Super-resolution head and neck MRA using deep machine learning

Imaging the cerebral vasculature at different scales: translational tools to investigate the neurovascular interfaces

The improvements in imaging technology opened up the possibility to investigate the structure and function of cerebral vasculature and the neurovascular unit with unprecedented precision and gaining deep insights not only on the morphology of the vessels but also regarding their function and regulation related to the cerebral activity. In this review, we will dissect the different imaging capabilities regarding the cerebrovascular tree, the neurovascular unit, the haemodynamic response function, and thus, the vascular-neuronal coupling. We will discuss both clinical and preclinical setting, with a final discussion on the current scenery in cerebrovascular imaging where magnetic resonance imaging and multimodal microscopy emerge as the most potent and versatile tools, respectively, in the clinical and preclinical context.

Robust resolution improvement of 3D UTE-MR angiogram of normal vasculatures using super-resolution convolutional neural network

Contrast-enhanced UTE-MRA provides detailed angiographic information but at the cost of prolonged scanning periods, which may impose moving artifacts and affect the promptness of diagnosis and treatment of time-sensitive diseases like stroke. This study aims to increase the resolution of rapidly acquired low-resolution UTE-MRA data to high-resolution using deep learning. A total of 20 and 10 contrast-enhanced 3D UTE-MRA data were collected from healthy control and stroke-bearing Wistar rats, respectively. A newly designed 3D convolutional neural network called ladder-shaped residual dense generator (LSRDG) and other state-of-the-art models (SR-ResNet, MRDG64) were implemented, trained, and validated on healthy control data and tested on stroke data. For healthy control data, significantly improved SSIM, PSNR, and MSE results were achieved using our proposed model, respectively 0.983, 36.80, and 0.00021, compared to 0.964, 34.38, and 0.00037 using SR-ResNet and 0.978, 35.47, and 0.00029 using MRDG64. For stroke data, respective SSIM, PSNR, and MSE scores of 0.963, 34.14, and 0.00041 were achieved using our proposed model compared to 0.953, 32.24, and 0.00061 (SR-ResNet) and 0.957, 32.90, and 0.00054 (MRDG64). Moreover, by combining a well-designed network, suitable loss function, and training with smaller patch sizes, the resolution of contrast-enhanced UTE-MRA was significantly improved from 2343 μm3 to 1173 μm3.

Accelerated intracranial time-of-flight MR angiography with image-based deep learning image enhancement reduces scan times and improves image quality at 3-T and 1.5-T

PURPOSE

Three-dimensional time-of-flight magnetic resonance angiography (TOF-MRA) is effective for cerebrovascular disease assessment, but clinical application is limited by long scan times and low spatial resolution. Recent advances in deep learning-based reconstruction have shown the potential to improve image quality and reduce scan times. This study aimed to evaluate the effectiveness of accelerated intracranial TOF-MRA using deep learning-based image enhancement (TOF-DL) compared to conventional TOF-MRA (TOF-Con) at both 3-T and 1.5-T.

MATERIALS AND METHODS

In this retrospective study, patients who underwent both conventional and 40% accelerated TOF-MRA protocols on 1.5-T or 3-T scanners from July 2022 to March 2023 were included. A commercially available DL-based image enhancement algorithm was applied to the accelerated MRA. Quantitative image quality assessments included signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), contrast ratio (CR), and vessel sharpness (VS), while qualitative assessments were conducted using a five-point Likert scale. Cohen's d was used to compare the quantitative image metrics, and a cumulative link mixed regression model analyzed the readers' scores.

RESULTS

A total of 129 patients (mean age, 64 years ± 12 [SD], 99 at 3-T and 30 at 1.5-T) were included. TOF-DL showed significantly higher SNR, CNR, CR, and VS compared to TOF-Con (CNR = 183.89 vs. 45.58; CR = 0.63 vs. 0.59; VS = 0.73 vs. 0.61; all p < 0.001). The improvement in VS was more pronounced at 1.5-T (Cohen's d = 2.39) compared to 3-T HR and routine (Cohen's d = 0.83 and 0.75, respectively). TOF-DL also outperformed TOF-Con in qualitative image parameters, enhancing the visibility of small- and medium-sized vessels, regardless of the degree of resolution and field strength. TOF-DL showed comparable diagnostic accuracy (AUC: 0.77-0.85) to TOF-Con (AUC: 0.79-0.87) but had higher specificity for steno-occlusive lesions.

CONCLUSIONS

Accelerated intracranial MRA with deep learning-based reconstruction reduces scan times by 40% and significantly enhances image quality over conventional TOF-MRA at both 3-T and 1.5-T.

Sub-1-min relaxation-enhanced non-contrast non-triggered cervical MRA using compressed SENSE with deep learning reconstruction in healthy volunteers

BACKGROUND

We evaluated the acceleration of a three-dimensional isotropic flow-independent magnetic resonance angiography (MRA) (relaxation-enhanced angiography without contrast and triggering, REACT) of neck arteries using compressed SENSE (CS) combined with deep learning (adaptive intelligence, AI)-based reconstruction (CS-AI).

METHODS

Thirty-four volunteers received 3-T REACT MRA, acquired threefold: (i) CS acceleration factor 7 (CS7), scan time 1:20 min:s; (ii) CS acceleration factor 10 (CS10), scan time 0:55 min:s; and (iii) CS-AI acceleration factor 10 (CS10-AI), scan time 0:55 min:s. Two radiologists rated the image quality of seven arterial segments and overall image noise. Additionally, a pairwise forced-choice comparison was conducted. Apparent signal-to-noise ratio (aSNR) and contrast-to-noise ratio (aCNR) were measured, and image sharpness was assessed using the edge-rise distance (ERD). Multiple t-tests and nonparametric tests with Bonferroni correction were performed for comparison to CS7 as the reference standard.

RESULTS

Compared to CS7, CS10 showed lower image quality (p < 0.001) while CS10-AI obtained higher scores (p = 0.010). Image noise was similar between CS7 and CS10 (p = 0.138) while CS10-AI yielded a lower noise (p = 0.008). Forced choice revealed preferences for CS7 over CS10 (p < 0.001), but no preference between CS7 and CS10-AI (p > 0.999). Compared to CS7, aSNR and aCNR were lower in CS10 (p < 0.001) and the ERD was longer (p = 0.004), while CS10-AI provided better aSNR and aCNR (p = 0.001) and showed no difference in ERD (p = 0.776).

CONCLUSION

Sub-1-min CS-AI cervical REACT MRA was acquired without compromising image quality.

RELEVANCE STATEMENT

The implementation of a fast and reliable non-contrast MRA has the potential to reduce costs and time while increasing patient comfort and safety. Clinical studies evaluating the diagnostic performance for stenosis or dissection are needed.

TRIAL REGISTRATION

DRKS00030210 (German Clinical Trials Register; https://drks.de/ ) KEY POINTS: Deep learning reconstruction enables sub-1-min non-contrast-enhanced MRA of extracranial arteries. Acceleration without deep learning reconstruction causes inferior image quality. Acceleration with deep learning reconstruction exceeds, in part, the clinical standard.

Simultaneous Luminal and Hemodynamic Evaluation of the Cervical Arteries Using Nonenhanced 3D Quantitative Quiescent-Interval Slice-Selective Magnetic Resonance Angiography

BACKGROUND

Luminal and hemodynamic evaluations of the cervical arteries inform the diagnosis and management of patients with cervical arterial disease.

PURPOSE

To demonstrate a 3D nonenhanced quantitative quiescent interval slice-selective (qQISS) magnetic resonance angiographic (MRA) strategy that provides simultaneous hemodynamic and luminal evaluation of the cervical arteries.

STUDY TYPE

Prospective.

SUBJECTS

Six healthy volunteers (3 female, 3 male, age = 35.7 ± 10.3 years) and 14 patients with cerebrovascular disease (12 female, 2 male, age = 56.6 ± 14.0 years).

FIELD STRENGTH/SEQUENCES

3 T, ungated 3D tilted-slab qQISS, pulse-gated 2D phase contrast (PC), ungated 3D PC, and 3D time-of-flight (TOF) gradient-echo protocols.

ASSESSMENT

Four readers scored 29 arterial segments on 3D qQISS volumes for image quality using a 4-point scale (1: non-diagnostic, 2: fair, 3: good, 4: excellent). Time-averaged arterial flow velocities and volume flow rates obtained with qQISS and PC protocols were compared. Arterial lumen area and radius measures obtained with 3D protocols were compared in a subgroup.

STATISTICAL TESTS

Gwet's AC2; intraclass correlation coefficient (ICC); Pearson's correlation; Bland-Altman. P values <0.05 were considered statistically significant.

RESULTS

3D qQISS provided good-to-excellent image quality for depicting the cervical arteries (mean scores of 3.72 ± 0.55, 3.55 ± 0.66, 3.42 ± 0.72, and 3.66 ± 0.73 for readers 1, 2, 3, and 4) with significant inter-reader agreement (AC2 = 0.91, ICC = 0.53) in image scoring, significantly agreed with pulse-gated 2D PC for time-averaged total flow velocity (ICC = 0.83) and volume flow rate (ICC = 0.92), and significantly agreed with 3D PC for total flow velocity (ICC = 0.70), volume flow rate (ICC = 0.91), and component flow velocity (ICC = 0.89). Compared with 3D PC, 3D qQISS better agreed with 3D TOF for arterial lumen area (ICC = 0.97 vs. 0.72) and radius (ICC = 0.94 vs. 0.74).

DATA CONCLUSION

Nonenhanced 3D qQISS provides high-quality sub-1 mm3 spatial resolution imaging of the cervical arteries, excellent agreement of arterial structural measures with respect to 3D TOF, and time-averaged hemodynamic data without the need for additional PC imaging.

PLAIN LANGUAGE SUMMARY

Magnetic resonance angiography (MRA), a method for depicting blood vessels within the body, can be used to evaluate arterial diseases and disorders of the neck. MRA methods routinely used to evaluate the neck arteries do not measure blood flow speed and volume, while other methods for obtaining this information provide less accurate pictures of arterial structure and are not routinely collected. This article reports a new method for MRA that clearly and efficiently portrays the neck arteries without using injected dyes, and provides measurements of arterial blood flow speed and volume.

EVIDENCE LEVEL

2 TECHNICAL EFFICACY: Stage 1.

Improved visualization of intracranial distal arteries with multiple 2D slice dynamic ASL-MRA and super-resolution convolutional neural network

PURPOSE

To develop a novel framework to improve the visualization of distal arteries in arterial spin labeling (ASL) dynamic MRA.

METHODS

The attenuation of ASL blood signal due to the repetitive application of excitation RF pulses was minimized by splitting the acquisition volume into multiple thin 2D (M2D) slices, thereby reducing the exposure of the arterial blood magnetization to RF pulses while it flows within the brain. To improve the degraded vessel visualization in the slice direction due to the limited minimum achievable 2D slice thickness, a super-resolution (SR) convolutional neural network (CNN) was trained by using 3D time-of-flight (TOF)-MRA images from a large public dataset. And then, we applied domain transfer from 3D TOF-MRA to M2D ASL-MRA, while avoiding acquiring a large number of ASL-MRA data required for CNN training.

RESULTS

Compared to the conventional 3D ASL-MRA, far more distal arteries were visualized with higher signal intensity by using M2D ASL-MRA. In general, however, the vessel visualization with a conventional interpolation was prone to be blurry and unclear due to the limited spatial resolution in the slice direction, particularly in small vessels. Application of CNN-based SR transferred from 3D TOF-MRA to M2D ASL-MRA successfully addressed such a limitation and achieved clearer visualization of small vessels than conventional interpolation.

CONCLUSION

This study demonstrated that the proposed framework provides improved visualization of distal arteries in later dynamic phases, which will particularly benefit the application of this approach in patients with cerebrovascular disease who have slow blood flow.

Ultra-low-field magnetic resonance angiography at 0.05 T: A preliminary study

We aim to explore the feasibility of head and neck time-of-flight (TOF) magnetic resonance angiography (MRA) at ultra-low-field (ULF). TOF MRA was conducted on a highly simplified 0.05 T MRI scanner with no radiofrequency (RF) and magnetic shielding. A flow-compensated three-dimensional (3D) gradient echo (GRE) sequence with a tilt-optimized nonsaturated excitation RF pulse, and a flow-compensated multislice two-dimensional (2D) GRE sequence, were implemented for cerebral artery and vein imaging, respectively. For carotid artery and jugular vein imaging, flow-compensated 2D GRE sequences were utilized with venous and arterial blood presaturation, respectively. MRA was performed on young healthy subjects. Vessel-to-background contrast was experimentally observed with strong blood inflow effect and background tissue suppression. The large primary cerebral arteries and veins, carotid arteries, jugular veins, and artery bifurcations could be identified in both raw GRE images and maximum intensity projections. The primary brain and neck arteries were found to be reproducible among multiple examination sessions. These preliminary experimental results demonstrated the possibility of artery TOF MRA on low-cost 0.05 T scanners for the first time, despite the extremely low MR signal. We expect to improve the quality of ULF TOF MRA in the near future through sequence development and optimization, ongoing advances in ULF hardware and image formation, and the use of vascular T1 contrast agents.

Deep Learning Accelerated Brain Diffusion-Weighted MRI with Super Resolution Processing

OBJECTIVES

To investigate the clinical feasibility and image quality of accelerated brain diffusion-weighted imaging (DWI) with deep learning image reconstruction and super resolution.

METHODS

85 consecutive patients with clinically indicated MRI at a 3 T scanner were prospectively included. Conventional diffusion-weighted data (c-DWI) with four averages were obtained. Reconstructions of one and two averages, as well as deep learning diffusion-weighted imaging (DL-DWI), were accomplished. Three experienced readers evaluated the acquired data using a 5-point Likert scale regarding overall image quality, overall contrast, diagnostic confidence, occurrence of artefacts and evaluation of the central region, basal ganglia, brainstem, and cerebellum. To assess interrater agreement, Fleiss' kappa (ϰ) was determined. Signal intensity (SI) levels for basal ganglia and the central region were estimated via automated segmentation, and SI values of detected pathologies were measured.

RESULTS

Intracranial pathologies were identified in 35 patients. DL-DWI was significantly superior for all defined parameters, independently from applied averages (p-value <0.001). Optimum image quality was achieved with DL-DWI by utilizing a single average (p-value <0.001), demonstrating very good (80.9%) to excellent image quality (14.5%) in nearly all cases, compared to 12.5% with very good and 0% with excellent image quality for c-MRI (p-value <0.001). Comparable results could be shown for diagnostic confidence. Inter-rater Fleiss' Kappa demonstrated moderate to substantial agreement for virtually all defined parameters, with good accordance, particularly for the assessment of pathologies (p = 0.74). Regarding SI values, no significant difference was found.

CONCLUSION

Ultra-fast diffusion-weighted imaging with super resolution is feasible, resulting in highly accelerated brain imaging while increasing diagnostic image quality.

Deep learning improves quality of intracranial vessel wall MRI for better characterization of potentially culprit plaques

Intracranial vessel wall imaging (VWI), which requires both high spatial resolution and high signal-to-noise ratio (SNR), is an ideal candidate for deep learning (DL)-based image quality improvement. Conventional VWI (Conv-VWI, voxel size 0.51 × 0.51 × 0.45 mm3) and denoised super-resolution DL-VWI (0.28 × 0.28 × 0.45 mm3) of 117 patients were analyzed in this retrospective study. Quality of the images were compared qualitatively and quantitatively. Diagnostic performance for identifying potentially culprit atherosclerotic plaques, using lesion enhancement and presence of intraplaque hemorrhage (IPH), was evaluated. DL-VWI significantly outperformed Conv-VWI in all image quality ratings (all P < .001). DL-VWI demonstrated higher SNR and contrast-to-noise ratio (CNR) than Conv-VWI, both in normal walls (basilar artery; SNR 4.83 ± 1.23 vs. 3.02 ± 0.59, P < .001) and lesions (contrast-enhanced images; SNR 22.12 ± 11.68 vs. 8.33 ± 3.26, P < .001). In the assessment of 86 lesions, DL-VWI showed higher confidence of detection (4.56 ± 0.55 vs. 2.62 ± 0.77, P < .001), more concordant IPH characterization (Cohen's Kappa 0.85 vs. 0.59) and greater enhancement. For culprit plaque identification, IPH exhibited higher sensitivity in DL-VWI compared to Conv-VWI (70.6% vs. 23.5%) and excellent specificity (94.3% vs. 94.3%). Deep learning application of intracranial vessel wall images successfully improved the quality and resolution of the images. This aided in detecting vessel wall lesions and intraplaque hemorrhage, and in identifying potentially culprit atherosclerotic plaques.

High Resolution TOF-MRA Using Compressed Sensing-based Deep Learning Image Reconstruction for the Visualization of Lenticulostriate Arteries: A Preliminary Study

PURPOSE

To investigate the visibility of the lenticulostriate arteries (LSAs) in time-of-flight (TOF)-MR angiography (MRA) using compressed sensing (CS)-based deep learning (DL) image reconstruction by comparing its image quality with that obtained by the conventional CS algorithm.

METHODS

Five healthy volunteers were included. High-resolution TOF-MRA images with the reduction (R)-factor of 1 were acquired as full-sampling data. Images with R-factors of 2, 4, and 6 were then reconstructed using CS-DL and conventional CS (the combination of CS and sensitivity conceding; CS-SENSE) reconstruction, respectively. In the quantitative assessment, the number of visible LSAs (identified by two radiologists), length of each depicted LSA (evaluated by one radiological technologist), and normalized mean squared error (NMSE) value were assessed. In the qualitative assessment, the overall image quality and the visibility of the peripheral LSA were visually evaluated by two radiologists.

RESULTS

In the quantitative assessment of the DL-CS images, the number of visible LSAs was significantly higher than those obtained with CS-SENSE in the R-factors of 4 and 6 (Reader 1) and in the R-factor of 6 (Reader 2). The length of the depicted LSAs in the DL-CS images was significantly longer in the R-factor 6 compared to the CS-SENSE result. The NMSE value in CS-DL was significantly lower than in CS-SENSE for R-factors of 4 and 6. In the qualitative assessment of DL-CS images, the overall image quality was significantly higher than that obtained with CS-SENSE in the R-factors 4 and 6 (Reader 1) and in the R-factor 4 (Reader 2). The visibility of the peripheral LSA was significantly higher than that shown by CS-SENSE in all R-factors (Reader 1) and in the R-factors 2 and 4 (Reader 2).

CONCLUSION

CS-DL reconstruction demonstrated preserved image quality for the depiction of LSAs compared to the conventional CS-SENSE when the R-factor is elevated.

Deep Learning-Based High-Resolution Magnetic Resonance Angiography (MRA) Generation Model for 4D Time-Resolved Angiography with Interleaved Stochastic Trajectories (TWIST) MRA in Fast Stroke Imaging

PURPOSE

The purpose of this study is to improve the qualitative and quantitative image quality of the time-resolved angiography with interleaved stochastic trajectories technique (4D-TWIST-MRA) using deep neural network (DNN)-based MR image reconstruction software.

MATERIALS AND METHODS

A total of 520 consecutive patients underwent 4D-TWIST-MRA for ischemic stroke or intracranial vessel stenosis evaluation. Four-dimensional DNN-reconstructed MRA (4D-DNR) was generated using commercially available software (SwiftMR v.3.0.0.0, AIRS Medical, Seoul, Republic of Korea). Among those evaluated, 397 (76.3%) patients received concurrent time-of-flight MRA (TOF-MRA) to compare the signal-to-noise ratio (SNR), image quality, noise, sharpness, vascular conspicuity, and degree of venous contamination with a 5-point Likert scale. Two radiologists independently evaluated the detection rate of intracranial aneurysm in TOF-MRA, 4D-TWIST-MRA, and 4D-DNR in separate sessions. The other 123 (23.7%) patients received 4D-TWIST-MRA due to a suspicion of acute ischemic stroke. The confidence level and decision time for large vessel occlusion were evaluated in these patients.

RESULTS

In qualitative analysis, 4D-DNR demonstrated better overall image quality, sharpness, vascular conspicuity, and noise reduction compared to 4D-TWIST-MRA. Moreover, 4D-DNR exhibited a higher SNR than 4D-TWIST-MRA. The venous contamination and aneurysm detection rates were not significantly different between the two MRA images. When compared to TOF-MRA, 4D-CE-MRA underestimated the aneurysm size (2.66 ± 0.51 vs. 1.75 ± 0.62, p = 0.029); however, 4D-DNR showed no significant difference in size compared to TOF-MRA (2.66 ± 0.51 vs. 2.10 ± 0.41, p = 0.327). In the diagnosis of large vessel occlusion, 4D-DNR showed a better confidence level and shorter decision time than 4D-TWIST-MRA.

CONCLUSION

DNN reconstruction may improve the qualitative and quantitative image quality of 4D-TWIST-MRA, and also enhance diagnostic performance for intracranial aneurysm and large vessel occlusion.

Exploring the impact of super-resolution deep learning on MR angiography image quality

PURPOSE

The aim of this study is to assess the effect of super-resolution deep learning-based reconstruction (SR-DLR), which uses k-space properties, on image quality of intracranial time-of-flight (TOF) magnetic resonance angiography (MRA) at 3 T.

METHODS

This retrospective study involved 35 patients who underwent intracranial TOF-MRA using a 3-T MRI system with SR-DLR based on k-space properties in October and November 2022. We reconstructed MRA with SR-DLR (matrix = 1008 × 1008) and MRA without SR-DLR (matrix = 336 × 336). We measured the signal-to-noise ratio (SNR), contrast, and contrast-to-noise ratio (CNR) in the basilar artery (BA) and the anterior cerebral artery (ACA) and the sharpness of the posterior cerebral artery (PCA) using the slope of the signal intensity profile curve at the half-peak points. Two radiologists evaluated image noise, artifacts, contrast, sharpness, and overall image quality of the two image types using a 4-point scale. We compared quantitative and qualitative scores between images with and without SR-DLR using the Wilcoxon signed-rank test.

RESULTS

The SNRs, contrasts, and CNRs were all significantly higher in images with SR-DLR than those without SR-DLR (p < 0.001). The slope was significantly greater in images with SR-DLR than those without SR-DLR (p < 0.001). The qualitative scores in MRAs with SR-DLR were all significantly higher than MRAs without SR-DLR (p < 0.001).

CONCLUSION

SR-DLR with k-space properties can offer the benefits of increased spatial resolution without the associated drawbacks of longer scan times and reduced SNR and CNR in intracranial MRA.

Ultrafast Brain MRI Protocol at 1.5 T Using Deep Learning and Multi-shot EPI

RATIONALE AND OBJECTIVES

To evaluate clinical feasibility and image quality of a comprehensive ultrafast brain MRI protocol with multi-shot echo planar imaging and deep learning-enhanced reconstruction at 1.5T.

MATERIALS AND METHODS

Thirty consecutive patients who underwent clinically indicated MRI at a 1.5 T scanner were prospectively included. A conventional MRI (c-MRI) protocol, including T1-, T2-, T2*-, T2-FLAIR, and diffusion-weighted images (DWI)-weighted sequences were acquired. In addition, ultrafast brain imaging with deep learning-enhanced reconstruction and multi-shot EPI (DLe-MRI) was performed. Subjective image quality was evaluated by three readers using a 4-point Likert scale. To assess interrater agreement, Fleiss' kappa (ϰ) was determined. For objective image analysis, relative signal intensity levels for grey matter, white matter, and cerebrospinal fluid were calculated.

RESULTS

Time of acquisition (TA) of c-MRI protocols added up to 13:55 minutes, whereas the TA of DLe-MRI-based protocol added up to 3:04 minutes, resulting in a time reduction of 78%. All DLe-MRI acquisitions yielded diagnostic image quality with good absolute values for subjective image quality. C-MRI demonstrated slight advantages for DWI in overall subjective image quality (c-MRI: 3.93 [+/- 0.25] vs DLe-MRI: 3.87 [+/- 0.37], P = .04) and diagnostic confidence (c-MRI: 3.93 [+/- 0.25] vs DLe-MRI: 3.83 [+/- 3.83], P = .01). For most evaluated quality scores, moderate interobserver agreement was found. Objective image evaluation revealed comparable results for both techniques.

CONCLUSION

DLe-MRI is feasible and allows for highly accelerated comprehensive brain MRI within 3minutes at 1.5 T with good image quality. This technique may potentially strengthen the role of MRI in neurological emergencies.

Physics-informed deep learning for T2-deblurred superresolution turbo spin echo MRI

PURPOSE

Deep learning superresolution (SR) is a promising approach to reduce MRI scan time without requiring custom sequences or iterative reconstruction. Previous deep learning SR approaches have generated low-resolution training images by simple k-space truncation, but this does not properly model in-plane turbo spin echo (TSE) MRI resolution degradation, which has variable T2 relaxation effects in different k-space regions. To fill this gap, we developed a T2 -deblurred deep learning SR method for the SR of 3D-TSE images.

METHODS

A SR generative adversarial network was trained using physically realistic resolution degradation (asymmetric T2 weighting of raw high-resolution k-space data). For comparison, we trained the same network structure on previous degradation models without TSE physics modeling. We tested all models for both retrospective and prospective SR with 3 × 3 acceleration factor (in the two phase-encoding directions) of genetically engineered mouse embryo model TSE-MR images.

RESULTS

The proposed method can produce high-quality 3 × 3 SR images for a typical 500-slice volume with 6-7 mouse embryos. Because 3 × 3 SR was performed, the image acquisition time can be reduced from 15 h to 1.7 h. Compared to previous SR methods without TSE modeling, the proposed method achieved the best quantitative imaging metrics for both retrospective and prospective evaluations and achieved the best imaging-quality expert scores for prospective evaluation.

CONCLUSION

The proposed T2 -deblurring method improved accuracy and image quality of deep learning-based SR of TSE MRI. This method has the potential to accelerate TSE image acquisition by a factor of up to 9.

Current State of Artificial Intelligence in Clinical Applications for Head and Neck MR Imaging

Due primarily to the excellent soft tissue contrast depictions provided by MRI, the widespread application of head and neck MRI in clinical practice serves to assess various diseases. Artificial intelligence (AI)-based methodologies, particularly deep learning analyses using convolutional neural networks, have recently gained global recognition and have been extensively investigated in clinical research for their applicability across a range of categories within medical imaging, including head and neck MRI. Analytical approaches using AI have shown potential for addressing the clinical limitations associated with head and neck MRI. In this review, we focus primarily on the technical advancements in deep-learning-based methodologies and their clinical utility within the field of head and neck MRI, encompassing aspects such as image acquisition and reconstruction, lesion segmentation, disease classification and diagnosis, and prognostic prediction for patients presenting with head and neck diseases. We then discuss the limitations of current deep-learning-based approaches and offer insights regarding future challenges in this field.

Application of deep learning-based super-resolution to T1-weighted postcontrast gradient echo imaging of the chest

OBJECTIVES

A deep learning-based super-resolution for postcontrast volume-interpolated breath-hold examination (VIBE) of the chest was investigated in this study. Aim was to improve image quality, noise, artifacts and diagnostic confidence without change of acquisition parameters.

MATERIALS AND METHODS

Fifty patients who received VIBE postcontrast imaging of the chest at 1.5 T were included in this retrospective study. After acquisition of the standard VIBE (VIBE_S), a novel deep learning-based algorithm and a denoising algorithm were applied, resulting in enhanced images (VIBE_DL). Two radiologists qualitatively evaluated both datasets independently, rating sharpness of soft tissue, vessels, bronchial structures, lymph nodes, artifacts, cardiac motion artifacts, noise levels and overall diagnostic confidence, using a Likert scale ranging from 1 to 4. In the presence of lung lesions, the largest lesion was rated regarding sharpness and diagnostic confidence using the same Likert scale as mentioned above. Additionally, the largest diameter of the lesion was measured.

RESULTS

The sharpness of soft tissue, vessels, bronchial structures and lymph nodes as well as the diagnostic confidence, the extent of artifacts, the extent of cardiac motion artifacts and noise levels were rated superior in VIBE_DL (all P < 0.001). There was no significant difference in the diameter or the localization of the largest lung lesion in VIBE_DL compared to VIBE_S. Lesion sharpness as well as detectability was rated significantly better by both readers with VIBE_DL (both P < 0.001).

CONCLUSION

The application of a novel deep learning-based super-resolution approach in T1-weighted VIBE postcontrast imaging resulted in an improvement in image quality, noise levels and diagnostic confidence as well as in a shortened acquisition time.

Combined Deep Learning-based Super-Resolution and Partial Fourier Reconstruction for Gradient Echo Sequences in Abdominal MRI at 3 Tesla: Shortening Breath-Hold Time and Improving Image Sharpness and Lesion Conspicuity

RATIONALE AND OBJECTIVES

To investigate the impact of a prototypical deep learning-based super-resolution reconstruction algorithm tailored to partial Fourier acquisitions on acquisition time and image quality for abdominal T1-weighted volume-interpolated breath-hold examination (VIBE_SR) at 3 Tesla. The standard T1-weighted images were used as the reference standard (VIBE_SD).

MATERIALS AND METHODS

Patients with diverse abdominal pathologies, who underwent a clinically indicated contrast-enhanced abdominal VIBE magnetic resonance imaging at 3T between March and June 2021 were retrospectively included. Following the acquisition of the standard VIBE_SD sequences, additional images for the non-contrast, dynamic contrast-enhanced and post-contrast T1-weighted VIBE acquisition were retrospectively reconstructed using the same raw data and employing a prototypical deep learning-based super-resolution reconstruction algorithm. The algorithm was designed to enhance edge sharpness by avoiding conventional k-space filtering and to perform a partial Fourier reconstruction in the slice phase-encoding direction for a predefined asymmetric sampling ratio. In the retrospective reconstruction, the asymmetric sampling was realized by omitting acquired samples at the end of the acquisition and therefore corresponding to a shorter acquisition. Four radiologists independently analyzed the image datasets (VIBE_SR and VIBE_SD) in a blinded manner. Outcome measures were: sharpness of abdominal organs, sharpness of vessels, image contrast, noise, hepatic lesion conspicuity and size, overall image quality and diagnostic confidence. These parameters were statistically compared and interrater reliability was computed using Fleiss' Kappa and intraclass correlation coefficient (ICC). Finally, the rate of detection of hepatic lesions was documented and was statistically compared using the paired Wilcoxon test.

RESULTS

A total of 32 patients aged 59 ± 16 years (23 men (72%), 9 women (28%)) were included. For VIBE_SR, breath-hold time was significantly reduced by approximately 13.6% (VIBE_SR 11.9 ± 1.2 seconds vs. VIBE_SD: 13.9 ± 1.4 seconds, p < 0.001). All readers rated sharpness of abdominal organs, sharpness of vessels to be superior in images with VIBE_SR (p values ranged between p = 0.005 and p < 0.001). Despite reduction of acquisition time, image contrast, noise, overall image quality and diagnostic confidence were not compromised, as there was no evidence of a difference between VIBE_SR and VIBE_SD (p > 0.05). The inter-reader agreement was substantial with a Fleiss' Kappa of >0.7 in all contrast phases. A total of 13 hepatic lesions were analyzed. The four readers observed a superior lesion conspicuity in VIBE_SR than in VIBE_SD (p values ranged between p = 0.046 and p < 0.001). In terms of lesion size, there was no significant difference between VIBE_SD and VIBE_SR for all readers. Finally, there was an excellent inter-reader agreement regarding lesion size (ICC > 0.9). For all readers, no statistically significant difference was observed regarding detection of hepatic lesions between VIBE_SD and VIBE_SR.

CONCLUSION

The deep learning-based super-resolution reconstruction with partial Fourier in the slice phase-encoding direction enabled a reduction of breath-hold time and improved image sharpness and lesion conspicuity in T1-weighted gradient echo sequences in abdominal magnetic resonance imaging at 3 Tesla. Faster acquisition time without compromising image quality or diagnostic confidence was possible by using this deep learning-based reconstruction technique.

Analysis of a Deep Learning-Based Superresolution Algorithm Tailored to Partial Fourier Gradient Echo Sequences of the Abdomen at 1.5 T: Reduction of Breath-Hold Time and Improvement of Image Quality

OBJECTIVES

The aim of this study was to investigate the feasibility and impact of a novel deep learning superresolution algorithm tailored to partial Fourier allowing retrospectively theoretical acquisition time reduction in 1.5 T T1-weighted gradient echo imaging of the abdomen.

MATERIALS AND METHODS

Fifty consecutive patients who underwent a 1.5 T contrast-enhanced magnetic resonance imaging examination of the abdomen between April and May 2021 were included in this retrospective study. After acquisition of a conventional T1-weighted volumetric interpolated breath-hold examination using Dixon for water-fat separation (VIBEStd), the acquired data were reprocessed including a superresolution algorithm that was optimized for partial Fourier acquisitions (VIBESR). To accelerate theoretically the acquisition process, a more aggressive partial Fourier setting was applied in VIBESR reconstructions practically corresponding to a shorter acquisition for the data included in the retrospective reconstruction. Precontrast, dynamic contrast-enhanced, and postcontrast data sets were processed. Image analysis was performed by 2 radiologists independently in a blinded random order without access to clinical data regarding the following criteria using a Likert scale ranging from 1 to 4 with 4 being the best: noise levels, sharpness and contrast of vessels, sharpness and contrast of organs and lymph nodes, overall image quality, diagnostic confidence, and lesion conspicuity.Wilcoxon signed rank test for paired data was applied to test for significance.

RESULTS

Mean patient age was 61 ± 14 years. Mean acquisition time for the conventional VIBEStd sequence was 15 ± 1 seconds versus theoretical 13 ± 1 seconds of acquired data used for the VIBESR reconstruction. Noise levels were evaluated to be better in VIBESR with a median of 4 (4-4) versus a median of 3 (3-3) in VIBEStd by both readers (P < 0.001). Sharpness and contrast of vessels as well as organs and lymph nodes were also evaluated to be superior in VIBESR compared with VIBEStd with a median of 4 (4-4) versus a median of 3 (3-3) (P < 0.001). Diagnostic confidence was also rated superior in VIBESR with a median of 4 (4-4) versus a median of 3.5 (3-4) in VIBEStd by reader 1 and with a median of 4 (4-4) for VIBESR and a median of 4 (4-4) for VIBEStd by reader 2 (both P < 0.001).

CONCLUSIONS

Image enhancement using deep learning-based superresolution tailored to partial Fourier acquisitions of T1-weighted gradient echo imaging of the abdomen provides improved image quality and diagnostic confidence in combination with more aggressive partial Fourier settings leading to shorter scan time.

Synthesizing high-resolution magnetic resonance imaging using parallel cycle-consistent generative adversarial networks for fast magnetic resonance imaging

PURPOSE

The common practice in acquiring the magnetic resonance (MR) images is to obtain two-dimensional (2D) slices at coarse locations while keeping the high in-plane resolution in order to ensure enough body coverage while shortening the MR scan time. The aim of this study is to propose a novel method to generate HR MR images from low-resolution MR images along the longitudinal direction. In order to address the difficulty of collecting paired low- and high-resolution MR images in clinical settings and to gain the advantage of parallel cycle consistent generative adversarial networks (CycleGANs) in synthesizing realistic medical images, we developed a parallel CycleGANs based method using a self-supervised strategy.

METHODS AND MATERIALS

The proposed workflow consists of two parallely trained CycleGANs to independently predict the HR MR images in the two planes along the directions that are orthogonal to the longitudinal MR scan direction. Then, the final synthetic HR MR images are generated by fusing the two predicted images. MR images, including T1-weighted (T1), contrast enhanced T1-weighted (T1CE), T2-weighted (T2), and T2 Fluid Attenuated Inversion Recovery (FLAIR), of the multimodal brain tumor segmentation challenge 2020 (BraTS2020) dataset were processed to evaluate the proposed workflow along the cranial-caudal (CC), lateral, and anterior-posterior directions. Institutional collected MR images were also processed for evaluation of the proposed method. The performance of the proposed method was investigated via both qualitative and quantitative evaluations. Metrics of normalized mean absolute error (NMAE), peak signal-to-noise ratio (PSNR), edge keeping index (EKI), structural similarity index measurement (SSIM), information fidelity criterion (IFC), and visual information fidelity in pixel domain (VIFP) were calculated.

RESULTS

It is shown that the proposed method can generate HR MR images visually indistinguishable from the ground truth in the investigations on the BraTS2020 dataset. In addition, the intensity profiles, difference images and SSIM maps can also confirm the feasibility of the proposed method for synthesizing HR MR images. Quantitative evaluations on the BraTS2020 dataset shows that the calculated metrics of synthetic HR MR images can all be enhanced for the T1, T1CE, T2, and FLAIR images. The enhancements in the numerical metrics over the low-resolution and bi-cubic interpolated MR images, as well as those genearted with a comparative deep learning method, are statistically significant. Qualitative evaluation of the synthetic HR MR images of the clinical collected dataset could also confirm the feasibility of the proposed method.

CONCLUSIONS

The proposed method is feasible to synthesize HR MR images using self-supervised parallel CycleGANs, which can be expected to shorten MR acquisition time in clinical practices.

Deep Learning Applications in Magnetic Resonance Imaging: Has the Future Become Present?

Deep learning technologies and applications demonstrate one of the most important upcoming developments in radiology. The impact and influence of these technologies on image acquisition and reporting might change daily clinical practice. The aim of this review was to present current deep learning technologies, with a focus on magnetic resonance image reconstruction. The first part of this manuscript concentrates on the basic technical principles that are necessary for deep learning image reconstruction. The second part highlights the translation of these techniques into clinical practice. The third part outlines the different aspects of image reconstruction techniques, and presents a review of the current literature regarding image reconstruction and image post-processing in MRI. The promising results of the most recent studies indicate that deep learning will be a major player in radiology in the upcoming years. Apart from decision and diagnosis support, the major advantages of deep learning magnetic resonance imaging reconstruction techniques are related to acquisition time reduction and the improvement of image quality. The implementation of these techniques may be the solution for the alleviation of limited scanner availability via workflow acceleration. It can be assumed that this disruptive technology will change daily routines and workflows permanently.

Tile-net for undersampled cardiovascular CINE magnetic resonance imaging

We propose the "Tile-net" method based on dividing an image into smaller tiles. Using the tile as the input to the neural network, the network is simplified substantially. The Tile-net learns at a much faster rate than the networks without tiling. The training and reconstruction times for the Tile-net are reduced by 40% and 33%, respectively compared to the networks without tiling. The Tile-net performance is evaluated through the normalized mean square error (NMSE), peak signal to noise ratio (PSNR), structure similarity index measure (SSIM) and the quality of the reconstructed image for test datasets. The Tile-net does not degrade performance; however, it reduces the NMSE by 0.3% compared to the networks without tiling.

Deep Learning and Its Application to Function Approximation for MR in Medicine: An Overview

This article presents an overview of deep learning (DL) and its applications to function approximation for MR in medicine. The aim of this article is to help readers develop various applications of DL. DL has made a large impact on the literature of many medical sciences, including MR. However, its technical details are not easily understandable for non-experts of machine learning (ML).The first part of this article presents an overview of DL and its related technologies, such as artificial intelligence (AI) and ML. AI is explained as a function that can receive many inputs and produce many outputs. ML is a process of fitting the function to training data. DL is a kind of ML, which uses a composite of many functions to approximate the function of interest. This composite function is called a deep neural network (DNN), and the functions composited into a DNN are called layers. This first part also covers the underlying technologies required for DL, such as loss functions, optimization, initialization, linear layers, non-linearities, normalization, recurrent neural networks, regularization, data augmentation, residual connections, autoencoders, generative adversarial networks, model and data sizes, and complex-valued neural networks.The second part of this article presents an overview of the applications of DL in MR and explains how functions represented as DNNs are applied to various applications, such as RF pulse, pulse sequence, reconstruction, motion correction, spectroscopy, parameter mapping, image synthesis, and segmentation.