Ultrasound Imaging Improved by the Context Encoder Reconstruction Generative Adversarial Network

The three-dimensional (3D) imaging performance of a cone-beam computed tomography (CBCT) system can be quantitatively characterized by its 3D point spread function (PSF). To avoid the pitfalls associated with 3D PSF measurement through the use of a micro point phantom, we adopt an edge-based technique, which iteratively blurs a step edge into a spread edge with a Gaussian blurring kernel. In experiment, a small Teflon solid ball (diameter ~6mm) is used to provide step edges along the scanlines across the ball center, in terms of x-ray linear attenuation coefficient. Correspondingly, the spread edge profiles are extracted from the digital volume ball reconstructed by a CBCT system. From a spread edge profile, a step edge can be established by a rectification procedure (essentially an edge identification). A 1D PSF along a scanline is modeled as a Gaussian distribution, which is determined by iterative edge-blurring technique. The 1D PSFs on a cross section of the ball constitute a 2D PSF, and the 2D PSFs at three orthogonal cross sections are used to represent the 3D PSF at the position of the ball. By repositioning the ball phantom and repeating the procedure, we measured the 3D PSFs at 126 positions over half of the support space. Experiment shows that a CBCT system is of spatial variance and anisotropy in terms of FWHMs (full width at half maximum) of the blurring kernel. Both numerical and graphical presentations of PSF results are provided. As a result, the FWHM of the 3D PSF in our CBCT system varies in the range between 0.66 mm and 1.39 mm.

Introduction: In yttrium-90 imaging, image quality is highly dependent on the selection of energy window and collimator design becausetheY-90 bremsstrahlung photons have a continuous and broad energy distribution. The current study aimed to optimize the bremsstrahlung energy window setting and collimator for the improvement of both resolution and sensitivity. Material and Methods: In the present study, simulation of medical imaging nuclear detectors (SIMIND) Monte Carlo program was used to simulate Siemens Medical System Symbia. The SIMIND was utilized to generate the Y-90 bremsstrahlung single-photon emission computed tomography (SPECT) projection of the point source. Six energy windows settings and two collimators denoting medium energy and high energy were used in order to assess the effect of the energy window on the resolution. Results: The experimental measurements and simulation results showed a similar pattern in the point spread functions with the energy window. The simulation data indicated that the geometric component reached 73%for the energy window within the range of51-120keVusingthe high-energy (HE) collimator. In addition, the obtained results showed that the full width at half maximum (FWHM) and full width at tenth maximum (FWTM)(FWHM=7mm and FWTM=35mm)were higher in this window in comparison to those reported for other windows. Conclusion: According to the obtained results of the present study, the optimal energy window for Y-90 bremsstrahlung SPECT imaging was within the range of 51-120 keV. The obtained optimal energy window and optimal HE collimator had the potential to improve the image resolution and sensitivity of Y-90 SPECT images

Current amorphous silicon electronic portal imaging devices (a-Si EPIDs) that are frequently used in radiotherapy applications employ a metal plate/phosphor screen configuration to optimize x-ray detection efficiency. The phosphor acts to convert x rays into an optical signal that is detected by an underlying photodiode array. The dosimetric response of EPIDs has been well characterized, in part through the development of computational models. Such models, however, have generally made simplifying assumptions with regards to the transport of optical photons within these detectors. The goal of this work was to develop and experimentally validate a new Monte Carlo (MC) model of an a-Si EPID that simulates both x-ray and optical photon transport in a self-contained manner. Using this model the authors establish a definitive characterization of the effects of optical transport on the dosimetric response of a-Si EPIDs employing gadolinium oxysulfide phosphor screens. The Geant4 MC toolkit was used to develop a model of an a-Si EPID that employs standard electromagnetic and optical physics classes. The sensitivity of EPID response to uncertainties in optical transport parameters was evaluated by investigating their effects on the EPID point spread function (PSF). An optical blur kernel was also calculated to isolate the component of the PSF resulting purely from optical transport. A 6 MV photon source model was developed and integrated into the MC model to investigate EPID dosimetric response. Field size output factors and relative dose profiles were calculated for a set of open fields by separately scoring energy deposited in the phosphor and optical absorption events in the photodiode. These were then compared to quantify effects resulting from optical photon transport. The EPID model was validated against experimental measurements taken using a research EPID. Optical photon scatter within the phosphor screen noticeably broadened the PSF. Variations in optical transport parameters reported in the literature caused fluctuations in the PSF full width at half maximum (FWHM) and full width at tenth maximum (FWTM) of less than 3% and 5%, respectively, confirming model robustness. Greater deviations (up to 9.5% and 36% for FWHM and FWTM, respectively) were observed when optical parameters were largely different from reference values. When scoring energy deposition in the phosphor, measured and calculated output factors agreed within statistical uncertainties and at least 94% of the MC simulated profile data points passed 3%/3 mm γ-index criterion for all field sizes considered. Despite statistical uncertainties in optical simulations arising from computational limitations, no differences were observed between optical and energy deposition profiles. Simulations demonstrated noticeable blurring of the EPID PSF when scoring optical absorption events in the photodiode relative to energy deposition in the phosphor. However, modeling the standard electromagnetic transport alone should suffice when using MC methods to predict EPID dose-response to static, open 6 MV fields with a standard a-Si photodiode array. Therefore, using energy deposition in the phosphor as a surrogate for EPID dose-response is a valid approach that should not require additional corrections for optical transport effects in current a-Si EPIDs employing phosphor screens.

Context. Modeling the optical point spread function (PSF) is particularly challenging for adaptive optics (AO)-assisted observations owing to the its complex shape and spatial variations. Aims. We aim to (i) exhaustively demonstrate the accuracy of a recent analytical model from comparison with a large sample of imaged PSFs, (ii) assess the conditions for which the model is optimal, and (iii) unleash the strength of this framework to enable the joint estimation of atmospheric parameters, AO performance and static aberrations. Methods. We gathered 4812 on-sky PSFs obtained from seven AO systems and used the same fitting algorithm to test the model on various AO PSFs and diagnose AO performance from the model outputs. Finally, we highlight how this framework enables the characterization of the so-called low wind effect on the Spectro-Polarimetic High contrast imager for Exoplanets REsearch (LWE; SPHERE) instrument and piston cophasing errors on the Keck II telescope. Results. Over 4812 PSFs, the model reaches down to 4% of error on both the Strehl-ratio (SR) and full width at half maximum (FWHM). We particularly illustrate that the estimation of the Fried’s parameter, which is one of the model parameters, is consistent with known seeing statistics and follows expected trends in wavelength using the Multi Unit Spectroscopic Explorer instrument (λ6/5) and field (no variations) from Gemini South Adaptive Optics Imager images with a standard deviation of 0.4 cm. Finally, we show that we can retrieve a combination of differential piston, tip, and tilt modes introduced by the LWE that compares to ZELDA measurements, as well as segment piston errors from the Keck II telescope and particularly the stair mode that has already been revealed from previous studies. Conclusions. This model matches all types of AO PSFs at the level of 4% error and can be used for AO diagnosis, post-processing, and wavefront sensing purposes.

An approach to breaking through the diffraction limitation in infrared microscopies is put forward in this paper. In this method, instead of Gaussian pump beam, an intensive vortex beam is first focused on the sample, leading to the saturation absorption of peripheral molecules in the point spread function (PSF). The vortex beam is followed by a Gaussian probe beam with the same wavelength. Because of the previous saturation absorption, the probe beam can only be absorbed by the molecules near the center, resulting in a shrunk PSF which means super-resolution. Furthermore, the PSF of a system based on this approach is numerically simulated. With a 100 nJ pulse energy vortex beam and a 0.1 nJ pulse energy probe beam, the theoretical resolution FWHM (full width at half maximum) is measured to be about 236 nm which is 14 times better than that of the traditional infrared microscopy.

Time‐of‐flight (TOF) and resolution modeling (RM) algorithms are frequently used in clinical PET images, and inclusion of these corrections should measurably improve image quality. We quantified the effects of these correction algorithms on reconstructed images via the following metrics: recovery coefficients (RCs), contrast‐to‐noise ratio (CNR), noise‐power spectrum (NPS), modulation transfer function (MTF), and the full width at half maximum (FWHM) of a point source. The goal of this experiment was to assess the effects of the correction algorithms when applied singly or together. Two different phantom tests were performed and analyzed by custom software. FWHM and MTF were measured using capillary tube point sources, while RCs, CNR, and NPS were measured using an image quality body phantom. Images were reconstructed with both TOF and RM, only TOF, only RM, or neither correction. The remaining reconstruction parameters used the standard clinical protocol. RM improved RCs, FWHM, and MTF, without increasing overall noise significantly. TOF improves CNR for small objects FWHM or MTF but did not decrease noise. RCs were not statistically improved by enabling these algorithms. Inclusion of both correction algorithms in image reconstruction provides an overall improvement to all metrics relative to the uncorrected image, but not by a significant margin in multiple aspects.

The ntCT nano tomography system is a geometrically magnifying X-ray microscopy system integrating the recent Excillum NanoTube nano-focus X-ray source and a CdTe photon counting detector from Dectris. The system's modulation transfer function (MTF) and corresponding point spread function (PSF) are characterized by analyzing the contrast visibility of periodic structures of a star pattern featuring line width from 150 nm to 1.5 μm. The results, which can be attributed to the characteristics of the source spot, are crosschecked by scanning the source's electron focus over an edge of the structured transmission target in order to obtain an independent measurement of its point spread function. For frequencies above 1000 linepairs/mm, the MTF is found to correspond to a Gaussian PSF of 250 nm full width at half maximum (FWHM) . The lower frequency range down to 340 linepairs/mm shows an additional Gaussian contribution of 1 μm FWHM . The resulting resolution ranges at 3200 linepairs/mm, which is consistent with the visual detectability of the smallest 150 nm structures within the imaged star pattern.

The penetration depth of ballistic optical imaging technologies is limited by light scattering. To study the effect of scattering on optical-resolution photoacoustic microscopy (OR-PAM), we divided the signals in OR-PAM into two classes: one is from the target volume defined by the optical resolution cell (Class I); the other is from the rest of the acoustic resolution cell (Class II). We developed a way to simulate the point spread function (PSF) of our OR-PAM system considering both optical illumination and acoustic detection, then used the PSF to calculate the contributions of each class of signal to the total signal at different focal depths. Our simulation results showed that: 1) The Class II signal decays much more slowly than the Class I signal; 2) The full width at half maximum (FWHM) of the PSF for the focal depth of 0.9 transport mean free path (TMFP) is not broadened much (~10%) compared with that for a clear medium; 3) Image contrast is degraded with increasing depth when there is a uniform absorption background.

In preparation of future Multi-Object Spectrographs (MOS) whose one of the major role is to provide an extensive statistical studies of high redshifted galaxies surveyed, the demonstrator Canary has been designed to tackle technical challenges related to open-loop Adaptive-Optics (AO) control with jointed Natural Guide Star (NGS) and Laser Guide Star (LGS) tomography. We have developed a Point Spread Function (PSF)-Reconstruction algorithm dedicated to MOAO systems using system telemetry to estimate the PSF potentially anywhere in the observed field, a prerequisite to post- process AO-corrected observations in Integral Field Spectroscopy (IFS). In this paper we show how to handle off-axis data to estimate the PSF using atmospheric tomography and compare it to a classical approach that uses on-axis residual phase from a truth sensor observing a natural bright source. We have reconstructed over 450 on-sky Canary PSFs and we get bias/1-${\sigma}$ standard-deviation (std) of 1.3/4.8 on the H-band Strehl ratio (SR) with 92.3% of correlation between reconstructed and sky SR. On the Full Width at Half Maximum (FWHM), we get respectively 2.94 mas, 19.9 mas and 88.3% for the bias, std and correlation. The reference method achieves 0.4/3.5/95% on the SR and 2.71 mas/14.9 mas/92.5% on the FWHM for the bias/std/correlation.

The visual quality of ultrasound (US) images is crucial for clinical diagnosis and treatment. The main source of image quality degradation is the inherent speckle noise generated during US image acquisition. Current deep learning-based methods cannot preserve the maximum boundary contrast when removing noise and speckle. In this paper, we address the issue by proposing a novel wavelet-based generative adversarial network (GAN) for real-time high-quality US image reconstruction, viz. WGAN-DUS. First, we propose a batch normalization module (BNM) to balance the importance of each sub-band image and fuse sub-band features simultaneously. Then, a wavelet reconstruction module (WRM) integrated with a cascade of wavelet residual channel attention block (WRCAB) is proposed to extract distinctive sub-band features used to reconstruct denoised images. A gradual tuning strategy is proposed to fine-tune our generator for better despeckling performance. We further propose a wavelet-based discriminator and a comprehensive loss function to effectively suppress speckle noise and preserve the image features. Besides, we have designed an algorithm to estimate the noise levels during despeckling of real US images. The performance of our network was then evaluated on natural, synthetic, simulated and clinical US images and compared against various despeckling methods. To verify the feasibility of WGAN-DUS, we further extend our work to uterine fibroid segmentation with the denoised US image of the proposed approach. Experimental result demonstrates that our proposed method is feasible and can be generalized to clinical applications for despeckling of US images in real-time without losing its fine details.

In ultrasound imaging using plane-wave (PW) techniques, image quality and contrast often suffer, especially when examining anechoic structures. Traditional beamforming methods like Delay-and-Sum or coherent PW compounding face limitations in balancing resolution and frame rate, which can result in suboptimal diagnosticaccuracy. This study aims to introduce a modular beamforming pipeline that overcomes these challenges and enhances PW image quality. By dividing the beamforming process into two modules: a multi-attention U-Net based model for capturing complex dependencies in time-delayed data and a super-resolution model for scaling up to the original B-mode image grid. We seek to improve PW image quality and modularity in the ultrasound imagingprocess. We implemented a modular beamforming approach, comprising a multi-attention U-Net model and a super-resolution model. We conducted experiments using simulated, experimental, and in vivo data from the PICMUS dataset to evaluate the performance of our pipeline against conventional methods such as PW1, PW9, and U-Net. Key metrics assessed included contrast-to-noise ratio (CNR), contrast ratio (CR), generalized Contrast-to-Noise Ratio (gCNR), andresolution. Our model demonstrated superior performance across all metrics. In simulated data, the model achieved a 0.99 improvement in CNR, a 3.5 dB increase in CR, and 18% in gCNR compared to PW1. Experimental data showed a 0.7 enhancement in CNR and a 8.6 dB improvement in CR and 24% increase in gCNR when compared to PW1. In vivo data also revealed significant improvements, with a 1.9 dB increase in CR and a 0.15 enhancement in CNR over PW1. The enhanced performance in the anechoic cyst region underscores the model's effectiveness in improving imagequality. The proposed modular beamforming approach offers significant advantages, including adaptability and improved image quality, despite the complexity of managing two models concurrently. The pipeline's flexibility in frame rate and image quality allows for customization based on specific clinical applications, making it a promising alternative to traditionalmethods.

We have developed and tested an MR-compatible gamma probe to simultaneously obtain anatomical information and functional information during surgery. The probe consists of a probe head, an optical fibre bundle and a photo-multiplier tube (PMT). The NaI(Tl) scintillator contained in the probe head is connected to a 7 m optical fibre bundle that transfers the scintillation photons produced in the NaI(Tl) to an area of low magnetic field or out of the MR-scanner's magnetic shielded room. Although the light loss due to the optical fibre bundle was more than 90%, the photo-peak of the gamma for Co-57 (122 keV) could be observed. The point spread function was 4.5 mm full width at half maximum (FWHM) at 5 mm from the collimator surface for 122 keV gamma photons. Furthermore, there was no sensitivity change outside the MR-scanner, inside the MR-scanner without scanning and inside the MR-scanner with scanning. The probe produced a small artefact on the phantom image of the MR-scanner due to the susceptibility difference of the alloy used for the collimator and the shield. However, the artefact was only limited to area surrounding the probe. These results indicate that the developed MR-compatible gamma probe will make it possible to realize the combined MR/RI guided surgery that provides surgeons with anatomical and RI distribution information simultaneously.

In this study, we evaluated the improvement of image quality in digital breast tomosynthesis under low-radiation dose conditions of pre-reconstruction processing using conditional generative adversarial networks [cGAN (pix2pix)]. Pix2pix pre-reconstruction processing with filtered back projection (FBP) was compared with and without multiscale bilateral filtering (MSBF) during pre-reconstruction processing. Noise reduction and preserve contrast rates were compared using full width at half-maximum (FWHM), contrast-to-noise ratio (CNR), peak signal-to-noise ratio (PSNR), and structural similarity (SSIM) in the in-focus plane using a BR3D phantom at various radiation doses [reference-dose (automatic exposure control reference dose: AECrd), 50% and 75% reduction of AECrd] and phantom thicknesses (40 mm, 50 mm, and 60 mm). The overall performance of pix2pix pre-reconstruction processing was effective in terms of FWHM, PSNR, and SSIM. At ~50% radiation-dose reduction, FWHM yielded good results independently of the microcalcification size used in the BR3D phantom, and good noise reduction and preserved contrast. PSNR results showed that pix2pix pre-reconstruction processing represented the minimum in the error with reference FBP images at an approximately 50% reduction in radiation-dose. SSIM analysis indicated that pix2pix pre-reconstruction processing yielded superior similarity when compared with and without MSBF pre-reconstruction processing at ~50% radiation-dose reduction, with features most similar to the reference FBP images. Thus, pix2pix pre-reconstruction processing is promising for reducing noise with preserve contrast and radiation-dose reduction in clinical practice.

The degradation of image quality caused by respiration is a major impediment to accurate lesion detection in single photon emission computed tomography (SPECT) imaging. This study was performed to evaluate the effects of lung motion on image quantification. A small animal SPECT system with NaI(Tl) was modeled in the Geant4 application for tomographic emission (GATE) simulation for a lung lesion using a 4D mouse whole-body phantom. SPECT images were obtained using 120 projection views acquired from 0o to 360o with a 3o step. Slices were reconstructed using ordered subsets expectation maximization (OS-EM) without attenuation correction with five iterations and four subsets. Image quality was compared between the static mode without respiratory motion, and dynamic mode with respiratory motion in terms of spatial resolution was measured by the full width at half maximum (FWHM), signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR). The FWHM of the non-gated image and the respiratory gated image were also compared. Spatial resolution improved as activity increased and lesion diameter decreased in the static and dynamic modes. The SNR and CNR increased significantly as lesion activity increased and lesion diameter decreased. Our results show that respiratory motion leads to reduced contrast and quantitative accuracy and that image quantification depends on both the amplitude and the pattern of the respiratory motion. We verified that respiratory motion can have a major effect on the accuracy of measurement of lung lesions and that respiratory gating can reduce activity smearing on SPECT images.

Chirp coded excitation is an effective method to improve the signal-to-noise ratio (SNR) and penetration depth of high-frequency endoscopic ultrasound (EUS) imaging. In coded excitation, pulse compression is applied to compress the elongated coded signals into a short pulse, which determines the final imaging performance, including spatial resolution and SNR. However, with the current pulse compression methods, it is hard to get high performance in the peak sidelobe level (PSL), image contrast, and axial resolution at the same time. To solve this problem, in this article, a new method named compressed pulse weighting method (CPWM) was proposed based on the combination of two kinds of pulse compression signals. A brief theoretical derivation proved the feasibility of method. The proposed method was evaluated by the simulation and phantom experiments. Compared with traditional method, the results showed that the proposed adaptive weighting method can provide increases of 32.42% in the penetration depth, 9.48 dB in the SNR, 5.60 dB in the contrast ratio (CR), 5.46 in the contrast-to-noise ratio (CNR), and 0.13 mm in the axial imaging resolution for 12-MHz EUS. Therefore, this method can effectively improve the ultrasound penetration depth and imaging quality, which made it have good potential for high-frequency ultrasound imaging.