Parallel hybrid bispectrum-multi-frame blind deconvolution image reconstruction technique

There is high demand for deploying real-time or near real-time implementation of image reconstruction techniques for intelligence, surveillance, and reconnaissance (ISR) systems, as well as biomedical systems, and border applications. While many image restoration techniques have been implemented, their computational complexity makes them unsuitable for real-time systems. This work presents massively parallelized algorithms for two and three-dimensional image reconstruction of objects using deconvolution and bispectrum techniques. Speckle imaging (SI) techniques are more preferred for phase reconstruction because of its high restoration quality at the lower values of signal-to-noise ratio (SNR), especially for smaller targets, and its shift-invariant property which makes it insensitive to random motion of an object. This work proposes, a novel strategy for parallelization of recursive bispectrum algorithm. Furthermore, parallel bispectrum speckle imaging technique is used as an initial step for parallelized multi-frame blind deconvolution (MFBD) image restoration algorithm in order to speed-up the reconstruction process for anisoplanatic, long horizontal path imaging. The parallel hybrid Bispectrum- MFBD (B-MFBD) technique, through massive parallelization, provides significantly large improvement to both bispectrum SI and MFBD parts of the hybrid algorithm. The work also investigates phase diversity (PD) technique as the special case of MFBD algorithm, where the object is observed with a stack of intentionally defocused images. The known defocused information between phases is used to estimate the object and aberration function. The multi-plane PD was implemented for wide field fluorescence microscopy imaging of three-dimensional objects. The PD algorithm was subsequently used to generate 3D point cloud data for the image from its single view.

Simultaneous multi-slice (SMS) bSSFP imaging enables stress myocardial perfusion imaging with high spatial resolution and increased spatial coverage. Standard parallel imaging techniques (e.g., TGRAPPA) can be used for image reconstruction but result in high noise level. Alternatively, iterative reconstruction techniques based on temporal regularization (ITER) improve image quality but are associated with reduced temporal signal fidelity and long computation time limiting their online use. The aim is to develop an image reconstruction technique for SMS-bSSFP myocardial perfusion imaging combining parallel imaging and image-based denoising using a novel noise map estimation network (NoiseMapNet), which preserves both sharpness and temporal signal profiles and that has low computational cost. The proposed reconstruction of SMS images consists of a standard temporal parallel imaging reconstruction (TGRAPPA) with motion correction (MOCO) followed by image denoising using NoiseMapNet. NoiseMapNet is a deep learning network based on a 2D Unet architecture and aims to predict a noise map from an input noisy image, which is then subtracted from the noisy image to generate the denoised image. This approach was evaluated in 17 patients who underwent stress perfusion imaging using a SMS-bSSFP sequence. Images were reconstructed with (a) TGRAPPA with MOCO (thereafter referred to as TGRAPPA), (b) iterative reconstruction with integrated motion compensation (ITER), and (c) proposed NoiseMapNet-based reconstruction. Normalized mean squared error (NMSE) with respect to TGRAPPA, myocardial sharpness, image quality, perceived SNR (pSNR), and number of diagnostic segments were evaluated. NMSE of NoiseMapNet was lower than using ITER for both myocardium (0.045 ± 0.021 vs. 0.172 ± 0.041, p < 0.001) and left ventricular blood pool (0.025 ± 0.014 vs. 0.069 ± 0.020, p < 0.001). There were no significant differences between all methods for myocardial sharpness (p = 0.77) and number of diagnostic segments (p = 0.36). ITER led to higher image quality than NoiseMapNet/TGRAPPA (2.7 ± 0.4 vs. 1.8 ± 0.4/1.3 ± 0.6, p < 0.001) and higher pSNR than NoiseMapNet/TGRAPPA (3.0 ± 0.0 vs. 2.0 ± 0.0/1.3 ± 0.6, p < 0.001). Importantly, NoiseMapNet yielded higher pSNR (p < 0.001) and image quality (p < 0.008) than TGRAPPA. Computation time of NoiseMapNet was only 20s for one entire dataset. NoiseMapNet-based reconstruction enables fast SMS image reconstruction for stress myocardial perfusion imaging while preserving sharpness and temporal signal profiles.

IntroductionLow back pain is a global health issue causing disability and missed work days. Commonly used MRI scans including T1-weighted and T2-weighted images provide detailed information of the spine and surrounding tissues. Artificial intelligence showed promise in improving image quality and simultaneously reducing scan time. This study evaluates the performance of deep learning (DL)-based T2 turbo spin-echo (TSE, T2DLR) and T1 TSE (T1DLR) in lumbar spine imaging regarding acquisition time, image quality, artifact resistance, and diagnostic confidence.Material and methodsThis retrospective monocentric study included 60 patients with lower back pain who underwent lumbar spinal MRI between February and April 2023. MRI parameters and DL reconstruction (DLR) techniques were utilized to acquire images. Two neuroradiologists independently evaluated image datasets based on various parameters using a 4-point Likert scale.ResultsAccelerated imaging showed significantly less image noise and artifacts, as well as better image sharpness, compared to standard imaging. Overall image quality and diagnostic confidence were higher in accelerated imaging. Relevant disk herniations and spinal fractures were detected in both DLR and conventional images. Both readers favored accelerated imaging in the majority of examinations. The lumbar spine examination time was cut by 61% in accelerated imaging compared to standard imaging.ConclusionIn conclusion, the utilization of deep learning-based image reconstruction techniques in lumbar spinal imaging resulted in significant time savings of up to 61% compared to standard imaging, while also improving image quality and diagnostic confidence. These findings highlight the potential of these techniques to enhance efficiency and accuracy in clinical practice for patients with lower back pain.

The authors are using a perceptual difference model (Case-PDM) to quantitatively evaluate image quality of the thousands of test images which can be created when optimizing fast magnetic resonance (MR) imaging strategies and reconstruction techniques. In this validation study, they compared human evaluation of MR images from multiple organs and from multiple image reconstruction algorithms to Case-PDM and similar models. The authors found that Case-PDM compared very favorably to human observers in double-stimulus continuous-quality scale and functional measurement theory studies over a large range of image quality. The Case-PDM threshold for nonperceptible differences in a 2-alternative forced choice study varied with the type of image under study, but was approximately 1.1 for diffuse image effects, providing a rule of thumb. Ordering the image quality evaluation models, we found in overall Case-PDM approximately IDM (Sarnoff Corporation) approximately SSIM [Wang et al. IEEE Trans. Image Process. 13, 600-612 (2004)] > mean squared error NR [Wang et al. (2004) (unpublished)] > DCTune (NASA) > IQM (MITRE Corporation). The authors conclude that Case-PDM is very useful in MR image evaluation but that one should probably restrict studies to similar images and similar processing, normally not a limitation in image reconstruction studies.

Image reconstruction in electrical impedance tomography using the sensitivity theorem is generally based on the assumption that the initial conductivity distribution of the body being imaged is uniform. The technique of image reconstruction using this method is described and reconstructed images are presented. Improvements in image quality and accuracy are demonstrated when accurate a priori `anatomical' information, in the form of a model of the distribution of conductivity within the region to be imaged, is used to construct the sensitivity matrix. In practice correct a priori information is not available, for example the conductivity values within the various anatomical regions will not be known. An iterative algorithm is presented which allows the conductivity parameters of the a priori model to be determined during reconstruction. Keywords: electrical impedance tomography, image reconstruction, a priori information

Ultrasound (US) imaging application in medicine and other fields is enormous. It has several advantages over other medical imaging modalities. The use of US in diagnosis is well established because of its noninvasive nature, portable, accurate, low-cost imaging modality, capable of forming real-time imaging and continuing improvement in image quality. Medical images and satellite images are usually degraded by noise during image acquisition and transmission process. It is estimated that 1 out of every form medical diagnostic image studies in the world involves ultrasonic techniques. The objectives of this article are to give an overview of the types of speckle reduction techniques in US imaging, present a new technique of speckle reduction, and carry out a comparative evaluation of despeckle filtering based on image quality metrics. A new speckle suppression method and coherence enhancement of medical US image are proposed: 9 despeckled filtering techniques and 5 edge detection operators, with the result evaluated by image quality metrics. The best edge detection hybrid with best filter, quality evaluation metrics has been found that the proposed method performance better than all other methods, while the structural details and result preserving of small and important image feature that contain diagnostic information in a better way than other despeckling filter.

Impact of noise-optimized virtual monoenergetic dual-energy computed tomography on image quality in patients with renal cell carcinoma

Unconventional computer aided imaging approaches, such as bi-spectrum or multi-frame blind deconvolution, have been used for some time to obtain high resolution images of space based objects from the ground. This looking up imaging scenario is characterized by r0 diameters usually much smaller than the telescope aperture, and isoplanatic angles on the same order as the object under observation. Air to ground imaging on the other hand, is often characterized by relatively large r0 values usually about the same size of the imaging aperture, but very small isoplanatic angles. In such a case, where the scene covers many isoplanatic patches, it has been shown that improvements in image quality can be obtained by dividing the field of view into sub-regions and applying processing algorithms to each sub-region. Because r0 is roughly the diameter of the aperture, the primary aberration over a subregion is tip-tilt. Block matching algorithms that correct only the local tip-tilt over the sub-regions have been developed to run in near real time. Correcting higher order local aberrations, using a blind deconvolution algorithm, can lead to improved results but requires increased computation. In this presentation we seek to compare air-to-ground imagery with different levels of processing; first global tip-tilt only correction; second local tip-tilt correction; and finally local tip-tilt plus high order correction. The goal is to determine if the increased detail obtained at each step is worth in increased processing complexity.

Computational acceleration on graphics processing units(GPUs) can make advanced magnetic resonance imaging(MRI) reconstruction algorithms attractive in clinical settings, thereby improving the quality of MR images across a broad spectrum of applications. At present, MR imaging is often limited by high noise levels, significant imaging artifacts, and/or long data acquisition (scan) times. Advanced image reconstruction algorithms can mitigate these limitations and improve image quality by simultaneously operating on scan data acquired with arbitrary trajectories and incorporating additional information such as anatomical constraints. However, the improvements in image quality come at the expense of a considerable increase in computation.This paper describes the acceleration of an advanced reconstruction algorithm on NVIDIA's Quadro FX 5600. Optimizations such as register allocating the voxel data, tiling the scan data, and storing the scan data in the Quadro's constant memory dramatically reduce the reconstruction's required bandwidth to on-chip memory. The Quadro's special functional units provide substantial acceleration of the trigonometric computations in the algorithm's inner loops, and experimentally-tuned code transformations increase the reconstruction's performance by an additional 20%.The reconstruction of a 3D image with 128^3 voxels ultimately achieves 150 GFLOPS and requires less than twominutes on the Quadro, while reconstruction on a quad-core CPU is thirteen times slower. Furthermore, relative to the true image, the error exhibited by the advanced reconstruction is only 12%, while conventional reconstruction techniques incur error of 42%. In short, the acceleration afforded by the GPU greatly increases the appeal of the advanced reconstruction for clinical MRI applications.

To evaluate objective and subjective image quality of a noise-optimized virtual monoenergetic imaging (VMI+) reconstruction technique in dual-energy computed tomography (DECT) angiography prior to transcatheter aortic valve replacement (TAVR). Datasets of 47 patients (35 men; 64.1 ± 10.9years) who underwent DECT angiography of heart and vascular access prior to TAVR were reconstructed with standard linear blending (F_0.5), VMI+, and traditional monoenergetic (VMI) algorithms in 10-keV intervals from 40-100keV. Signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) of 564 arterial segments were evaluated. Subjective analysis was rated by three blinded observers using a Likert scale. Mean SNR and CNR were highest in 40keV VMI+ series (SNR, 27.8 ± 13.0; CNR, 26.3 ± 12.7), significantly (all p < 0.001) superior to all VMI series, which showed highest values at 70keV (SNR, 18.5 ± 7.6; CNR, 16.0 ± 7.4), as well as linearly-blended F_0.5 series (SNR, 16.8 ± 7.3; CNR, 13.6 ± 6.9). Highest subjective image quality scores were observed for 40, 50, and 60keV VMI+ reconstructions (all p > 0.05), significantly superior to all VMI and standard linearly-blended images (all p < 0.01). Low-keV VMI+ reconstructions significantly increase CNR and SNR compared to VMI and standard linear-blending image reconstruction and improve subjective image quality in preprocedural DECT angiography in the context of TAVR planning. • VMI+ combines increased contrast with reduced image noise. • VMI+ shows substantially less image noise than traditional VMI. • 40-keV reconstructions show highest SNR/CNR of the aortic and iliofemoral access route. • Observers overall prefer 60keV VMI+ images. • VMI+ DECT imaging helps improve image quality for TAVR planning.

Deep learning (DL) reconstruction has been successful in realizing otherwise impracticable acceleration factors and improving image quality in conventional MRI field strengths; however, there has been limited application to ultra-high-field MRI. The objective of this study was to evaluate the performance of a prototype DL-based image reconstruction technique in 7T MRI of the brain utilizing magnetization-prepared 2 rapid acquisition gradient echoes (MP2RAGE) and sampling perfection with application-optimized contrasts using different flip angle evolutions (SPACE) acquisitions, in comparison with reconstructions in conventional compressed sensing and controlled aliasing in parallel imaging techniques. This retrospective study involved 60 patients who underwent 7T brain MRI between June 2024 and October 2024, comprising 30 patients with MP2RAGE data and 30 patients with SPACE FLAIR data. Each set of raw data was reconstructed with DL-based reconstruction and conventional reconstruction. Image quality was independently assessed by 2 neuroradiologists by using a 5-point Likert scale, which included overall image quality, artifacts, sharpness, structural conspicuity, and noise level. Interobserver agreement was determined by using top-box analysis. Contrast-to-noise ratio (CNR) and noise levels were quantitatively evaluated and compared by using the Wilcoxon signed-rank test. DL-based reconstruction resulted in a significant increase in overall image quality and a reduction in subjective noise level for both MP2RAGE and SPACE FLAIR data (all P < .001), with no significant differences in image artifacts (all P > .05). When compared with standard reconstruction, the implementation of DL-based reconstruction yielded an increase in CNR of 49.5% (95% CI, 33.0%-59.0%) for MP2RAGE data and 90.6% (95% CI, 73.2%-117.7%) for SPACE FLAIR data, along with a decrease in noise of 33.5% (95% CI, 23.0%-38.0%) for MP2RAGE data and 47.5% (95% CI, 41.9%-52.6%) for SPACE FLAIR data. DL-based reconstruction of 7T MRI significantly enhanced image quality compared with conventional reconstruction without introducing image artifacts. The achievable high acceleration factors have the potential to substantially improve image quality and resolution in 7T MRI.

Introduction CT coronary angiography (CTCA) is widely applied in coronary artery disease. It is therefore important that imaging protocols are optimised to obtain diagnostic images at the lowest possible radiation dose. We aimed to establish the feasibility of single heart-beat imaging in 320-multidetector CTCA, and assess variables that affect image quality. Methods Consecutive patients (n=249, 38% male) underwent contrast enhanced prospective electrocardiogram-gated single heart-beat CTCA. We assessed images before and after the introduction of sublingual glyceryl trinitrate to the CTCA protocol. Tube current and voltage were selected based on body-mass index (BMI) and acquisition window was selected based on heart rate (70–80% or 30–80% if the heart rate was below or above 65 beats/min respectively). Images were assessed by two trained observers and image quality was graded on a 4-point scale (1, excellent; 4, poor). Obstructive disease was defined as a stenosis of greater than 70% diameter. Results The mean heart rate was 60 beats/min (95% CI 59 to 62), BMI 29 kg/m2 (28, 30), age 58 years (56, 59) and dose-length product 283 mGy cm (266, 301). During scanning, 133(51%) received sublingual glyceryl trinitrate (GTN), 9(4%) had ectopic beats, and 12(5%) had atrial fibrillation. Diagnostic image quality was obtained in 99% with mean image quality 1.4 (1.3, 1.5). For patients with a BMI of 30 or above (n=111) diagnostic image quality was obtained in 109 scans (98%) and excellent image quality was obtained in 78 scans (70%). For patients with a heart rate above 65 bpm (n=58) diagnostic image quality was obtained in 98% (57 scans) and excellent image quality was obtained in 36% (21 scans). The mean dose-length product was 283 mGy cm (266, 301). An increase in heart rate led to an increase in radiation dose to 377 mGy cm (286, 268) for patients with a heart rate of greater than 80 beats/min (p&lt;0.001). A higher radiation dose was also associated with a BMI above 30 (DLP 231 (213, 248) vs 353 (330, 377), p&lt;0.001) and the presence of arrhythmia (DLP 275 (259, 290) vs 384 (301, 467), p 0.013). Age, sex, atrial fibrillation, ectopics, diabetes mellitus (12%) and the presence of obstructive coronary disease were not related to image quality. In multivariate regression analysis a lower heart rate and GTN were associated with improved image quality (β 0.439, p&lt;0.001 and β −0.199, p=0.001 respectively). Figure 1 The effect of heart rate, BMI and arrhythmia on the proportion of CTCA with diagnostic image quality. Conclusions Optimal image quality in single heart-beat 320-multidetector CTCA is achievable at low radiation doses in 99% of unselected patients. Image quality is further improved by lower heart rate and sublingual GTN. This study has established the use of half-segment reconstruction in all patients undergoing wide volume CTCA irrespective of heart rate, ectopic beats and the presence of atrial fibrillation. A widened window of acquisition can be used to improve image quality in patients with a heart rate greater than 65 beats per minute. In addition we have established that the use of GTN improves image quality in 320 multidetector CTCA.

We report a multiframe blind deconvolution algorithm that we have developed for imaging through the atmosphere. The algorithm has been parallelized to a significant degree for execution on high-performance computers, with an emphasis on distributed-memory systems so that it can be hosted on commodity clusters. As a result, image restorations can be obtained in seconds to minutes. We have compared and quantified the quality of its image restorations relative to the associated Cramér-Rao lower bounds (when they can be calculated). We describe the algorithm and its parallelization in detail, demonstrate the scalability of its parallelization across distributed-memory computer nodes, discuss the results of comparing sample variances of its output to the associated Cramér-Rao lower bounds, and present image restorations obtained by using data collected with ground-based telescopes.

Cardiovascular magnetic resonance (CMR) imaging is a powerful tool for assessing cardiac anatomy and function but remains limited by average image quality due to artefacts and long acquisition times, and complex and often too long breath-holds. Deep learning methods have recently been applied and show potential to shorten scan times by 70-80% while improving image quality, enhancing clinical efficiency. The aim of this study is to summarise the different AI-enabled methods for improving CMR image quality, including scanning time, as a key determinant for artefact reduction. A scoping review was conducted according to PRISMA guidelines. The articles were screened and reviewed by two researchers. A qualitative thematic synthesis was conducted and a CASP-mediated risk of bias assessment was performed. The eligible articles were thirty-one. These articles were thematically categorised in four subgroups, based on emerging themes: scan acceleration, artefact detection, artefact reduction, image reconstruction. A table with significant results for each theme has been presented and results were discussed qualitatively. AI demonstrated consistent improvements across the four subgroups. For scan acceleration, deep learning achieved approximately a 70-80% reduction in scan duration maintaining or even improving image quality. For artefact detection, convolutional neural networks achieved on average a 90% accuracy in detecting artefacts, across multiple metrics, indicating reliable artefact identification and strong agreement with human experts. AI models effectively reduce artefacts and enhance image quality, achieving consistently better reconstruction accuracy, sharper edges, and faster processing compared to conventional methods. Finally, for image reconstruction, generative adversarial networks enhanced structural similarity by approximately 56% (SSIM 0.591 → 0.925). Together, these results illustrate the potential of AI to optimise CMR image quality. AI can be an effective tool in addressing many of the CMR imaging challenges and thus improving image quality.

Summary This case study summarizes the data reprocessing effort that has been applied to seismic data located in the Algerian on-shore over a structurally complex, low porosity, low permeability reservoir and the improvements in final image quality that have been achieved. Subsequent to processing, the seismic data were inverted to rock properties. The final results show seismic data with significant improvements in image quality and structural definition, whilst the inversion to rock properties has enabled lithology and porosity characterization with increased confidence. Furthermore, borehole seismic surveys were acquired and integrated in the 3D volume to design the trajectory of horizontal and side-track wells.