing noise levels across dynamic frames. To address these issues, we propose a Unified Noise-aware Network (UNN) that combines multiple sub-networks with varying denoising power to generate optimal denoised results regardless of the input noise levels. Evaluated using large-scale data from two medical centers with different vendors, presented results showed that the UNN can consistently produce promising denoised results regardless of input noise levels, and demonstrate superior performance over networks trained on single noise level data, especially for extremely low-count data.

Performance Evaluation of a Preclinical SPECT Scanner with a Collimator Designed for Medium-Sized Animals

Background

Equipped with two stationary detectors, a large bore collimator for medium-sized animals has been recently introduced for dedicated preclinical single-photon emission computed tomography (SPECT) imaging. We aimed to evaluate the basic performance of the system using phantoms and healthy rabbits.

Methods

A general-purpose medium-sized animal (GP-MSA) collimator with 135 mm bore diameter and thirty-three holes of 2.5 mm diameter was installed on an ultrahigh-resolution scanner equipped with two large stationary detectors (U-SPECT5-E/CT). The sensitivity and uniformity were investigated using a point source and a cylinder phantom containing ^99mTc-pertechnetate, respectively. Uniformity (in %) was derived using volumes of interest (VOIs) on images of the cylinder phantom and calculated as [(maximum count − minimum count)/(maximum count + minimum count) × 100], with lower values of % indicating superior performance. The spatial resolution and contrast-to-noise ratios (CNRs) were evaluated with images of a hot-rod Derenzo phantom using different activity concentrations. Feasibility of in vivo SPECT imaging was finally confirmed by rabbit imaging with the most commonly used clinical myocardial perfusion SPECT agent [^99mTc]Tc-sestamibi (dynamic acquisition with a scan time of 5 min).

Results

In the performance evaluation, a sensitivity of 790 cps/MBq, a spatial resolution with the hot-rod phantom of 2.5 mm, and a uniformity of 39.2% were achieved. The CNRs of the rod size 2.5 mm were 1.37, 1.24, 1.20, and 0.85 for activity concentration of 29.2, 1.0, 0.5, and 0.1 MBq/mL, respectively. Dynamic SPECT imaging in rabbits allowed to visualize most of the thorax and to generate time-activity curves of the left myocardial wall and ventricular cavity.

Conclusion

Preclinical U-SPECT5-E/CT equipped with a large bore collimator demonstrated adequate sensitivity and resolution for in vivo rabbit imaging. Along with its unique features of SPECT molecular functional imaging is a superior collimator technology that is applicable to medium-sized animal models and thus may promote translational research for diagnostic purposes and development of novel therapeutics.

Postreconstruction filtering of 3D PET images by using weighted higher-order singular value decomposition

Background

Positron emission tomography (PET) always suffers from high levels of noise due to the constraints of the injected dose and acquisition time, especially in the studies of dynamic PET imaging. To improve the quality of PET image, several approaches have been introduced to suppress noise. However, traditional filters often blur the image edges, or erase small detail, or rely on multiple parameters. In order to solve such problems, nonlocal denoising methods have been adapted to denoise PET images.

Methods

In this paper, we propose to use the weighted higher-order singular value decomposition for PET image denoising. We first modeled the noise in the PET image as Poisson distribution. Then, we transformed the noise to an additive Gaussian noise by use of the anscombe root transformation. Finally, we denoised the transformed image using the proposed higher-order singular value decomposition (HOSVD)-based algorithms. The denoised results were compared with results from some general filters by performing physical phantom and mice studies.

Results

Compared to other commonly used filters, HOSVD-based denoising algorithms can preserve boundaries and quantitative accuracy better. The spatial resolution and the low activity features in PET image also can be preserved by use of HOSVD-based methods. Comparing with the standard HOSVD-based algorithm, the proposed weighted HOSVD algorithm can suppress the stair-step artifact, and the time-consumption is about half of that needed by the Wiener-augmented HOSVD algorithm.

Conclusions

The proposed weighted HOSVD denoising algorithm can suppress noise while better preserving of boundary and quantity in PET images.

Electronic supplementary material

The online version of this article (doi:10.1186/s12938-016-0221-y) contains supplementary material, which is available to authorized users.

uPAR-targeted multimodal tracer for pre- and intraoperative imaging in cancer surgery

Pre- and intraoperative diagnostic techniques facilitating tumor staging are of paramount importance in colorectal cancer surgery. The urokinase receptor (uPAR) plays an important role in the development of cancer, tumor invasion, angiogenesis, and metastasis and over-expression is found in the majority of carcinomas. This study aims to develop the first clinically relevant anti-uPAR antibody-based imaging agent that combines nuclear (¹¹¹In) and real-time near-infrared (NIR) fluorescent imaging (ZW800-1). Conjugation and binding capacities were investigated and validated in vitro using spectrophotometry and cell-based assays. In vivo, three human colorectal xenograft models were used including an orthotopic peritoneal carcinomatosis model to image small tumors. Nuclear and NIR fluorescent signals showed clear tumor delineation between 24h and 72h post-injection, with highest tumor-to-background ratios of 5.0 ± 1.3 at 72h using fluorescence and 4.2 ± 0.1 at 24h with radioactivity. 1-2 mm sized tumors could be clearly recognized by their fluorescent rim. This study showed the feasibility of an uPAR-recognizing multimodal agent to visualize tumors during image-guided resections using NIR fluorescence, whereas its nuclear component assisted in the pre-operative non-invasive recognition of tumors using SPECT imaging. This strategy can assist in surgical planning and subsequent precision surgery to reduce the number of incomplete resections.

Quarter-millimeter-resolution molecular mouse imaging with U-SPECT⁺

Limited spatial resolution of preclinical positron emission tomography (PET) and single-photon emission computed tomography (SPECT) has slowed down applications of molecular imaging in small animals. Here we present the latest-generation U-SPECT system (U-SPECT⁺, MILabs, Utrecht, the Netherlands) enabling radionuclide imaging of mice with quarter-millimeter resolution. The system was equipped with the newest high-resolution collimator with 0.25 mm diameter circular pinholes. It was calibrated with technetium-99 m point source measurements from which the system matrix was calculated. Images were reconstructed using pixel-based ordered subset expectation maximization (OSEM). Various phantoms and mouse SPECT scans were acquired. The reconstructed spatial resolution (the smallest visible capillary diameter in a hot-rod resolution phantom) was 0.25 mm. Knee joint images show tiny structures such as the femur epicondyle sulcus, as well as a clear separation between cortical and trabecular bone structures. In addition, time-activity curves of the lumbar spine illustrated that tracer dynamics in tiny tissue amounts could be measured. U-SPECT⁺ allows discrimination between molecular concentrations in adjacent volumes of as small as 0.015 μL, which is significantly better than can be imaged by any existing SPECT or PET system. This increase in the level of detail makes it more and more attractive to replace ex vivo methods and allows monitoring biological processes in tiny parts of organs in vivo.

Evaluation of bias and variance in low-count OSEM list mode reconstruction

Statistical algorithms have been widely used in PET image reconstruction. The maximum likelihood expectation maximization reconstruction has been shown to produce bias in applications where images are reconstructed from a relatively small number of counts. In this study, image bias and variability in low-count OSEM reconstruction are investigated on images reconstructed with MOLAR (motion-compensation OSEM list-mode algorithm for resolution-recovery reconstruction) platform. A human brain ([(11)C]AFM) and a NEMA phantom are used in the simulation and real experiments respectively, for the HRRT and Biograph mCT. Image reconstructions were repeated with different combinations of subsets and iterations. Regions of interest were defined on low-activity and high-activity regions to evaluate the bias and noise at matched effective iteration numbers (iterations × subsets). Minimal negative biases and no positive biases were found at moderate count levels and less than 5% negative bias was found using extremely low levels of counts (0.2 M NEC). At any given count level, other factors, such as subset numbers and frame-based scatter correction may introduce small biases (1-5%) in the reconstructed images. The observed bias was substantially lower than that reported in the literature, perhaps due to the use of point spread function and/or other implementation methods in MOLAR.