Reduced Scan Time Research Articles

106 The effective study: development and application of a question-driven, time-effective cardiac magnetic resonance scanning protocol

Abstract Aims Long scanning times impede cardiac magnetic resonance (CMR) clinical uptake. A ‘one-size-fits-all’ shortened, focused protocol [e.g. only function and late-gadolinium enhancement (LGE)] reduces scanning time and costs, but provide less information. We developed two question-driven CMR and stress-CMR protocols, including tailored advanced tissue characterization, and tested their effectiveness in reducing scanning time while retaining the diagnostic performances of standard protocols. Methods and results Eighty-three consecutive patients with cardiomyopathy or ischaemic heart disease underwent the tailored CMR. Each scan consisted of standard cines, LGE imaging, native T1-mapping, and extracellular volume. Fat/oedema modules, right ventricle cine, and in-line quantitative perfusion mapping were performed as clinically required. Workflow was optimized to avoid gaps. See Figure 1 for protocol details. Time target was <30 min for a CMR and <35 min for a stress-CMR. CMR was considered impactful when its results drove changes in diagnosis or management. Advanced tissue characterization was considered impactful when it changed the confidence level in the diagnosis. Images’ quality was assessed. A ‘control group’ of 137 patients was identified among scans performed before February 2020. Compared to standard protocols, the average scan duration dropped by > 30% (CMR: from 42 ± 8 to 28 ± 6min; stress-CMR: from 50 ± 10 to 34 ± 6min, both P < 0.0001). Independent on the protocol, CMR was impactful in ∼60% cases, and advanced tissue characterization was impactful in > 45% of cases. Quality grading was similar between the two protocols. Tailored protocols did not require additional staff. Conclusions Tailored CMR and stress-CMR protocols including advanced tissue characterization are accurate and time-effective for cardiomyopathies and ischaemic heart disease.

Abdominal computed tomography localizer image generation: A deep learning approach

Background and ObjectiveComputed Tomography (CT) has become an important clinical imaging modality, as well as the leading source of radiation dose from medical imaging procedures. Modern CT exams are usually led by two quick orthogonal localization scans, which are used for patient positioning and diagnostic scan parameter definition. These two localization scans contribute to the patient dose but are not used for diagnosis purposes. In this study, we investigate the possibility of using deep learning models to reconstruct one localization scan image from the other, thus reducing the patient dose and simplifying the clinical workflow. MethodsWe propose a modified encoder-decoder network and a scaled mixture loss function specifically for the focal task. In this study, 12,487 clinical abdominal exams were retrieved from a clinical medical imaging storage system and randomly split for training, validation, and test in the ratio of 7:1:2. Reconstructed images were compared with the ground truth in terms of location prediction error, profile prediction error, and attenuation prediction error. ResultsThe average location error, profile error, and attenuation error were 1.02±3.37 mm, 4.43±2.02%, and 6.2 ± 2.94% for lateral prediction, and 6.46±6.43 mm, 3.9 ± 2.32%, and 7.12±3.54% for AP prediction, respectively. ConclusionsWe conclude that although the reconstructed abdominal CT localization images may lack some details on the internal organ structures, they could be used effectively for tube current modulation calculation and patient positioning purposes, leading to a reduction of radiation dose and scan time in clinical CT exams.

High-Resolution DWI with Simultaneous Multi-Slice Readout-Segmented Echo Planar Imaging for the Evaluation of Malignant and Benign Breast Lesions.

To investigate the feasibility and effectiveness of high-resolution readout-segmented echo planar imaging (rs-EPI), diffusion-weighted imaging (DWI) is used simultaneously with multi-slice (SMS) imaging (SMS rs-EPI) for the differentiation of breast malignant and benign lesions in comparison to conventional rs-EPI on a 3T MR scanner. A total of 102 patients with 113 breast lesions underwent bilateral breast MRI using a prototype SMS rs-EPI sequence and a conventional rs-EPI sequence. Subjective image quality was assessed using a 5-point Likert scale (1 = poor, 5 = excellent). Signal-to-noise ratio (SNR), lesion contrast-to-noise ratio (CNR) and apparent diffusion coefficients (ADC) value of the lesion were measured for comparison. Receiver operating characteristic curve analysis was performed to evaluate the diagnosis performance of ADC, and the corresponding area under curve (AUC) was calculated. The image quality scores in anatomic distortion, lesion conspicuity, sharpness of anatomical details and overall image quality of SMS rs-EPI were significantly higher than those of conventional rs-EPI. CNR was enhanced in the high-resolution SMS rs-EPI acquisition (6.48 ± 1.71 vs. 4.23 ± 1.49; p < 0.001). The mean ADC value was comparable in SMS rs-EPI and conventional rs-EPI (benign 1.45 × 10−3 vs. 1.43 × 10−3 mm2/s, p = 0.702; malignant 0.91 × 10−3 vs. 0.89 × 10−3 mm2/s, p = 0.076). The AUC was 0.957 in SMS rs-EPI and 0.983 in conventional rs-EPI. SMS rs-EPI technique allows for higher spatial resolution and slight reduction of scan time in comparison to conventional rs-EPI, which has potential for better differentiation between malignant and benign lesions of the breast.

Clinical feasibility of simultaneous multislice acceleration in knee MRI

PurposeTo find the best simultaneous multislice (SMS) accelerated setting for clinical application in knee MRI. Material and methodsThirty-three patients (mean age, 54 years; 21 women) who underwent knee MRI (conventional/SMS sequences) between June and October 2020 were enrolled. Two radiologists retrospectively evaluated sagittal T1- and T2-weighted conventional (2-fold parallel acquisition technique [PAT-2]) and SMS (SMS-2 [PAT-2 with 2-fold SMS], SMS-3, and SMS-4) images. For qualitative analysis, artifacts (zebra/residual aliasing) and diagnostic confidence for internal derangement of knee (bone marrow, cartilage, meniscus, anterior cruciate ligament, and synovium abnormalities) were evaluated. For quantitative analysis, contrast-to-noise ratios of bone marrow, meniscus, joint effusion, and ligament were evaluated. ResultsCompared to PAT-2 (2 min 32 s), mean acquisition time was reduced by 47% in SMS-2; 64%, SMS-3; and 70%, SMS-4. In qualitative analysis, zebra artifacts were only seen on T2-weighted SMS images. The more SMS was applied, the more zebra and residual aliasing artifacts were seen and the lower diagnostic confidence was for internal derangement. However, qualitative analysis showed acceptable image quality in SMS-2 and SMS-3 images, but not in SMS-4 images. In quantitative analysis, SMS-4 images showed the lowest contrast-to-noise ratios and there were no significant differences among PAT-2, SMS-2, and SMS-3 images. ConclusionApplying SMS-3 to knee MRI reduced scan time and showed acceptable image quality compared to conventional (PAT-2). However, when evaluating SMS images, radiologists should know that when more SMS is applied, more zebra and residual aliasing artifacts appear.

Anisotropic neural deblurring for MRI acceleration.

MRI has become the tool of choice for brain imaging, providing unrivalled contrast between soft tissues, as well as a wealth of information about anatomy, function, and neurochemistry. Image quality, in terms of spatial resolution and noise, is strongly dependent on acquisition duration. A typical brain MRI scan may last several minutes, with total protocol duration often exceeding 30 minutes. Long scan duration leads to poor patient experience, long waiting time for appointments, and high costs. Therefore, shortening MRI scans is crucial. In this paper, we investigate the enhancement of low-resolution (LR) brain MRI scanning, to enable shorter acquisition times without compromising the diagnostic value of the images. We propose a novel fully convolutional neural enhancement approach. It is optimized for accelerated LR MRI acquisitions obtained by reducing the acquisition matrix size only along phase encoding direction. The network is trained to transform the LR acquisitions into corresponding high-resolution (HR) counterparts in an end-to-end manner. In contrast to previous neural-based MRI enhancement algorithms, such as DAGAN, the LR images used for training are real acquisitions rather than smoothed, downsampled versions of the HR images. The proposed method is validated qualitatively and quantitatively for an acceleration factor of 4. Favourable comparison is demonstrated against the state-of-the-art DeblurGAN and DAGAN algorithms in terms of PSNR and SSIM scores. The result was further confirmed by an image quality rating experiment performed by four senior neuroradiologists. The proposed method may become a valuable tool for scan time reduction in brain MRI. In continuation of this research, the validation should be extended to larger datasets acquired for different imaging protocols, and considering several MRI machines produced by different vendors.

Impact of reduced injected dose on the quantification of [18F]RO948 and [18F]Flortaucipir PET for in vivo tau pathology

AbstractBackgroundPrevious research has demonstrated that the injected dose in PET examinations can be reduced without substantial effects on quantitative outcomes. These reductions can lead to lower radiation burden for subjects, reduced costs for institutions and potential for additional research scans of the same subject. Here, we investigated the effect of reduced injected doses of [18F]RO948 and [18F]Flortaucipir (FTP) on standardised uptake value ratios (SUVRs) and associated outcomes.MethodImages were manipulated from list‐mode data to simulate injected doses of 70%, 50%, 25% and 12.5% of the true injected dose. Initial analyses were scan time reduction for RO948 and dose reduction for FTP with future harmonisation of processing intended. Inferior cerebellum‐grey was used to compute SUVRs in regions‐of‐interest (ROIs) corresponding to the in vivo Braak regions of tau spread. Differences between true and reduced injected doses (given as a percentage) were calculated in individual regions as the mean of [SUVRreduced – SUVRtrue] over the means of SUVRreduced and SUVRtrue values across subjects.ResultPreviously reported test‐retest variability for RO948 and FTP are 2‐6% (±6%) and 1.5‐3.3% (±3%) respectively. Regionally, RO948 remained below this TRT variability at 25% injected dose, with an average SUVR difference of 2.3% (±8%) (figure 2), whereas FTP remained below TRT variability only at 70% injected dose with an average SUVR difference of 2.9% (±3%). However, differences were above TRT variability for FTP at lower injected doses. FTP average SUVR differences were observed to be higher in CN compared to AD patients at reduced injected doses but with greater variability in AD than CN. There were no differences observed between Ab+ and Ab‐ controls.ConclusionPreliminary findings suggest that lower injected tracer‐doses for RO948 and FTP yield robust SUVRs. However, the degree to which injected doses can be reduced must be investigated further for different demographics and for different tracers. Ongoing work aims to harmonise the simulation list‐mode processing, add MCI/AD RO948 subjects, and evaluate dose reduction in [18F]GTP1 data. Further validation will be performed against demographic and clinical variables, as well as for longitudinal data.

Deep Learning Image Processing Enables 40% Faster Spinal MR Scans Which Match or Exceed Quality of Standard of Care

ObjectiveThis prospective multicenter multireader study evaluated the performance of 40% scan-time reduced spinal magnetic resonance imaging (MRI) reconstructed with deep learning (DL).MethodsA total of 61 patients underwent standard of care (SOC) and accelerated (FAST) spine MRI. DL was used to enhance the accelerated set (FAST-DL). Three neuroradiologists were presented with paired side-by-side datasets (666 series). Datasets were blinded and randomized in sequence and left-right display order. Image features were preference rated. Structural similarity index (SSIM) and per pixel L1 was assessed for the image sets pre and post DL-enhancement as a quantitative assessment of image integrity impact.ResultsFAST-DL was qualitatively better than SOC for perceived signal-to-noise ratio (SNR) and artifacts and equivalent for other features. Quantitative SSIM was high, supporting the absence of image corruption by DL processing.ConclusionDL enables 40% spine MRI scan time reduction while maintaining diagnostic integrity and image quality with perceived benefits in SNR and artifact reduction, suggesting potential for clinical practice utility.

Adjustable shrinkage-thresholding projection algorithm for compressed sensing magnetic resonance imaging

Compressed sensing (CS) aims to reconstruct a high quality images with as little sample data as possible. Magnetic resonance imaging (MRI) plays an important role in medical imaging tools but has a slower data acquisition process. Applying CS to MRI offers significant scan time reductions. In this paper, we proposed a fast and efficient algorithm for compressed sensing magnetic resonance imaging (CS-MRI) reconstruction, denoted as adjustable shrinkage-thresholding projection algorithm (ASTP). It is designed to use adjustable shrinkage rules for lp-norm based CS-MRI model. This algorithm is established by using an iterative projection and acceleration scheme. In each iteration, the proposed adjustable shrinkage-thresholding rules are employed to ensure global convergence to accurate solution. Furthermore, the parameter p can be selected flexibly according to different practical application situations, and the orthogonal projection operation is used to reduce the dimension of the solution space to accelerate the convergence speed and improve the reconstruction quality. Numerical experiments show that proposed ASTP algorithm provides a higher accuracy, convergence speed and ability to suppress noise compared with some certain state-of-the-art algorithms.

Deep Learning Enables 60% Accelerated Volumetric Brain MRI While Preserving Quantitative Performance: A Prospective, Multicenter, Multireader Trial.

In this prospective, multicenter, multireader study, we evaluated the impact on both image quality and quantitative image-analysis consistency of 60% accelerated volumetric MR imaging sequences processed with a commercially available, vendor-agnostic, DICOM-based, deep learning tool (SubtleMR) compared with that of standard of care. Forty subjects underwent brain MR imaging examinations on 6 scanners from 5 institutions. Standard of care and accelerated datasets were acquired for each subject, and the accelerated scans were enhanced with deep learning processing. Standard of care, accelerated scans, and accelerated-deep learning were subjected to NeuroQuant quantitative analysis and classified by a neuroradiologist into clinical disease categories. Concordance of standard of care and accelerated-deep learning biomarker measurements were assessed. Randomized, side-by-side, multiplanar datasets (360 series) were presented blinded to 2 neuroradiologists and rated for apparent SNR, image sharpness, artifacts, anatomic/lesion conspicuity, image contrast, and gray-white differentiation to evaluate image quality. Accelerated-deep learning was statistically superior to standard of care for perceived quality across imaging features despite a 60% sequence scan-time reduction. Both accelerated-deep learning and standard of care were superior to accelerated scans for all features. There was no difference in quantitative volumetric biomarkers or clinical classification for standard of care and accelerated-deep learning datasets. Deep learning reconstruction allows 60% sequence scan-time reduction while maintaining high volumetric quantification accuracy, consistent clinical classification, and what radiologists perceive as superior image quality compared with standard of care. This trial supports the reliability, efficiency, and utility of deep learning-based enhancement for quantitative imaging. Shorter scan times may heighten the use of volumetric quantitative MR imaging in routine clinical settings.

Spiral gradient echo versus cartesian turbo spin echo imaging for sagittal contrast-enhanced fat-suppressed T1 weighted spine MRI: an inter-individual comparison study.

To compare a novel 3D spiral gradient echo (GRE) sequence with a conventional 2D cartesian turbo spin echo (TSE) sequence for sagittal contrast-enhanced (CE) fat-suppressed (FS) T1 weighted (T1W) spine MRI. In this inter-individual comparison study, 128 patients prospectively underwent sagittal CE FS T1W spine MRI with either a 2D cartesian TSE ("TSE", 285 s, 64 patients) or a 3D spiral GRE sequence ("Spiral", 93 s, 64 patients). Between both groups, patients were matched in terms of anatomical region (cervical/thoracic/lumbar spine and sacrum). Three readers used 4-point Likert scales to assess images qualitatively in terms of overall image quality, presence of artifacts, spinal cord visualization, lesion conspicuity and quality of fat suppression. Spiral achieved a 67.4% scan time reduction compared to TSE. Interreader agreement was high (alpha=0.868-1). Overall image quality (4;[3,4] vs 3;[3,4], p<0.001 - p=0.002 for all readers), presence of artifacts (4;[3,4] vs 3;[3,4] p=0.027 - p=0.046 for all readers), spinal cord visualization (4;[4,4] vs 4;[3,4], p<0.001 for all readers), lesion conspicuity (4;[4,4] vs 4;[4,4], p=0.016 for all readers) and quality of fat suppression (4;[4,4] vs 4;[4,4], p=0.027 - p=0.033 for all readers), were all deemed significantly improved by all three readers on Spiral images as compared to TSE images. We demonstrate the feasibility of a novel 3D spiral GRE sequence for improved and rapid sagittal CE FS T1W spine MRI. A 3D spiral GRE sequence allows for improved sagittal CE FS T1W spine MRI at very short scan times.

Dual-energy CT imaging over non-overlapping, orthogonal arcs of limited-angular ranges.

Interest exists in dual-energy computed tomography (DECT) imaging with scanning arcs of limited-angular ranges (LARs) for reducing scan time and radiation dose, and for enabling scan configurations of C-arm CT that can avoid possible collision between the rotating X-ray tube/detector and the imaged subject. In this work, we investigate image reconstruction for a type of configurations of practical DECT interest, referred to as the two-orthogonal-arc configuration, in which low- and high-kVp data are collected over two non-overlapping arcs of equal LAR α, ranging from 30° to 90°, separated by 90°. The configuration can readily be implemented, e.g., on CT with dual sources separated by 90° or with the slow-kVp-switching technique. The directional-total-variation (DTV) algorithm developed previously for image reconstruction in conventional, single-energy CT is tailored to enable image reconstruction in DECT with two-orthogonal-arc configurations. Performing visual inspection and quantitative analysis of monochromatic images obtained and effective atomic numbers estimated, we observe that the monochromatic images of the DTV algorithm from LAR data are with substantially reduced LAR artifacts, which are observed otherwise in those of existing algorithms, and thus visually correlate reasonably well, in terms of metrics PCC and nMI, with their reference images obtained from full-angular-range data. In addition, effective atomic numbers estimated from LAR data of DECT with two-orthogonal-arc configurations are in reasonable agreement, with relative errors up to ∼ 10%, with those estimated from full-angular-range data in DECT. The results acquired in the work may yield insights into the design of LAR configurations of practical dual-energy application relevance in diagnostic CT or C-arm CT imaging.

Construction of super-rapid brain MRA using oblique transverse acquisition phase contrast angiography with tilted optimized non-saturated excitation pulse

[Background] Magnetic resonance angiography (MRA) is one of the most important sequences to estimate a cerebrovascular disease. We often encounter poor image quality due to slow arterial flow related to aging and motion artifact caused by disturbance of consciousness. We focused on phase contrast angiography (PCA) to overcome these difficulties. PCA can reduce scan time drastically by combining transverse acquisition and partial slab setting covering entire brain arteries. However, transverse acquisition in PCA has a large difference in signal intensity between proximal and distal vessels. Therefore, we apply tilted optimized non-saturated excitation (TONE) to improve image quality. [Purpose] The purpose of this study to investigate the usefulness of TONE for PCA. [Method] We estimated the efficacy of TONE in transverse acquisition PCA using measurement of signal intensity in arteries. We compared image quality among 1 min PCA with/without TONE and time-of flight (TOF)-MRA, by visual. [Result] TONE improved the signal inhomogeneity in entire brain arteries. PCA with TONE (5°–9°) demonstrated the highest image quality. [Conclusion] Oblique transverse acquisition PCA with TONE provides superior image quality compared with TOF with similar scan time. TONE improved image quality by the homogenizing signal intensity of vessels from proximal to distal in oblique transvers acquisition PCA. Our MRA can be performed in about 1 min and provides sufficient quality to estimate brain vessels.

Enhanced hyperspectral tomography for bioimaging by spatiospectral reconstruction

Here we apply hyperspectral bright field imaging to collect computed tomographic images with excellent energy resolution (~ 1 keV), applying it for the first time to map the distribution of stain in a fixed biological sample through its characteristic K-edge. Conventionally, because the photons detected at each pixel are distributed across as many as 200 energy channels, energy-selective images are characterised by low count-rates and poor signal-to-noise ratio. This means high X-ray exposures, long scan times and high doses are required to image unique spectral markers. Here, we achieve high quality energy-dispersive tomograms from low dose, noisy datasets using a dedicated iterative reconstruction algorithm. This exploits the spatial smoothness and inter-channel structural correlation in the spectral domain using two carefully chosen regularisation terms. For a multi-phase phantom, a reduction in scan time of 36 times is demonstrated. Spectral analysis methods including K-edge subtraction and absorption step-size fitting are evaluated for an ex vivo, single (iodine)-stained biological sample, where low chemical concentration and inhomogeneous distribution can affect soft tissue segmentation and visualisation. The reconstruction algorithms are available through the open-source Core Imaging Library. Taken together, these tools offer new capabilities for visualisation and elemental mapping, with promising applications for multiply-stained biological specimens.

Spiral 2D T2-Weighted TSE Brain MR Imaging: Initial Clinical Experience.

Spiral MR imaging may enable improved image quality and higher scan speeds than Cartesian trajectories. We sought to compare a novel spiral 2D T2-weighted TSE sequence with a conventional Cartesian and an artifact-robust, non-Cartesian sequence named MultiVane for routine clinical brain MR imaging. Thirty-one patients were scanned with all 3 sequences (Cartesian, 4 minutes 14 seconds; MultiVane, 2 minutes 49 seconds; spiral, 2 minutes 12 seconds) on a standard clinical 1.5T MR scanner. Three readers described the presence and location of abnormalities and lesions and graded images qualitatively in terms of overall image quality, the presence of motion and pulsation artifacts, gray-white matter differentiation, lesion conspicuity, and subjective preference. Image quality was objectivized by measuring the SNR and the coefficients of variation for CSF, GM, and WM. Spiral achieved a scan time reduction of 51.9% and 21.9% compared with Cartesian and MultiVane, respectively. The number and location of lesions were identical among all sequences. As for the qualitative analysis, interreader agreement was high (Krippendorff α > .75). Spiral and MultiVane both outperformed the Cartesian sequence in terms of overall image quality, the presence of motion artifacts, and subjective preference (P < .001). In terms of the presence of pulsation artifacts, gray-white matter differentiation, and lesion conspicuity, all 3 sequences performed similarly well (P > .15). Spiral and MultiVane outperformed the Cartesian sequence in coefficient of variation WM and SNR (P < .01). Spiral 2D T2WI TSE is feasible for routine structural brain MR imaging and offers high-quality, artifact-robust brain imaging in short scan times.

Quantitative accuracy in total-body imaging using the uEXPLORER PET/CT scanner

Absolute quantification of regional tissue concentration of radioactivity in positron emission tomography (PET) is a critical parameter-of-interest across various clinical and research applications and is affected by a complex interplay of factors including scanner calibration, data corrections, and image reconstruction. The emergence of long axial field-of-view (FOV) PET systems widens the dynamic range accessible to PET and creates new opportunities in reducing scan time and radiation dose, delayed or low radioactivity imaging, as well as kinetic modeling of the entire human. However, these imaging regimes impose challenging conditions for accurate quantification due to constraints from image reconstruction, low count conditions, as well as large and rapidly changing radioactivity distribution across a large axial FOV. We comprehensively evaluated the quantitative accuracy of the uEXPLORER total-body scanner in conditions that encompass existing and potential imaging applications (such as dynamic imaging and ultralow-dose imaging) using a set of total-body specific phantom and human measurements. Through these evaluations we demonstrated a relative count rate accuracy of ±3%–4% using the NEMA NU 2-2018 protocol, an axial uniformity spread of ±3% across the central 90% axial FOV, and a 3% activity bias spread from 17 to 474 MBq 18F-FDG in a 210 cm long cylindrical phantom. Region-of-interest quantification spread of 1% was found by simultaneously scanning three NEMA NU 2 image quality phantoms, as well as relatively stable volume-of-interest quantification across 0.2%–100% of total counts through re-sampled datasets. In addition, an activity bias spread of −2% to +1% post-bolus injections in human subjects was found. Larger bias changes during the bolus injection phase in humans indicated the difficulty in providing accurate PET data corrections for complex activity distributions across a large dynamic range. Our results overall indicated that the quantitative performance achieved with the uEXPLORER scanner was uniform across the axial FOV and provided the accuracy necessary to support a wide range of imaging applications.

Feasibility evaluation of PET scan-time reduction for diagnosing amyloid-β levels in Alzheimer's disease patients using a deep-learning-based denoising algorithm

PurposeTo shorten positron emission tomography (PET) scanning time in diagnosing amyloid-β levels thus increasing the workflow in centers involving Alzheimer's Disease (AD) patients. MethodsPET datasets were collected for 25 patients injected with 18F-AV45 radiopharmaceutical. To generate necessary training data, PET images from both normal-scanning-time (20-min) as well as so-called “shortened-scanning-time” (1-min, 2-min, 5-min, and 10-min) were reconstructed for each patient. Building on our earlier work on MCDNet (Monte Carlo Denoising Net) and a new Wasserstein-GAN algorithm, we developed a new denoising model called MCDNet-2 to predict normal-scanning-time PET images from a series of shortened-scanning-time PET images. The quality of the predicted PET images was quantitatively evaluated using objective metrics including normalized-root-mean-square-error (NRMSE), structural similarity (SSIM), and peak signal-to-noise ratio (PSNR). Furthermore, two radiologists performed subjective evaluations including the qualitative evaluation and a five-point grading evaluation. The denoising performance of the proposed MCDNet-2 was finally compared with those of U-Net, MCDNet, and a traditional denoising method called Gaussian Filtering. ResultsThe proposed MCDNet-2 can yield good denoising performance in 5-min PET images. In the comparison of denoising methods, MCDNet-2 yielded the best performance in the subjective evaluation although it is comparable with MCDNet in objective comparison (NRMSE, PSNR, and SSIM). In the qualitative evaluation of amyloid-β positive or negative results, MCDNet-2 was found to achieve a classification accuracy of 100%. ConclusionsThe proposed denoising method has been found to reduce the PET scan time from the normal level of 20 min to 5 min but still maintaining acceptable image quality in correctly diagnosing amyloid-β levels. These results suggest strongly that deep learning-based methods such as ours can be an attractive solution to the clinical needs to improve PET imaging workflow.

Detection of Mandibular Canal in Human Dry Mandibles with Cone Beam Computed Tomography using 270° and 360° Protocols under Continuous and Pulse Modes – A Comparative Study

Background: Cone beam computed tomography (CBCT) enables practitioner to accurately localize and mark the nerve canal. The rotation of the X-ray tube and the detector around the patient's head produces multiple projection images. A high number of projections is associated with increased radiation dose to the patient, higher spatial and contrast resolution. Shorter scan arc results in reduced scan time, reduced dose and smaller file sizes. The pulsed X-ray beam reduces the actual exposure time than the scanning time and the patient radiation exposure. Aim: (1) To assess and compare the diagnostic efficacy between pulse mode and continuous mode (2) To assess and compare the diagnostic efficacy between 270° and 360° CBCT rotational protocols Subjects and Methods: A total of 30 Intact Dentate or Partially dentate dry human mandibles consisting first, second, and third molars were subjected to a total of 4 CBCT scans: 270° rotation scans under continuous and pulse mode, 360° rotation scans under continuous and pulse mode using two CBCT units. Distance from the inferior cortical border of mandible (ICBM) to the roof of the inferior alveolar nerve canal (IANC) was measured at four locations: mental foramen, first, second and third molar. The results were then compared. Results: Comparison of the IANC length measurements for the two modes revealed no significant difference. Conclusion: Shorter scan arc and pulse mode can be adopted in patients finding it difficult to remain stable for prolonged periods of time, while maintaining appreciable clarity and quality of image. This could apply to geriatric patients, children, mentally challenged, in patients with trauma, neurological diseases, anxiety, claustrophobia, wherein the patient's movements might need re-scanning, which would lead to an extra radiation dose to the patient.

Fast data-driven learning of parallel MRI sampling patterns for large scale problems

In this study, a fast data-driven optimization approach, named bias-accelerated subset selection (BASS), is proposed for learning efficacious sampling patterns (SPs) with the purpose of reducing scan time in large-dimensional parallel MRI. BASS is applicable when Cartesian fully-sampled k-space measurements of specific anatomy are available for training and the reconstruction method for undersampled measurements is specified; such information is used to define the efficacy of any SP for recovering the values at the non-sampled k-space points. BASS produces a sequence of SPs with the aim of finding one of a specified size with (near) optimal efficacy. BASS was tested with five reconstruction methods for parallel MRI based on low-rankness and sparsity that allow a free choice of the SP. Three datasets were used for testing, two of high-resolution brain images (text {T}_{2}-weighted images and, respectively, text {T}_{1rho }-weighted images) and another of knee images for quantitative mapping of the cartilage. The proposed approach has low computational cost and fast convergence; in the tested cases it obtained SPs up to 50 times faster than the currently best greedy approach. Reconstruction quality increased by up to 45% over that provided by variable density and Poisson disk SPs, for the same scan time. Optionally, the scan time can be nearly halved without loss of reconstruction quality. Quantitative MRI and prospective accelerated MRI results show improvements. Compared with greedy approaches, BASS rapidly learns effective SPs for various reconstruction methods, using larger SPs and larger datasets; enabling better selection of sampling-reconstruction pairs for specific MRI problems.

Reproducibility of Longitudinal Changes in Cortical Thickness Determined by Surface-Based Morphometry Between Non-Accelerated and Accelerated MR Imaging.

Scan acceleration such as parallel imaging reduces scan time, but shorter scan time may reduce the signal-to-noise ratio and affect image quality. The reproducibility of longitudinal changes in the brain structure between non-accelerated and accelerated imaging by surface-based analysis is unclear. To determine the reproducibility of longitudinal changes in cortical thickness, measured by surface-based morphometry, between non-accelerated and accelerated structural T1 -weighted imaging in the healthy elderly and those with mild cognitive impairment (MCI) and Alzheimer's disease (AD). Retrospective. Fifty healthy elderly subjects (age=73 ± 5 years, 29 females, 21 males), 54 MCI patients (age=71 ± 7 years, 23 females, 31 males), and 8 AD patients (age=78 ± 6 years, 6 females, 2 males). 3 T, magnetization-prepared rapid gradient-echo. Longitudinal changes in cortical thickness estimated by the longitudinal stream in FreeSurfer from 2-year interval data, and visual assessment of image quality by three radiologists. Intraclass correlation coefficient (ICC) and Kruskal-Wallis test. A P value <0.05 was considered significant. Healthy elderly subjects, MCI patients, and AD patients showed different patterns in the ICC maps. For the smoothing of 20 mm full width at half maximum, the mean ICC was 0.45 overall (healthy elderly, 0.33; MCI patients, 0.49; AD patients, 0.31). The within-subject SDs of the symmetrized percent changes were similar between healthy elderly subjects (mean, 1.3%/year) and MCI patients (mean, 1.3%/year) but larger in AD patients (mean, 1.7%/year). Image quality did not significantly differ per group (P=0.18). The results of this study indicate the reproducibility of longitudinal changes in cortical thickness measured by surface-based morphometry between non-accelerated and accelerated imaging, and that the reproducibility varies by disease and region. 3 TECHNICAL EFFICACY: Stage 1.

Image Quality and Diagnostic Performance of Accelerated Shoulder MRI With Deep Learning-Based Reconstruction.

BACKGROUND. Shoulder MRI using standard multiplanar sequences requires long scan times. Accelerated sequences have tradeoffs in noise and resolution. Deep learning-based reconstruction (DLR) may allow reduced scan time with preserved image quality. OBJECTIVE. The purpose of this study was to compare standard shoulder MRI sequences and accelerated sequences without and with DLR in terms of image quality and diagnostic performance. METHODS. This retrospective study included 105 patients (45 men, 60 women; mean age, 57.6 ± 10.9 [SD] years) who underwent a total of 110 3-T shoulder MRI examinations. Examinations included standard sequences (scan time, 9 minutes 23 seconds) and accelerated sequences (3 minutes 5 seconds; 67% reduction), both including fast spin-echo sequences in three planes. Standard sequences were reconstructed using the conventional pipeline; accelerated sequences were reconstructed using both the conventional pipeline and a commercially available DLR pipeline. Two radiologists independently assessed three image sets (standard sequence, accelerated sequence without DLR, and accelerated sequence with DLR) for subjective image quality and artifacts using 4-point scales (4 = highest quality) and identified pathologies of the subscapularis tendon, supraspinatus-infraspinatus tendon, long head of the biceps brachii tendon, and glenoid labrum. Interobserver agreement and agreement between image sets for the evaluated pathologies were assessed using weighted kappa statistics. In 27 patients who underwent arthroscopy, diagnostic performance was calculated using arthroscopic findings as a reference standard. RESULTS. Mean subjective image quality scores for readers 1 and 2 were 10.6 ± 1.2 and 10.5 ± 1.4 for the standard sequence, 8.1 ± 1.3 and 7.2 ± 1.1 for the accelerated sequence without DLR, and 10.7 ± 1.2 and 10.5 ± 1.6 for the accelerated sequence with DLR. Mean artifact scores for readers 1 and 2 were 9.3 ± 1.2 and 10.0 ± 1.0 for the standard sequence, 7.3 ± 1.3 and 9.1 ± 0.8 for the accelerated sequence without DLR, and 9.4 ± 1.2 and 9.8 ± 0.8 for the accelerated sequence with DLR. Interobserver agreement ranged from kappa of 0.813-0.951 except for accelerated sequence without DLR for the supraspinatus-infraspinatus tendon (κ = 0.673). Agreement between image sets ranged from kappa of 0.809-0.957 except for reader 1 for supraspinatus-infraspinatus tendon (κ = 0.663-0.700). Accuracy, sensitivity, and specificity for tears of the four structures were not different (p > .05) among image sets. CONCLUSION. Accelerated sequences with DLR provide 67% scan time reduction with similar subjective image quality, artifacts, and diagnostic performance to standard sequences. CLINICAL IMPACT. Accelerated sequences with DLR may provide an alternative to standard sequences for clinical shoulder MRI.