Deep Learning-based Image Reconstruction Research Articles

Accelerated Diffusion-Weighted Magnetic Resonance Imaging of the Liver at 1.5 T With Deep Learning-Based Image Reconstruction: Impact on Image Quality and Lesion Detection.

To perform image quality comparison between deep learning-based multiband diffusion-weighted sequence (DL-mb-DWI), accelerated multiband diffusion-weighted sequence (accelerated mb-DWI), and conventional multiband diffusion-weighted sequence (conventional mb-DWI) in patients undergoing clinical liver magnetic resonance imaging (MRI). Fifty consecutive patients who underwent clinical MRI of the liver at a 1.5-T scanner, between September 1, 2021, and January 31, 2022, were included in this study. Three radiologists independently reviewed images using a 5-point Likert scale for artifacts and image quality factors, in addition to assessing the presence of liver lesions and lesion conspicuity. DL-mb-DWI acquisition time was 65.0 ± 2.4 seconds, significantly (P < 0.001) shorter than conventional mb-DWI (147.5 ± 19.2 seconds) and accelerated mb-DWI (94.3 ± 1.8 seconds). DL-mb-DWI received significantly higher scores than conventional mb-DWI for conspicuity of the left lobe (P < 0.001), sharpness of intrahepatic vessel margin (P < 0.001), sharpness of the pancreatic contour (P < 0.001), in-plane motion artifact (P = 0.002), and overall image quality (P = 0.005) by reader 2. DL-mb-DWI received significantly higher scores for conspicuity of the left lobe (P = 0.006), sharpness of the pancreatic contour (P = 0.020), and in-plane motion artifact (P = 0.042) by reader 3. DL-mb-DWI received significantly higher scores for strength of fat suppression (P = 0.004) and sharpness of the pancreatic contour (P = 0.038) by reader 1. The remaining quality parameters did not reach statistical significance for reader 1. Novel diffusion-weighted MRI sequence with deep learning-based image reconstruction demonstrated significantly decreased acquisition times compared with conventional and accelerated mb-DWI sequences, while maintaining or improving image quality for routine abdominal MRI. DL-mb-DWI offers a potential alternative to conventional mb-DWI in routine clinical liver MRI.

Advanced deep learning-based image reconstruction in lumbar spine MRI at 0.55 T – Effects on image quality and acquisition time in comparison to conventional deep learning-based reconstruction

ObjectivesTo evaluate an optimized deep leaning-based image post-processing technique in lumbar spine MRI at 0.55 T in terms of image quality and image acquisition time. Materials and methodsLumbar spine imaging was conducted on 18 patients using a 0.55 T MRI scanner, employing conventional (CDLR) and advanced (ADLR) deep learning-based post-processing techniques. Two musculoskeletal radiologists visually evaluated the images using a 5-point Likert scale to assess image quality and resolution. Quantitative assessment in terms of signal intensities (SI) and contrast ratios was performed by region of interest measurements in different body-tissues (vertebral bone, intervertebral disc, spinal cord, cerebrospinal fluid and autochthonous back muscles) to investigate differences between CDLR and ADLR sequences. ResultsThe images processed with the advanced technique (ADLR) were rated superior to the conventional technique (CDLR) in terms of signal/contrast, resolution, and assessability of the spinal canal and neural foramen. The interrater agreement was moderate for signal/contrast (ICC = 0.68) and good for resolution (ICC = 0.77), but moderate for spinal canal and neuroforaminal assessability (ICC = 0.55). Quantitative assessment showed a higher contrast ratio for fluid-sensitive sequences in the ADLR images. The use of ADLR reduced image acquisition time by 44.4%, from 14:22 min to 07:59 min. ConclusionsAdvanced deep learning-based image reconstruction algorithms improve the visually perceived image quality in lumbar spine imaging at 0.55 T while simultaneously allowing to substantially decrease image acquisition times. Clinical relevanceAdvanced deep learning-based image post-processing techniques (ADLR) in lumbar spine MRI at 0.55 T significantly improves image quality while reducing image acquisition time.

Res-U2Net: untrained deep learning for phase retrieval and image reconstruction

Conventional deep learning-based image reconstruction methods require a large amount of training data, which can be hard to obtain in practice. Untrained deep learning methods overcome this limitation by training a network to invert a physical model of the image formation process. Here we present a novel, to our knowledge, untrained Res-U2Net model for phase retrieval. We use the extracted phase information to determine changes in an object’s surface and generate a mesh representation of its 3D structure. We compare the performance of Res-U2Net phase retrieval against UNet and U2Net using images from the GDXRAY dataset.

Evaluation of data uncertainty for deep-learning-based CT noise reduction using ensemble patient data and a virtual imaging trial framework.

Deep learning-based image reconstruction and noise reduction (DLIR) methods have been increasingly deployed in clinical CT. Accurate assessment of their data uncertainty properties is essential to understand the stability of DLIR in response to noise. In this work, we aim to evaluate the data uncertainty of a DLIR method using real patient data and a virtual imaging trial framework and compare it with filtered-backprojection (FBP) and iterative reconstruction (IR). The ensemble of noise realizations was generated by using a realistic projection domain noise insertion technique. The impact of varying dose levels and denoising strengths were investigated for a ResNet-based deep convolutional neural network (DCNN) model trained using patient images. On the uncertainty maps, DCNN shows more detailed structures than IR although its bias map has less structural dependency, which implies that DCNN is more sensitive to small changes in the input. Both visual examples and histogram analysis demonstrated that hotspots of uncertainty in DCNN may be associated with a higher chance of distortion from the truth than IR, but it may also correspond to a better detection performance for some of the small structures.

Comparison of model-based versus deep learning-based image reconstruction for thin-slice T2-weighted spin-echo prostate MRI.

To compare a previous model-based image reconstruction (MBIR) with a newly developed deep learning (DL)-based image reconstruction for providing improved signal-to-noise ratio (SNR) in high through-plane resolution (1mm) T2-weighted spin-echo (T2SE) prostate MRI. Large-area contrast and high-contrast spatial resolution of the reconstruction methods were assessed quantitatively in experimental phantom studies. The methods were next evaluated radiologically in 17 subjects at 3.0 Tesla for whom prostate MRI was clinically indicated. For each subject, the axial T2SE raw data were directed to MBIR and to the DL reconstruction at three vendor-provided levels: (L)ow, (M)edium, and (H)igh. Thin-slice images from the four reconstructions were compared using evaluation criteria related to SNR, sharpness, contrast fidelity, and reviewer preference. Results were compared using the Wilcoxon signed-rank test using Bonferroni correction, and inter-reader comparisons were done using the Cohen and Krippendorf tests. Baseline contrast and resolution in phantom studies were equivalent for all four reconstruction pathways as desired. In vivo, all three DL levels (L, M, H) provided improved SNR versus MBIR. For virtually, all other evaluation criteria DL L and M were superior to MBIR. DL L and M were evaluated as superior to DL H in fidelity of contrast. For 44 of the 51 evaluations, the DL M reconstruction was preferred. The deep learning reconstruction method provides significant SNR improvement in thin-slice (1mm) T2SE images of the prostate while retaining image contrast. However, if taken to too high a level (DL High), both radiological sharpness and fidelity of contrast diminish.

Deep learning-based image reconstruction for the multi-arterial phase images: improvement of the image quality to assess the small hypervascular hepatic tumor on gadoxetic acid-enhanced liver MRI.

To evaluated the impact of a deep learning (DL)-based image reconstruction on multi-arterial-phase magnetic resonance imaging (MA-MRI) for small hypervascular hepatic masses in patients who underwent gadoxetic acid-enhanced liver MRI. We retrospectively enrolled 55 adult patients (aged ≥ 18years) with small hepatic hypervascular mass (≤ 3cm) between December 2022 and February 2023. All patients underwent MA-MRI, subsequently reconstructed with a DL-based application. Qualitative assessment with Linkert scale including motion artifact (MA), liver edge (LE), hepatic vessel clarity (HVC) and image quality (IQ) was performed. Quantitative image analysis including signal to noise ratio (SNR), contrast to noise ratio (CNR) and noise was performed. On both arterial phases (APs), all qualitative parameters were significantly improved after DL-based image reconstruction. (LE on 1st AP, 1.22 vs 1.61; LE on 2nd AP, 1.21 vs 1.65; HVC on 1st AP, 1.24 vs 1.39; HVC on 2nd AP, 1.24 vs 1.44; IQ on 1st AP, 1.17 vs 1.45; IQ on 2nd AP, 1.17 vs 1.47, all p values < 0.05). The SNR, CNR and noise were significantly improved after DL-based image reconstruction. (SNR on AP1, 279.08 vs 176.14; SNR on AP2, 334.34 vs 199.24; CNR on AP1, 106.09 vs 64.14; CNR on AP2, 129.66 vs 73.73; noise on AP1, 1.51 vs 2.33; noise on AP2, 1.45 vs 2.28, all p values < 0.05). Gadoxetic acid-enhanced MA-MRI with DL-based image reconstruction improved the qualitative and quantitative parameters. Despite the short acquisition time, high-quality MA-MRI is now achievable.

Modern acceleration in musculoskeletal MRI: applications, implications, and challenges.

Magnetic resonance imaging (MRI) is crucial for accurately diagnosing a wide spectrum of musculoskeletal conditions due to its superior soft tissue contrast resolution. However, the long acquisition times of traditional two-dimensional (2D) and three-dimensional (3D) fast and turbo spin-echo (TSE) pulse sequences can limit patient access and comfort. Recent technical advancements have introduced acceleration techniques that significantly reduce MRI times for musculoskeletal examinations. Key acceleration methods include parallel imaging (PI), simultaneous multi-slice acquisition (SMS), and compressed sensing (CS), enabling up to eightfold faster scans while maintaining image quality, resolution, and safety standards. These innovations now allow for 3- to 6-fold accelerated clinical musculoskeletal MRI exams, reducing scan times to 4 to 6min for joints and spine imaging. Evolving deep learning-based image reconstruction promises even faster scans without compromising quality. Current research indicates that combining acceleration techniques, deep learning image reconstruction, and superresolution algorithms will eventually facilitate tenfold accelerated musculoskeletal MRI in routine clinical practice. Such rapid MRI protocols can drastically reduce scan times by 80-90% compared to conventional methods. Implementing these rapid imaging protocols does impact workflow, indirect costs, and workload for MRI technologists and radiologists, which requires careful management. However, the shift from conventional to accelerated, deep learning-based MRI enhances the value of musculoskeletal MRI by improving patient access and comfort and promoting sustainable imaging practices. This article offers a comprehensive overview of the technical aspects, benefits, and challenges of modern accelerated musculoskeletal MRI, guiding radiologists and researchers in this evolving field.

CGAN-ECT: Reconstruction of Electrical Capacitance Tomography images from capacitance measurements using Conditional Generative Adversarial Networks

Due to the rapid growth of applications of Electrical Capacitance Tomography (ECT) across diverse industrial sectors, there is a pressing demand for swift and high-fidelity methodologies for image reconstruction from raw capacitance measurements. Deep learning, as an effective non-linear mapping tool for complicated functions, has been going viral in many fields including electrical tomography. This paper introduces a pioneering contribution by proposing a Conditional Generative Adversarial Network (CGAN) model explicitly tailored for the purpose of reconstructing ECT images from capacitance measurements. We propose a novel modulator to transform the raw input capacitance measurements into a 2D image. To the best of our knowledge, this is the first time to represent the capacitance measurements in an image form. We have created a new massive ECT dataset of synthetic image-measurements pairs for training and testing purposes of the proposed model. This extensive ECT dataset comprises 320,000 synthetic image-measurement pairs. It forms the basis for evaluating the feasibility and generalizability of the CGAN-ECT model, with assessments conducted on testing data, contaminated data, and flow patterns which are intentionally excluded from the model’s training phase. The evaluation results prove the efficacy of the proposed CGAN model in generating ECT images with heightened accuracy compared to traditional methods and alternative deep learning-based image reconstruction algorithms. The CGAN-ECT model has succeeded in a achieving an average image correlation coefficient exceeding 99.3%, coupled with an average relative image error of approximately 0.07.

Enhanced visualization in endoleak detection through iterative and AI-noise optimized spectral reconstructions

To assess the image quality parameters of dual-energy computed tomography angiography (DECTA) 40-, and 60 keV virtual monoenergetic images (VMIs) combined with deep learning-based image reconstruction model (DLM) and iterative reconstructions (IR). CT scans of 28 post EVAR patients were enrolled. The 60 s delayed phase of DECTA was evaluated. Objective [noise, contrast-to-noise ratio (CNR), signal-to-noise ratio (SNR)] and subjective (overall image quality and endoleak conspicuity – 3 blinded readers assessment) image quality analyses were performed. The following reconstructions were evaluated: VMI 40, 60 keV VMI; IR VMI 40, 60 keV; DLM VMI 40, 60 keV. The noise level of the DLM VMI images was approximately 50% lower than that of VMI reconstruction. The highest CNR and SNR values were measured in VMI DLM images. The mean CNR in endoleak in 40 keV was accounted for as 1.83 ± 1.2; 2.07 ± 2.02; 3.6 ± 3.26 in VMI, VMI IR, and VMI DLM, respectively. The DLM algorithm significantly reduced noise and increased lesion conspicuity, resulting in higher objective and subjective image quality compared to other reconstruction techniques. The application of DLM algorithms to low-energy VMIs significantly enhances the diagnostic value of DECTA in evaluating endoleaks. DLM reconstructions surpass traditional VMIs and IR in terms of image quality.

Deep learning-accelerated image reconstruction in back pain-MRI imaging: reduction of acquisition time and improvement of image quality

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

Improved image quality in CT pulmonary angiography using deep learning-based image reconstruction

We investigated the effect of deep learning-based image reconstruction (DLIR) compared to iterative reconstruction on image quality in CT pulmonary angiography (CTPA) for suspected pulmonary embolism (PE). For 220 patients with suspected PE, CTPA studies were reconstructed using filtered back projection (FBP), adaptive statistical iterative reconstruction (ASiR-V 30%, 60% and 90%) and DLIR (low, medium and high strength). Contrast-to-noise ratio (CNR) served as the primary parameter of objective image quality. Subgroup analyses were performed for normal weight, overweight and obese individuals. For patients with confirmed PE (n = 40), we further measured PE-specific CNR. Subjective image quality was assessed independently by two experienced radiologists. CNR was lowest for FBP and enhanced with increasing levels of ASiR-V and, even more with increasing strength of DLIR. High strength DLIR resulted in an additional improvement in CNR by 29–67% compared to ASiR-V 90% (p < 0.05). PE-specific CNR increased by 75% compared to ASiR-V 90% (p < 0.05). Subjective image quality was significantly higher for medium and high strength DLIR compared to all other image reconstructions (p < 0.05). In CT pulmonary angiography, DLIR significantly outperforms iterative reconstruction for increasing objective and subjective image quality. This may allow for further reductions in radiation exposure in suspected PE.

Improving Diagnostic Performance of MRI for Temporal Lobe Epilepsy With Deep Learning-Based Image Reconstruction in Patients With Suspected Focal Epilepsy.

To evaluate the diagnostic performance and image quality of 1.5-mm slice thickness MRI with deep learning-based image reconstruction (1.5-mm MRI + DLR) compared to routine 3-mm slice thickness MRI (routine MRI) and 1.5-mm slice thickness MRI without DLR (1.5-mm MRI without DLR) for evaluating temporal lobe epilepsy (TLE). This retrospective study included 117 MR image sets comprising 1.5-mm MRI + DLR, 1.5-mm MRI without DLR, and routine MRI from 117 consecutive patients (mean age, 41 years; 61 female; 34 patients with TLE and 83 without TLE). Two neuroradiologists evaluated the presence of hippocampal or temporal lobe lesions, volume loss, signal abnormalities, loss of internal structure of the hippocampus, and lesion conspicuity in the temporal lobe. Reference standards for TLE were independently constructed by neurologists using clinical and radiological findings. Subjective image quality, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) were analyzed. Performance in diagnosing TLE, lesion findings, and image quality were compared among the three protocols. The pooled sensitivity of 1.5-mm MRI + DLR (91.2%) for diagnosing TLE was higher than that of routine MRI (72.1%, P < 0.001). In the subgroup analysis, 1.5-mm MRI + DLR showed higher sensitivity for hippocampal lesions than routine MRI (92.7% vs. 75.0%, P = 0.001), with improved depiction of hippocampal T2 high signal intensity change (P = 0.016) and loss of internal structure (P < 0.001). However, the pooled specificity of 1.5-mm MRI + DLR (76.5%) was lower than that of routine MRI (89.2%, P = 0.004). Compared with 1.5-mm MRI without DLR, 1.5-mm MRI + DLR resulted in significantly improved pooled accuracy (91.2% vs. 73.1%, P = 0.010), image quality, SNR, and CNR (all, P < 0.001). The use of 1.5-mm MRI + DLR enhanced the performance of MRI in diagnosing TLE, particularly in hippocampal evaluation, because of improved depiction of hippocampal abnormalities and enhanced image quality.

NeRP: Implicit Neural Representation Learning With Prior Embedding for Sparsely Sampled Image Reconstruction.

Image reconstruction is an inverse problem that solves for a computational image based on sampled sensor measurement. Sparsely sampled image reconstruction poses additional challenges due to limited measurements. In this work, we propose a methodology of implicit Neural Representation learning with Prior embedding (NeRP) to reconstruct a computational image from sparsely sampled measurements. The method differs fundamentally from previous deep learning-based image reconstruction approaches in that NeRP exploits the internal information in an image prior and the physics of the sparsely sampled measurements to produce a representation of the unknown subject. No large-scale data is required to train the NeRP except for a prior image and sparsely sampled measurements. In addition, we demonstrate that NeRP is a general methodology that generalizes to different imaging modalities such as computed tomography (CT) and magnetic resonance imaging (MRI). We also show that NeRP can robustly capture the subtle yet significant image changes required for assessing tumor progression.

Kiosk 9R-TC-09 - Retrospective Phase-map Synthesis for CMR Datasets FBom Magnitude-only DICOM Images Enabled by A.I. Generative Models to Create Large Training Datasets for Deep Learning-based Image Reconstruction

Kiosk 9R-TC-09 - Retrospective Phase-map Synthesis for CMR Datasets FBom Magnitude-only DICOM Images Enabled by A.I. Generative Models to Create Large Training Datasets for Deep Learning-based Image Reconstruction

Subspace Model-Assisted Deep Learning for Improved Image Reconstruction.

Image reconstruction from limited and/or sparse data is known to be an ill-posed problem and a priori information/constraints have played an important role in solving the problem. Early constrained image reconstruction methods utilize image priors based on general image properties such as sparsity, low-rank structures, spatial support bound, etc. Recent deep learning-based reconstruction methods promise to produce even higher quality reconstructions by utilizing more specific image priors learned from training data. However, learning high-dimensional image priors requires huge amounts of training data that are currently not available in medical imaging applications. As a result, deep learning-based reconstructions often suffer from two known practical issues: a) sensitivity to data perturbations (e.g., changes in data sampling scheme), and b) limited generalization capability (e.g., biased reconstruction of lesions). This paper proposes a new method to address these issues. The proposed method synergistically integrates model-based and data-driven learning in three key components. The first component uses the linear vector space framework to capture global dependence of image features; the second exploits a deep network to learn the mapping from a linear vector space to a nonlinear manifold; the third is an unrolling-based deep network that captures local residual features with the aid of a sparsity model. The proposed method has been evaluated with magnetic resonance imaging data, demonstrating improved reconstruction in the presence of data perturbation and/or novel image features. The method may enhance the practical utility of deep learning-based image reconstruction.

Insights about cervical lymph nodes: Evaluating deep learning–based reconstruction for head and neck computed tomography scan

PurposeThis study aimed to investigate differences in cervical lymph node image quality on dual-energy computed tomography (CT) scan with datasets reconstructed using filter back projection (FBP), hybrid iterative reconstruction (IR), and deep learning–based image reconstruction (DLIR) in patients with head and neck cancer. MethodSeventy patients with head and neck cancer underwent follow-up contrast-enhanced dual-energy CT examinations. All datasets were reconstructed using FBP, hybrid IR with 30 % adaptive statistical IR (ASiR-V), and DLIR with three selectable levels (low, medium, and high) at 2.5- and 0.625-mm slice thicknesses. Herein, signal, image noise, signal-to-noise ratio, and contrast-to-noise ratio of lymph nodes and overall image quality, artifact, and noise of selected regions of interest were evaluated by two radiologists. Next, cervical lymph node sharpness was evaluated using full width at half maximum. ResultsDLIR exhibited significantly reduced noise, ranging from 3.8 % to 35.9 % with improved signal-to-noise ratio (11.5–105.6 %) and contrast-to-noise ratio (10.5–107.5 %) compared with FBP and ASiR-V, for cervical lymph nodes (p < 0.001). Further, 0.625-mm-thick images reconstructed using DLIR-medium and DLIR-high had a lower noise than 2.5-mm-thick images reconstructed using FBP and ASiR-V. The lymph node margins and vessels on DLIR-medium and DLIR-high were sharper than those on FBP and ASiR-V (p < 0.05). Both readers agreed that DLIR had a better image quality than the conventional reconstruction algorithms. ConclusionDLIR-medium and -high provided superior cervical lymph node image quality in head and neck CT. Improved image quality affords thin-slice DLIR images for dose-reduction protocols in the future.

Can 1.25 mm thin-section images generated with Deep Learning Image Reconstruction technique replace standard-of-care 5 mm images in abdominal CT?

CT image reconstruction has evolved from filtered back projection to hybrid- and model-based iterative reconstruction. Deep learning-based image reconstruction is a relatively new technique that uses deep convolutional neural networks to improve image quality. To evaluate and compare 1.25 mm thin-section abdominal CT images reconstructed with deep learning image reconstruction (DLIR) with 5 mm thick images reconstructed with adaptive statistical iterative reconstruction (ASIR-V). This retrospective study included 52 patients (31 F; 56.9±16.9 years) who underwent abdominal CT scans between August-October 2019. Image reconstruction was performed to generate 5 mm images at 40% ASIR-V and 1.25 mm DLIR images at three strengths (low [DLIR-L], medium [DLIR-M], and high [DLIR-H]). Qualitative assessment was performed to determine image noise, contrast, visibility of small structures, sharpness, and artifact based on a 5-point-scale. Image preference determination was based on a 3-point-scale. Quantitative assessment included measurement of attenuation, image noise, and contrast-to-noise ratios (CNR). Thin-section images reconstructed with DLIR-M and DLIR-H yielded better image quality scores than 5 mm ASIR-V reconstructed images. Mean qualitative scores of DLIR-H for noise (1.77 ± 0.71), contrast (1.6 ± 0.68), small structure visibility (1.42 ± 0.66), sharpness (1.34 ± 0.55), and image preference (1.11 ± 0.34) were the best (p<0.05). DLIR-M yielded intermediate scores. All DLIR reconstructions showed superior ratings for artifacts compared to ASIR-V (p<0.05), whereas each DLIR group performed comparably (p>0.05, 0.405-0.763). In the quantitative assessment, there were no significant differences in attenuation values between all reconstructions (p>0.05). However, DLIR-H demonstrated the lowest noise (9.17 ± 3.11) and the highest CNR (CNRliver = 26.88 ± 6.54 and CNRportal vein = 7.92 ± 3.85) (all p<0.001). DLIR allows generation of thin-section (1.25 mm) abdominal CT images, which provide improved image quality with higher inter-reader agreement compared to 5 mm thick images reconstructed with ASIR-V. Improved image quality of thin-section CT images reconstructed with DLIR has several benefits in clinical practice, such as improved diagnostic performance without radiation dose penalties.

Accelerated Cardiac MRI Cine with Use of Resolution Enhancement Generative Adversarial Inline Neural Network.

Background Cardiac cine can benefit from deep learning-based image reconstruction to reduce scan time and/or increase spatial and temporal resolution. Purpose To develop and evaluate a deep learning model that can be combined with parallel imaging or compressed sensing (CS). Materials and Methods The deep learning model was built on the enhanced super-resolution generative adversarial inline neural network, trained with use of retrospectively identified cine images and evaluated in participants prospectively enrolled from September 2021 to September 2022. The model was applied to breath-hold electrocardiography (ECG)-gated segmented and free-breathing real-time cine images collected with reduced spatial resolution with use of generalized autocalibrating partially parallel acquisitions (GRAPPA) or CS. The deep learning model subsequently restored spatial resolution. For comparison, GRAPPA-accelerated cine images were collected. Diagnostic quality and artifacts were evaluated by two readers with use of Likert scales and compared with use of Wilcoxon signed-rank tests. Agreement for left ventricle (LV) function, volume, and strain was assessed with Bland-Altman analysis. Results The deep learning model was trained on 1616 patients (mean age ± SD, 56 years ± 16; 920 men) and evaluated in 181 individuals, 126 patients (mean age, 57 years ± 16; 77 men) and 55 healthy subjects (mean age, 27 years ± 10; 15 men). In breath-hold ECG-gated segmented cine and free-breathing real-time cine, the deep learning model and GRAPPA showed similar diagnostic quality scores (2.9 vs 2.9, P = .41, deep learning vs GRAPPA) and artifact score (4.4 vs 4.3, P = .55, deep learning vs GRAPPA). Deep learning acquired more sections per breath-hold than GRAPPA (3.1 vs one section, P < .001). In free-breathing real-time cine, the deep learning showed a similar diagnostic quality score (2.9 vs 2.9, P = .21, deep learning vs GRAPPA) and lower artifact score (3.9 vs 4.3, P < .001, deep learning vs GRAPPA). For both sequences, the deep learning model showed excellent agreement for LV parameters, with near-zero mean differences and narrow limits of agreement compared with GRAPPA. Conclusion Deep learning-accelerated cardiac cine showed similarly accurate quantification of cardiac function, volume, and strain to a standardized parallel imaging method. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Vannier and Wang in this issue.

Clinical Feasibility of Deep Learning-Based Image Reconstruction on Coronary Computed Tomography Angiography.

This study evaluated the feasibility of deep-learning-based image reconstruction (DLIR) on coronary computed tomography angiography (CCTA). By using a 20 cm water phantom, the noise reduction ratio and noise power spectrum were evaluated according to the different reconstruction methods. Then 46 patients who underwent CCTA were retrospectively enrolled. CCTA was performed using the 16 cm coverage axial volume scan technique. All CT images were reconstructed using filtered back projection (FBP); three model-based iterative reconstructions (MBIR) of 40%, 60%, and 80%; and three DLIR algorithms: low (L), medium (M), and high (H). Quantitative and qualitative image qualities of CCTA were compared according to the reconstruction methods. In the phantom study, the noise reduction ratios of MBIR-40%, MBIR-60%, MBIR-80%, DLIR-L, DLIR-M, and DLIR-H were 26.7 ± 0.2%, 39.5 ± 0.5%, 51.7 ± 0.4%, 33.1 ± 0.8%, 43.2 ± 0.8%, and 53.5 ± 0.1%, respectively. The pattern of the noise power spectrum of the DLIR images was more similar to FBP images than MBIR images. In a CCTA study, CCTA yielded a significantly lower noise index with DLIR-H reconstruction than with the other reconstruction methods. DLIR-H showed a higher SNR and CNR than MBIR (p < 0.05). The qualitative image quality of CCTA with DLIR-H was significantly higher than that of MBIR-80% or FBP. The DLIR algorithm was feasible and yielded a better image quality than the FBP or MBIR algorithms on CCTA.

MDST: multi-domain sparse-view CT reconstruction based on convolution and swin transformer

Objective.Sparse-view computed tomography (SVCT), which can reduce the radiation doses administered to patients and hasten data acquisition, has become an area of particular interest to researchers. Most existing deep learning-based image reconstruction methods are based on convolutional neural networks (CNNs). Due to the locality of convolution and continuous sampling operations, existing approaches cannot fully model global context feature dependencies, which makes the CNN-based approaches less efficient in modeling the computed tomography (CT) images with various structural information. Approach. To overcome the above challenges, this paper develops a novel multi-domain optimization network based on convolution and swin transformer (MDST). MDST uses swin transformer block as the main building block in both projection (residual) domain and image (residual) domain sub-networks, which models global and local features of the projections and reconstructed images. MDST consists of two modules for initial reconstruction and residual-assisted reconstruction, respectively. The sparse sinogram is first expanded in the initial reconstruction module with a projection domain sub-network. Then, the sparse-view artifacts are effectively suppressed by an image domain sub-network. Finally, the residual assisted reconstruction module to correct the inconsistency of the initial reconstruction, further preserving image details. Main results. Extensive experiments on CT lymph node datasets and real walnut datasets show that MDST can effectively alleviate the loss of fine details caused by information attenuation and improve the reconstruction quality of medical images. Significance. MDST network is robust and can effectively reconstruct images with different noise level projections. Different from the current prevalent CNN-based networks, MDST uses transformer as the main backbone, which proves the potential of transformer in SVCT reconstruction.