Deep Learning-based Reconstruction Research Articles

An efficient deep unrolling network for sparse-view CT reconstruction via alternating optimization of dense-view sinograms and images

Objective. Recently, there have been many advancements in deep unrolling methods for sparse-view computed tomography (SVCT) reconstruction. These methods combine model-based and deep learning-based reconstruction techniques, improving the interpretability and achieving significant results. However, they are often computationally expensive, particularly for clinical raw projection data with large sizes. This study aims to address this issue while maintaining the quality of the reconstructed image.Approach. The SVCT reconstruction task is decomposed into two subproblems using the proximal gradient method: optimizing dense-view sinograms and optimizing images. Then dense-view sinogram inpainting, image-residual learning, and image-refinement modules are performed at each iteration stage using deep neural networks. Unlike previous unrolling methods, the proposed method focuses on optimizing dense-view sinograms instead of full-view sinograms. This approach not only reduces computational resources and runtime but also minimizes the challenge for the network to perform sinogram inpainting when the sparse ratio is extremely small, thereby decreasing the propagation of estimation error from the sinogram domain to the image domain.Main results. The proposed method successfully reconstructs an image (512 × 512 pixels) from real-size (2304 × 736) projection data, with 3.39 M training parameters and an inference time of 0.09 s per slice on a GPU. The proposed method also achieves superior quantitative and qualitative results compared with state-of-the-art deep unrolling methods on datasets with sparse ratios of 1/12 and 1/18, especially in suppressing artifacts and preserving structural details. Additionally, results show that using dense-view sinogram inpainting not only accelerates the computational speed but also leads to faster network convergence and further improvements in reconstruction results.Significance. This research presents an efficient dual-domain deep unrolling technique that produces excellent results in SVCT reconstruction while requiring small computational resources. These findings have important implications for speeding up deep unrolling CT reconstruction methods and making them more practical for processing clinical CT projection data.

Optimizing hip MRI: enhancing image quality and elevating inter-observer consistency using deep learning-powered reconstruction

BackgroundConventional hip joint MRI scans necessitate lengthy scan durations, posing challenges for patient comfort and clinical efficiency. Previously, accelerated imaging techniques were constrained by a trade-off between noise and resolution. Leveraging deep learning-based reconstruction (DLR) holds the potential to mitigate scan time without compromising image quality.MethodsWe enrolled a cohort of sixty patients who underwent DL-MRI, conventional MRI, and No-DL MRI examinations to evaluate image quality. Key metrics considered in the assessment included scan duration, overall image quality, quantitative assessments of Relative Signal-to-Noise Ratio (rSNR), Relative Contrast-to-Noise Ratio (rCNR), and diagnostic efficacy. Two experienced radiologists independently assessed image quality using a 5-point scale (5 indicating the highest quality). To gauge interobserver agreement for the assessed pathologies across image sets, we employed weighted kappa statistics. Additionally, the Wilcoxon signed rank test was employed to compare image quality and quantitative rSNR and rCNR measurements.ResultsScan time was significantly reduced with DL-MRI and represented an approximate 66.5% reduction. DL-MRI consistently exhibited superior image quality in both coronal T2WI and axial T2WI when compared to both conventional MRI (p < 0.01) and No-DL-MRI (p < 0.01). Interobserver agreement was robust, with kappa values exceeding 0.735. For rSNR data, coronal fat-saturated(FS) T2WI and axial FS T2WI in DL-MRI consistently outperformed No-DL-MRI, with statistical significance (p < 0.01) observed in all cases. Similarly, rCNR data revealed significant improvements (p < 0.01) in coronal FS T2WI of DL-MRI when compared to No-DL-MRI. Importantly, our findings indicated that DL-MRI demonstrated diagnostic performance comparable to conventional MRI.ConclusionIntegrating deep learning-based reconstruction methods into standard clinical workflows has the potential to the promise of accelerating image acquisition, enhancing image clarity, and increasing patient throughput, thereby optimizing diagnostic efficiency.Trial registrationRetrospectively registered.

A practical approach to the spatial-domain calculation of nonprewhitening model observers in computed tomography.

Modern reconstruction algorithms for computed tomography (CT) can exhibit nonlinear properties, including non-stationarity of noise and contrast dependence of both noise and spatial resolution. Model observers have been recommended as a tool for the task-based assessment of image quality (Samei E etal., Med Phys. 2019; 46(11): e735-e756), but the common Fourier domain approach to their calculation assumes quasi-stationarity. A practical spatial-domain approach is proposed for the calculation of the nonprewhitening (NPW) family of model observers in CT, avoiding the disadvantages of the Fourier domain. The methodology avoids explicit estimation of a noise covariance matrix. A formula is also provided for the uncertainty on estimates of detectability index, for a given number of slices and repeat scans. The purpose of this work is to demonstrate the method and provide comparisons to the conventional Fourier approach for both iterative reconstruction (IR) and a deep Learning-based reconstruction (DLR) algorithm. Acquisitions were made on a Revolution CT scanner (GE Healthcare, Waukesha, Wisconsin, USA) and reconstructed using the vendor's IR and DLR algorithms (ASiR-V and TrueFidelity). Several reconstruction kernels were investigated (Standard, Lung, and Bone for IR and Standard for DLR). An in-house developed phantom with two flat contrast levels (2 and 8 mgI/mL) and varying feature size (1-10mm diameter) was used. Two single-energy protocols (80 and 120kV) were investigated with two dose levels (CTDIvol=5 and 13 mGy). The spatial domain calculations relied on repeated scanning, region-of-interest placement and simple operations with image matrices. No more repeat scans were utilized than required for Fourier domain estimations. Fourier domain calculations were made using techniques described in a previous publication (Thor D etal., Med Phys. 2023;50(5):2775-2786). Differences between the calculations in the two domains were assessed using the normalized root-mean-square discrepancy (NMRSD). Fourier domain calculations agreed closely with those in the spatial domain for all zero-strength IR reconstructions, which most closely resemble traditional filtered backprojection. The Fourier-based calculations, however, displayed higher detectability compared to those in the spatial domain for IR with strong iterative strength and for the DLR algorithm. The NRMSD remained within 10% for the NPW model observer without eye filter, but reached larger values when an eye filter was included. The formula for the uncertainty on the detectability index was validated by bootstrap estimates. A practical methodology was demonstrated for calculating NPW observers in the spatial domain. In addition to being a valuable tool for verifying the applicability of typical Fourier-based methodologies, it lends itself to routine calculations for features embedded in a phantom. Higher estimates of detectability were observed when adopting the Fourier domain methodology for IR and for a DLR algorithm, demonstrating that use of the Fourier domain can indicate greater benefit to noise suppression than suggested by spatial domain calculations. This is consistent with the results of previous authors for the Fourier domain, who have compared to human and other model observers, but not, as in this study, to the NPW model observer calculated in the spatial domain.

Clinical 7Tesla magnetic resonance imaging: Impact and patient value in neurological disorders.

Magnetic resonance imaging (MRI) is a cornerstone of non-invasive diagnostics and treatment monitoring, particularly for diseases of the central nervous system. Although 1.5- and 3Tesla (T) field strengths remain the clinical standard, the advent of 7T MRI represents a transformative step forward, offering superior spatial resolution, contrast, and sensitivity for visualizing neuroanatomy, metabolism, and function. Recent innovations, including parallel transmission and deep learning-based reconstruction, have resolved many prior technical challenges of 7T MRI, enabling its routine clinical use. This review examines the diagnostic impact, patient value, and practical considerations of 7T MRI, emphasizing its role in facilitating earlier diagnoses and improving care in conditions, such as amyotrophic lateral sclerosis (ALS), epilepsy, multiple sclerosis (MS), dementia, parkinsonism, tumors, and vascular diseases. Based on insights from over 1200 clinical scans with a second-generation 7T system, the review highlights disease-specific biomarkers such as the motor band sign in ALS and the new diagnostic markers in MS, the central vein sign, and paramagnetic rim lesions. The unparalleled ability of 7T MRI to study neurological diseases ex vivo at ultra-high resolution is also explored, offering new opportunities to understand pathophysiology and identify novel treatment targets. Additionally, the review provides a clinical perspective on patient handling and safety considerations, addressing challenges and practicalities associated with clinical 7T MRI. By bridging research and clinical practice, 7T MRI has the potential to redefine neuroimaging and advance the understanding and management of complex neurological disorders.

Prospective Evaluation of Accelerated Brain MRI Using Deep Learning-Based Reconstruction: Simultaneous Application to 2D Spin-Echo and 3D Gradient-Echo Sequences.

To prospectively evaluate the effect of accelerated deep learning-based reconstruction (Accel-DL) on improving brain magnetic resonance imaging (MRI) quality and reducing scan time compared to that in conventional MRI. This study included 150 participants (51 male; mean age 57.3 ± 16.2 years). Each group of 50 participants was scanned using one of three 3T scanners from three different vendors. Conventional and Accel-DL MRI images were obtained from each participant and compared using 2D T1- and T2-weighted and 3D gradient-echo sequences. Accel-DL acquisition was achieved using optimized scan parameters to reduce the scan time, with the acquired images reconstructed using U-Net-based software to transform low-quality, undersampled k-space data into high-quality images. The scan times of Accel-DL and conventional MRI methods were compared. Four neuroradiologists assessed the overall image quality, structural delineation, and artifacts using Likert scale (5- and 3-point scales). Inter-reader agreement was assessed using Fleiss' kappa coefficient. Signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were calculated, and volumetric quantification of regional structures and white matter hyperintensities (WMHs) was performed. Accel-DL showed a mean scan time reduction of 39.4% (range, 24.2%-51.3%). Accel-DL improved overall image quality (3.78 ± 0.71 vs. 3.36 ± 0.61, P < 0.001), structure delineation (2.47 ± 0.61 vs. 2.35 ± 0.62, P < 0.001), and artifacts (3.73 ± 0.72 vs. 3.71 ± 0.69, P = 0.016). Inter-reader agreement was fair to substantial (κ = 0.34-0.50). SNR and CNR increased in Accel-DL (82.0 ± 23.1 vs. 31.4 ± 10.8, P = 0.02; 12.4 ± 4.1 vs. 4.4 ± 11.2, P = 0.02). Bland-Altman plots revealed no significant differences in the volumetric measurements of 98.2% of the relevant regions, except in the deep gray matter, including the thalamus. Five of the six lesion categories showed no significant differences in WMH segmentation, except for leukocortical lesions (r = 0.64 ± 0.29). Accel-DL substantially reduced the scan time and improved the quality of brain MRI in both spin-echo and gradient-echo sequences without compromising volumetry, including lesion quantification.

Transforming traffic accident investigations: a virtual-real-fusion framework for intelligent 3D traffic accident reconstruction

The daily occurrence of traffic accidents has led to the development of 3D reconstruction as a key tool for reconstruction, investigation, and insurance claims. This study proposes a novel virtual-real-fusion simulation framework that integrates traffic accident generation, unmanned aerial vehicle (UAV)-based image collection, and a 3D traffic accident reconstruction pipeline with advanced computer vision techniques and unsupervised 3D point cloud clustering algorithms. Specifically, a micro-traffic simulator and an autonomous driving simulator are co-simulated to generate high-fidelity traffic accidents. Subsequently, a deep learning-based reconstruction method, i.e., 3D Gaussian splatting (3D-GS), is utilized to construct 3D digitized traffic accident scenes from UAV-based image datasets collected in the traffic simulation environment. While visual rendering by 3D-GS struggles under adverse conditions like nighttime or rain, a clustering parameter stochastic optimization model and mixed-integer programming Bayesian optimization (MIPBO) algorithm are proposed to enhance the segmentation of large-scale 3D point clouds. In the numerical experiments, 3D-GS produces high-quality, seamless, and real-time rendered traffic accident scenes achieve a structural similarity index measure of up to 0.90 across different towns. Furthermore, the proposed MIPDBO algorithm exhibits a remarkably fast convergence rate, requiring only 3–5 iterations to identify well-performing parameters and achieve a high R2 value of 0.8 on a benchmark cluster problem. Finally, the Gaussian Mixture Model assisted by MIPBO accurately separates various traffic elements in the accident scenes, demonstrating higher effectiveness compared to other classical clustering algorithms.

Multidisciplinary quantitative and qualitative assessment of IDH-mutant gliomas with full diagnostic deep learning image reconstruction

Rationale and Objectives: Diagnostic accuracy and therapeutic decision-making for IDH-mutant gliomas in tumor board reviews are based on MRI and multidisciplinary interactions. Materials and Methods:This study explores the feasibility of deep learning-based reconstruction (DLR) in MRI for IDH-mutant gliomas. The research utilizes a multidisciplinary approach, engaging neuroradiologists, neurosurgeons, neuro-oncologists, and radiotherapists to evaluate qualitative aspects of DLR and conventional reconstructed (CR) sequences. Furthermore, quantitative image quality and tumor volumes according to Response Assessment in Neuro-Oncology (RANO) 2.0 standards were assessed. ResultsAll DLR sequences consistently outperformed CR sequences (median of 4 for all) in qualitative image quality across all raters (p < 0.001 for all) and revealed higher SNR and CNR values (p < 0.001 for all). Preference for all DLR over CR was overwhelming, with ratings of 84 % from the neuroradiologist, 100 % from the neurosurgeon, 92 % from the neuro-oncologist, and 84 % from the radiation oncologist. The RANO 2.0 compliant measurements showed no significant difference between the CR and DRL sequences (p = 0.142). ConclusionThis study demonstrates the clinical feasibility of DLR in MR imaging of IDH-mutant gliomas, with significant time savings of 29.6 % on average and non-inferior image quality to CR. DLR sequences received strong multidisciplinary preference, underscoring their potential for enhancing neuro-oncological decision-making and suitability for clinical implementation.

Deep learning-based reconstruction: a reliability assessment in preoperative magnetic resonance imaging for primary rectal cancer.

Deep learning has developed rapidly, and deep learning reconstruction (DLR) methods in magnetic resonance imaging (MRI) are gaining attention for their potential to improve efficacy in clinical work. The preoperative MRI assessment of rectal cancer is crucial for patient management, but the imaging quality is currently limited by a number of factors. DLR could be applied to the preoperative MRI assessment of primary rectal cancer, but research about its specific reliability is limited. Thus, this study aimed to evaluate the reliability of DLR in the preoperative MRI examination of primary rectal cancer. This cross-sectional study was conducted at Ruijin Hospital, Shanghai Jiaotong University School of Medicine from March 2022 to October 2022. Patients with primary rectal cancer underwent routine MRI scans on a 3.0T magnetic resonance scanner (SIGNA Architect, GE Healthcare, USA) with 32-channels flexible coil with conventional reconstruction (ConR) and DLR. The DLR method had three noise reduction levels: DLR-H: 75% noise reduction reconstruction; DLR-M: 50% noise reduction reconstruction; and DLR-L: 25% noise reduction reconstruction. Three components were evaluated: objective image quality; subjective image quality; and diagnostic performance. The objective image quality assessment included the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR). The subjective image quality assessment involved evaluating five subjective image quality parameters based on a 4-point Likert scale. The diagnostic performance assessment included tumour (T) staging, node (N) staging, as well as the circumferential resection margin and extramural vascular invasion evaluation. The images were evaluated in a blinded manner by two radiologists with different levels of experience. The paired sample Wilcoxon signed-rank test, Kappa test, interclass correlation coefficient, Chi-square test, Friedman test, and weighted kappa coefficients were used for the statistical analysis. In total, 61 patients (mean age: 65±12 years; 38 men) were enrolled in the study. The DLR method improved the SNR and CNR values of the images relative to the ConR method, while the DLR-H produced the greatest improvement (P<0.040). The subjective image quality of the DLR-H images was superior to that of the ConR images (P<0.001), but there was no significant difference between the DLR-H and DLR-M images (P≥0.075). The evaluators showed good agreement in subjective scoring, and in the DLR image scoring, the evaluators have the best consistency in the DLR-H images scoring (kappa =0.921, P<0.001). The diagnostic efficacy of the DLR images was comparable to that of the ConR images in terms of T staging [Reader 1 (R1): P=0.603; Reader 2 (R2): P=0.206] and N staging (R1: P=0.990; R2: P=0.884). The DLR method improved the quality of the images, and had comparable diagnostic efficacy without additional scanning time to that of the ConR method, and thus could be a feasible option for replacing the ConR method in the preoperative MRI examination of primary rectal cancer.

TD-STrans: Tri-domain sparse-view CT reconstruction based on sparse transformer.

Sparse-view computed tomography (CT) speeds up scanning and reduces radiation exposure in medical diagnosis. However, when the projection views are severely under-sampled, deep learning-based reconstruction methods often suffer from over-smoothing of the reconstructed images due to the lack of high-frequency information. To address this issue, we introduce frequency domain information into the popular projection-image domain reconstruction, proposing a Tri-Domain sparse-view CT reconstruction model based on Sparse Transformer (TD-STrans). TD-STrans integrates three essential modules: the projection recovery module completes the sparse-view projection, the Fourier domain filling module mitigates artifacts and over-smoothing by filling in missing high-frequency details; the image refinement module further enhances and preserves image details. Additionally, a multi-domain joint loss function is designed to simultaneously enhance the reconstruction quality in the projection domain, image domain, and frequency domain, thereby further improving the preservation of image details. The results of simulation experiments on the lymph node dataset and real experiments on the walnut dataset consistently demonstrate the effectiveness of TD-STrans in artifact removal, suppression of over-smoothing, and preservation of structural fidelity. The reconstruction results of TD-STrans indicate that sparse transformer across multiple domains can alleviate over-smoothing and detail loss caused by reduced views, offering a novel solution for ultra-sparse-view CT imaging.

Contrast-enhanced thin-slice abdominal CT with super-resolution deep learning reconstruction technique: evaluation of image quality and visibility of anatomical structures.

To compare image quality and visibility of anatomical structures on contrast-enhanced thin-slice abdominal CT images reconstructed using super-resolution deep learning reconstruction (SR-DLR), deep learning-based reconstruction (DLR), and hybrid iterative reconstruction (HIR) algorithms. This retrospective study included 54 consecutive patients who underwent contrast-enhanced abdominal CT. Thin-slice images (0.5mm thickness) were reconstructed using SR-DLR, DLR, and HIR. Objective image noise and contrast-to-noise ratio (CNR) for liver parenchyma relative to muscle were assessed. Two radiologists independently graded image quality using a 5-point rating scale for image noise, sharpness, artifact/blur, and overall image quality. They also graded the visibility of small vessels, main pancreatic duct, ureters, adrenal glands, and right adrenal vein on a 5-point scale. SR-DLR yielded significantly lower objective image noise and higher CNR than DLR and HIR (P < .001). The visual scores of SR-DLR for image noise, sharpness, and overall image quality were significantly higher than those of DLR and HIR for both readers (P < .001). Both readers scored significantly higher on SR-DLR than on HIR for visibility for all structures (P < .01), and at least one reader scored significantly higher on SR-DLR than on DLR for visibility for all structures (P < .05). SR-DLR reduced image noise and improved image quality of thin-slice abdominal CT images compared to HIR and DLR. This technique is expected to enable further detailed evaluation of small structures.

Accelerating Brain MR Imaging With Multisequence and Convolutional Neural Networks.

Magnetic resonance imaging (MRI) refers to one of the critical image modalities for diagnosis, whereas its long acquisition time limits its application. In this study, the aim was to investigate whether deep learning-based techniques are capable of using the common information in different MRI sequences to reduce the scan time of the most time-consuming sequences while maintaining the image quality. Fully sampled T1-FLAIR, T2-FLAIR, and T2WI brain MRI raw data originated from 217 patients and 105 healthy subjects. The T1-FLAIR and T2-FLAIR sequences were subsampled using Cartesian masks based on four different acceleration factors. The fully sampled T1/T2-FLAIR images were predicted from undersampled T1/T2-FLAIR images and T2WI images through deep learning-based reconstruction. They were qualitatively assessed by two senior radiologists in accordance with the diagnosis decision and a four-point scale image quality score. Furthermore, the images were quantitatively assessed based on regional signal-to-noise ratios (SNRs) and contrast-to-noise ratios (CNRs). The chi-square test was performed, where p<0.05 indicated a difference with statistical significance. The diagnosis decisions from two senior radiologists remained unchanged in accordance with the accelerated and fully sampled images. There were no significant differences in the regional SNRs and CNRs of most assessed regions (p>0.05) between the accelerated and fully sampled images. Moreover, no significant difference was identified in the image quality assessed by two senior radiologists (p>0.05). Deep learning-based image reconstruction is capable of significantly expediting the brain MR imaging process and producing acceptable image quality without affecting diagnosis decisions.

Exploring the feasibility of FOCUS DWI with deep learning reconstruction for breast cancer diagnosis: A comparative study with conventional DWI.

This study compared field-of-view (FOV) optimized and constrained undistorted single-shot diffusion-weighted imaging (FOCUS DWI) with deep-learning-based reconstruction (DLR) to conventional DWI for breast imaging. This study prospectively enrolled 49 female patients suspected of breast cancer from July to December 2023. The patients underwent conventional and FOCUS breast DWI and data were reconstructed with and without DLR. Two radiologists independently evaluated three images per patient using a 5-point Likert scale. Objective evaluations, including signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and apparent diffusion coefficient (ADC), were conducted using manual region of interest-based analysis. The subjective and objective evaluations were compared using the Friedman test. The scores for the overall image quality, anatomical details, lesion conspicuity, artifacts, and distortion in FOCUS-DLR DWI were higher than in conventional DWI (all P < 0.001). The SNR of FOCUS-DLR DWI was higher than that of conventional and FOCUS DWI (both P < 0.001), while FOCUS and conventional DWI were similar (P = 0.096). Conventional, FOCUS, and FOCUS-DLR DWI had similar CNR and ADC values. Our findings indicate that images produced by FOCUS-DLR DWI were superior to conventional DWI, supporting the applicability of this technique in clinical practice. DLR provides a new approach to optimize breast DWI.

Deep Learning-Based Reconstruction of 3D Morphology of Geomaterial Particles from Single-View 2D Images.

The morphology of particles formed in different environments contains critical information. Thus, the rapid and effective reconstruction of their three-dimensional (3D) morphology is crucial. This study reconstructs the 3D morphology from two-dimensional (2D) images of particles using artificial intelligence (AI). More than 100,000 particles were sampled from three sources: naturally formed particles (desert sand), manufactured particles (lunar soil simulant), and numerically generated digital particles. A deep learning approach based on a voxel representation of the morphology and multi-dimensional convolutional neural networks was proposed to rapidly upscale and reconstruct particle morphology. The trained model was tested using the three particle types and evaluated using different multi-scale morphological descriptors. The results demonstrated that the statistical properties of the morphological descriptors were consistent for the real 3D particles and those derived from the 2D images and the model. This finding confirms the model's validity and generalizability in upscaling and reconstructing diverse particle samples. This study provides a method for generating 3D numerical representations of geological particles, facilitating in-depth analysis of properties, such as mechanical behavior and transport characteristics, from 2D images.

A dynamic approach for MR T2-weighted pelvic imaging

T2-weighted 2D fast spin echo sequence serves as the standard sequence in&#xD;clinical pelvic MR imaging protocols. However, motion artifacts and blurring caused&#xD;by peristalsis present significant challenges. Patient preparation such as administering&#xD;antiperistaltic agents is often required before examination to reduce artifacts, which&#xD;discomfort the patients. This work introduce a novel dynamic approach for T2&#xD;weighted pelvic imaging to address peristalsis-induced motion issue without any patient&#xD;preparation.&#xD;Approach: A rapid dynamic data acquisition strategy with complementary sampling&#xD;trajectory is designed to enable highly undersampled motion-resistant data sampling,&#xD;and an unrolling method based on deep equilibrium model is leveraged to reconstruct&#xD;images from the dynamic sampled k-space data. Moreover, the fix-point convergence of&#xD;the equilibrium model ensures the stability of the reconstruction. The high acceleration&#xD;factor in each temporal phase, which is much higher than that in traditional static&#xD;imaging, has the potential to effectively freeze pelvic motion, thereby transforming&#xD;the imaging problem from conventional motion prevention or removal to motion&#xD;reconstruction.&#xD;Main results: Experiments on both retrospective and prospective data have&#xD;demonstrated the superior performance of the proposed dynamic approach in reducing&#xD;motion artifacts and accurately depicting structural details compared to standard static&#xD;imaging.&#xD;Significance: The proposed dynamic approach effectively captures motion states&#xD;through dynamic data acquisition and deep learning-based reconstruction, addressing&#xD;motion-related challenges in pelvic imaging.

Enhanced image quality and lesion detection in FLAIR MRI of white matter hyperintensity through deep learning-based reconstruction

ObjectiveTo delve deeper into the study of degenerative diseases, it becomes imperative to investigate whether deep-learning reconstruction (DLR) can improve the evaluation of white matter hyperintensity (WMH) on 3.0T scanners, and compare its lesion detection capabilities with conventional reconstruction (CR). MethodsA total of 131 participants (mean age, 46 years ±17; 46 men) were included in the study. The images of these participants were evaluated by readers blinded to clinical data. Two readers independently assessed subjective image indicators on a 4-point scale. The severity of WMH was assessed by four raters using the Fazekas scale. To evaluate the relative detection capabilities of each method, we employed the Wilcoxon signed rank test to compare scores between the DLR and the CR group. Additionally, we assessed interrater reliability using weighted k statistics and intraclass correlation coefficient to test consistency among the raters. ResultsIn terms of subjective image scoring, the DLR group exhibited significantly better scores compared to the CR group (P < 0.001). Regarding the severity of WMH, the DL group demonstrated superior performance in detecting lesions. Majority readers agreed that the DL group provided clearer visualization of the lesions compared to the conventional group. ConclusionDLR exhibits notable advantages over CR, including subjective image quality, lesion detection sensitivity, and inter reader reliability.

Reliability of brain volume measures of accelerated 3D T1-weighted images with deep learning-based reconstruction.

The time-intensive nature of acquiring 3D T1-weighted MRI and analyzing brain volumetry limits quantitative evaluation of brain atrophy. We explore the feasibility and reliability of deep learning-based accelerated MRI scans for brain volumetry. This retrospective study collected 3D T1-weighted data using 3T from 42 participants for the simulated acceleration dataset and 48 for the validation dataset. The simulated acceleration dataset consists of three sets at different simulated acceleration levels (Simul-Accel) corresponding to level 1 (65% undersampling), 2 (70%), and 3 (75%). These images were then subjected to deep learning-based reconstruction (Simul-Accel-DL). Conventional images (Conv) without acceleration and DL were set as the reference. In the validation dataset, DICOM images were collected from Conv and accelerated scan with DL-based reconstruction (Accel-DL). The image quality of Simul-Accel-DL was evaluated using quantitative error metrics. Volumetric measurements were evaluated using intraclass correlation coefficients (ICCs) and linear regression analysis in both datasets. The volumes were estimated by two software, NeuroQuant and DeepBrain. Simul-Accel-DL across all acceleration levels revealed comparable or better error metrics than Simul-Accel. In the simulated acceleration dataset, ICCs between Conv and Simul-Accel-DL in all ROIs exceeded 0.90 for volumes and 0.77 for normative percentiles at all acceleration levels. In the validation dataset, ICCs for volumes > 0.96, ICCs for normative percentiles > 0.89, and R2 > 0.93 at all ROIs except pallidum demonstrated good agreement in both software. DL-based reconstruction achieves clinical feasibility of 3D T1 brain volumetric MRI by up to 75% acceleration relative to full-sampled acquisition.

Deep learning-based reconstruction improves the image quality of low-dose CT enterography in patients with inflammatory bowel disease.

Lifelong re-examination of CT enterography (CTE) in patients with inflammatory bowel disease (IBD) may be necessary, and reducing radiation exposure during CT examinations is crucial. We investigated the potential application of deep learning reconstruction (DLR) in CTE to reduce radiation dose and improve image quality in IBD. Thirty-six patients with known or suspected IBD were prospectively recruited to the low-dose CTE (LDCTE) group, while forty patients were retrospectively selected from previous clinical standard-dose CTE (STDCTE) scans as controls. STDCTE images were reconstructed with hybrid-IR (adaptive iterative dose reduction 3-dimensional [AIDR3D], standard setting); LDCTE images were reconstructed with AIDR3D and DLR (Advanced Intelligence ClearIQ Engine [AiCE], Body mild/standard/strong, Sharp Body mild/standard/strong setting). The effective radiation dose (ED), image noise, signal-to-noise ratio (SNR), overall image quality, subjective image noise, and diagnostic effectiveness were compared between the LDCTE and STDCTE groups. Compared with STDCTE, the ED of LDCTE was lower by 54.1% (p<0.001). Compared with STDCTE-AIDR3D, LDCTE-AIDR3D reconstruction objective image noise and SNR were greater (p<0.05), the subjective overall image quality was lower (p<0.05), and the diagnostic efficiency was lower (AUC=0.52, p<0.05). The SNRs of reconstructedimages of LDCTE-AiCE Body Strong and LDCTE-AiCE Body Sharp standard/strong groups were greater than that of STDCTE-AIDR3D group (all p<0.05), and the diagnostic performance was better than or comparable to that of STDCTE; the AUCs were 0.83, 0.76 and 0.76, respectively CONCLUSION: Compared with STDCTE with AIDR3D, LDCTE with DLR effectively reduced the radiation dose and improve image quality in IBD patients.

Clinical feasibility of deep learning–accelerated single-shot turbo spin echo sequence with enhanced denoising for pancreas MRI at 3 Tesla

PurposeTo assess the feasibility of the single-shot turbo spin echo sequence using deep learning-based reconstruction (DLR) (HASTEDL) with enhanced denoising for pancreas MRI. MethodsPatients who underwent pancreas MRI from March to April 2021 were included. Four T2-weighted images (non-FS conventional HASTE vs. HASTEDL with enhanced denoising and FS HASTEDL with enhanced denoising vs. HASTEDL) were acquired. Two readers independently assessed the image quality parameters of the two non-FS image sets using a 4-point Likert scale. The signal-to-noise ratio (SNR) of the cystic lesions and pancreatic parenchyma and the contrast-to-noise ratio between the cystic lesion and pancreatic parenchyma were calculated for all four image sets. The size of the largest cystic lesion and the diameter of pancreatic duct were measured. ResultsA total of 63 patients were included, 48 (76.2 %) of whom had 136 pancreatic cystic lesion(s). The acquisition times of conventional HASTE and HASTEDL were 69 and 18 sec, respectively. All image quality parameters except artifacts for reader 2 were significantly better for HASTEDL with enhanced denoising. Those images also received scores for overall image quality that were significantly higher than those for the conventional HASTE (3.26 ± 0.54 vs. 2.47 ± 0.56, p < 0.001). The SNR of the pancreatic cystic lesion and pancreatic parenchyma was significantly higher in the HASTEDL with enhanced denoising (p < 0.001 for both). Inter-reader variability for measuring the pancreatic cyst size (ICC, 0.999 vs. 0.995; 95 % LoA, −0.13481 to 0.14743 vs. −0.24097 to 0.27404) and duct diameter (ICC, 0.994 vs. 0.969; 95 % LoA, −0.11684 to 0.36026 vs. −0.45544 to 0.44664) was lower in HASTEDL with enhanced denoising than in the conventional HASTE. ConclusionHASTEDL with enhanced denoising could be useful for reducing the acquisition time of pancreas MRI while improving the image quality for the evaluation of pancreatic cystic lesions.

High-resolution in vivo 4D-OCT fish-eye imaging using 3D-UNet with multi-level residue decoder.

Optical coherence tomography (OCT) allows high-resolution volumetric imaging of biological tissues in vivo. However, 3D-image acquisition often suffers from motion artifacts due to slow frame rates and involuntary and physiological movements of living tissue. To solve these issues, we implement a real-time 4D-OCT system capable of reconstructing near-distortion-free volumetric images based on a deep learning-based reconstruction algorithm. The system initially collects undersampled volumetric images at a high speed and then upsamples the images in real-time by a convolutional neural network (CNN) that generates high-frequency features using a deep learning algorithm. We compare and analyze both dual-2D- and 3D-UNet-based networks for the OCT 3D high-resolution image reconstruction. We refine the network architecture by incorporating multi-level information to accelerate convergence and improve accuracy. The network is optimized by utilizing the 16-bit floating-point precision for network parameters to conserve GPU memory and enhance efficiency. The result shows that the refined and optimized 3D-network is capable of retrieving the tissue structure more precisely and enable real-time 4D-OCT imaging at a rate greater than 10 Hz with a root mean square error (RMSE) of ∼0.03.

CineVN: Variational network reconstruction for rapid functional cardiac cine MRI.

To develop a reconstruction method for highly accelerated cardiac cine MRI with high spatiotemporal resolution and low temporal blurring, and to demonstrate accurate estimation of ventricular volumes and myocardial strain in healthy subjects and in patients. The proposed method, called CineVN, employs a spatiotemporal Variational Network combined with conjugate gradient descent for optimized data consistency and improved image quality. The method is first evaluated on retrospectively undersampled cine MRI data in terms of image quality. Then, prospectively accelerated data are acquired in 18 healthy subjects both segmented over two heartbeats per slice as well as in real time with 1.6 mm isotropic resolution. Ventricular volumes and strain parameters are computed and compared to a compressed sensing reconstruction and to a conventional reference cine MRI acquisition. Lastly, the method is demonstrated in 46 patients and ventricular volumes and strain parameters are evaluated. CineVN outperformed compressed sensing in image quality metrics on retrospectively undersampled data. Functional parameters and myocardial strain were the most accurate for CineVN compared to two state-of-the-art compressed sensing methods. Deep learning-based reconstruction using our proposed method enables accurate evaluation of cardiac function in real-time cine MRI with high spatiotemporal resolution. This has the potential to improve cardiac imaging particularly for patients with arrhythmia or impaired breath-hold capability.