Advancing machine learning for MR image reconstruction with an open competition: Overview of the 2019 fastMRI challenge

Steerable Conditional Diffusion for Out-of-Distribution Adaptation in Medical Image Reconstruction

Denoising diffusion models have emerged as the go-to generative framework for solving inverse problems in imaging. A critical concern regarding these models is their performance on out-of-distribution tasks, which remains an under-explored challenge. Using a diffusion model on an out-of-distribution dataset, realistic reconstructions can be generated, but with hallucinating image features that are uniquely present in the training dataset. To address this discrepancy and improve reconstruction accuracy, we introduce a novel test-time adaptation sampling framework called Steerable Conditional Diffusion. Specifically, this framework adapts the diffusion model, concurrently with image reconstruction, based solely on the information provided by the available measurement. Utilising the proposed method, we achieve substantial enhancements in out-of-distribution performance across diverse imaging modalities, advancing the robust deployment of denoising diffusion models in real-world applications.

JotlasNet: Joint tensor low-rank and attention-based sparse unrolling network for accelerating dynamic MRI

Joint low-rank and sparse unrolling networks have shown superior performance in dynamic MRI reconstruction. However, existing works mainly utilized matrix low-rank priors, neglecting the tensor characteristics of dynamic MRI images, and only a global threshold is applied for the sparse constraint to the multi-channel data, limiting the flexibility of the network. Additionally, most of them have inherently complex network structure, with intricate interactions among variables. In this paper, we propose a novel deep unrolling network, JotlasNet, for dynamic MRI reconstruction by jointly utilizing tensor low-rank and attention-based sparse priors. Specifically, we utilize tensor low-rank prior to exploit the structural correlations in high-dimensional data. Convolutional neural networks are used to adaptively learn the low-rank and sparse transform domains. A novel attention-based soft thresholding operator is proposed to assign a unique learnable threshold to each channel of the data in the CNN-learned sparse domain. The network is unrolled from the elaborately designed composite splitting algorithm and thus features a simple yet efficient parallel structure. Extensive experiments on two datasets (OCMR, CMRxRecon) demonstrate the superior performance of JotlasNet in dynamic MRI reconstruction.

Deep Learning-Driven Abbreviated Shoulder MRI Protocols: Diagnostic Accuracy in Clinical Practice

BACKGROUND

Deep learning (DL) reconstruction techniques have shown promise in reducing MRI acquisition times while maintaining image quality. However, the impact of different acceleration factors on diagnostic accuracy in shoulder MRI remains unexplored in clinical practice.

PURPOSE

The purpose of this study was to evaluate the diagnostic accuracy of 2-fold and 4-fold DL-accelerated shoulder MRI protocols compared to standard protocols in clinical practice.

MATERIALS AND METHODS

In this prospective single-center study, 88 consecutive patients (49 males, 39 females; mean age, 51 years) underwent shoulder MRI examinations using standard, 2-fold (DL2), and 4-fold (DL4) accelerated protocols between June 2023 and January 2024. Four independent radiologists (experience range: 4-25 years) evaluated the presence of bone marrow edema (BME), rotator cuff tears, and labral lesions. The sensitivity, specificity, and interobserver agreement were calculated. Diagnostic confidence was assessed using a 4-point scale. The impact of reader experience was analyzed by stratifying the radiologists into ≤10 and >10 years of experience.

RESULTS

Both accelerated protocols demonstrated high diagnostic accuracy. For BME detection, DL2 and DL4 achieved 100% sensitivity and specificity. In rotator cuff evaluation, DL2 showed a sensitivity of 98-100% and specificity of 99-100%, while DL4 maintained a sensitivity of 95-98% and specificity of 99-100%. Labral tear detection showed perfect sensitivity (100%) with DL2 and slightly lower sensitivity (89-100%) with DL4. Interobserver agreement was excellent across the protocols (Kendall's W = 0.92-0.98). Reader experience did not significantly impact diagnostic performance. The area under the ROC curve was 0.94 for DL2 and 0.90 for DL4 (p = 0.32).

CLINICAL IMPLICATIONS

The implementation of DL-accelerated protocols, particularly DL2, could improve workflow efficiency by reducing acquisition times by 50% while maintaining diagnostic reliability. This could increase patient throughput and accessibility to MRI examinations without compromising diagnostic quality.

CONCLUSIONS

DL-accelerated shoulder MRI protocols demonstrate high diagnostic accuracy, with DL2 showing performance nearly identical to that of the standard protocol. While DL4 maintains acceptable diagnostic accuracy, it shows a slight sensitivity reduction for subtle pathologies, particularly among less experienced readers. The DL2 protocol represents an optimal balance between acquisition time reduction and diagnostic confidence.

Spatial-frequency aware zero-centric residual unfolding network for MRI reconstruction

Magnetic Resonance Imaging is a cornerstone of medical diagnostics, providing high-quality soft tissue contrast through non-invasive methods. However, MRI technology faces critical limitations in imaging speed and resolution. Prolonged scan times not only increase patient discomfort but also contribute to motion artifacts, further compromising image quality. Compressed Sensing (CS) theory has enabled the acquisition of partial k-space data, which can then be effectively reconstructed to recover the original image using advanced reconstruction algorithms. Recently, deep learning has been widely applied to MRI reconstruction, aiming to reduce the artifacts in the image domain caused by undersampling in k-space and enhance image quality. As deep learning continues to evolve, the undersampling factors in k-space have gradually increased in recent years. However, these layers are limited in compensating for reconstruction errors in the unsampled areas, impeding further performance improvements. To address this, we propose a learnable spatial-frequency difference-aware module that complements the learnable data consistency layer, mapping k-space domain differences to the spatial image domain for perceptual compensation. Additionally, inspired by wavelet decomposition, we introduce explicit priors by decomposing images into mean and residual components, enforcing a refined zero-mean constraint on the residuals while maintaining computational efficiency. Comparative experiments on the FastMRI and Calgary-Campinas datasets demonstrate that our method achieves superior reconstruction performance against seven state-of-the-art techniques. Ablation studies further confirm the efficacy of our model's architecture, establishing a new pathway for enhanced MRI reconstruction.

Speed and efficiency: evaluating pulmonary nodule detection with AI-enhanced 3D gradient echo imaging

OBJECTIVES

Evaluating the diagnostic feasibility of accelerated pulmonary MR imaging for detection and characterisation of pulmonary nodules with artificial intelligence-aided compressed sensing.

MATERIALS AND METHODS

In this prospective trial, patients with benign and malignant lung nodules admitted between December 2021 and December 2022 underwent chest CT and pulmonary MRI. Pulmonary MRI used a respiratory-gated 3D gradient echo sequence, accelerated with a combination of parallel imaging, compressed sensing, and deep learning image reconstruction with three different acceleration factors (CS-AI-7, CS-AI-10, and CS-AI-15). Two readers evaluated image quality (5-point Likert scale), nodule detection and characterisation (size and morphology) of all sequences compared to CT in a blinded setting. Reader agreement was determined using the intraclass correlation coefficient (ICC).

RESULTS

Thirty-seven patients with 64 pulmonary nodules (solid n = 57 [3-107 mm] part-solid n = 6 [ground glass/solid 8 mm/4-28 mm/16 mm] ground glass nodule n = 1 [20 mm]) were analysed. Nominal scan times were CS-AI-7 3:53 min; CS-AI-10 2:34 min; CS-AI-15 1:50 min. CS-AI-7 showed higher image quality, while quality remained diagnostic even for CS-AI-15. Detection rates of pulmonary nodules were 100%, 98.4%, and 96.8% for CS-AI factors 7, 10, and 15, respectively. Nodule morphology was best at the lowest acceleration and was inferior to CT in only 5% of cases, compared to 10% for CS-AI-10 and 23% for CS-AI-15. The nodule size was comparable for all sequences and deviated on average < 1 mm from the CT size.

CONCLUSION

The combination of compressed sensing and AI enables a substantial reduction in the scan time of lung MRI while maintaining a high detection rate of pulmonary nodules.

CLINICAL RELEVANCE STATEMENT

Incorporating compressed sensing and AI in pulmonary MRI achieves significant time savings without compromising nodule detection or characteristics. This advancement holds clinical promise, enhancing efficiency in lung cancer screening without sacrificing diagnostic quality.

KEY POINTS

Lung cancer screening by MRI may be possible but would benefit from scan time optimisation. Significant scan time reduction, high detection rates, and preserved nodule characteristics were achieved across different acceleration factors. Integrating compressed sensing and AI in pulmonary MRI offers efficient lung cancer screening without compromising diagnostic quality.

Improving radiologist detection of meniscal abnormality on undersampled, deep learning reconstructed knee MRI

Background

Accurate interpretation of meniscal anomalies on knee MRI is critical for diagnosis and treatment planning, with artificial intelligence emerging as a promising tool to support and enhance this process through automated anomaly detection.

Purpose

To evaluate the impact of an artificial intelligence (AI) anomaly detection assistant on radiologists' interpretation of meniscal anomalies in undersampled, deep learning (DL)-reconstructed knee MRI and assess the relationship between reconstruction quality metrics and anomaly detection performance.

Materials and Methods

This retrospective study included 947 knee MRI examinations; 51 were excluded for poor image quality, leaving 896 participants (mean age, 44.7 ± 15.3 years; 472 women). Using 8-fold undersampled data, DL-based reconstructed images were generated. An object detection model was trained on original, fully sampled images and evaluated on 1 original and 14 DL-reconstructed test sets to identify meniscal lesions. Standard reconstruction metrics (normalized root mean square error, peak signal-to-noise ratio, and structural similarity index) and anomaly detection metrics (mean average precision, F1 score) were quantified and compared. Two radiologists independently reviewed a stratified sample of 50 examinations unassisted and assisted with AI-predicted anomaly boxes. McNemar's test evaluated differences in diagnostic performance; Cohen's kappa assessed interrater agreement.

Results

On the original images, the anomaly detection model achieved the following: 70.53% precision, 72.17% recall, 63.09% mAP, and a 71.34% F1 score. Comparing performance among the undersampled reconstruction datasets, box-based reconstruction metrics showed better correlation with detection performance than traditional image-based metrics (mAP to box-based SSIM, r = 0.81, P < .01; mAP to image-based SSIM, r = 0.64, P = .01). In 50 participants, AI assistance improved radiologists' accuracy on reconstructed images. Sensitivity increased from 77.27% (95% CI, 65.83-85.72; 51/66) to 80.30% (95% CI, 69.16-88.11; 53/66), and specificity improved from 88.46% (95% CI, 83.73-91.95; 207/234) to 90.60% (95% CI, 86.18-93.71; 212/234) (P < .05).

Conclusion

AI-assisted meniscal anomaly detection enhanced radiologists' interpretation of undersampled, DL-reconstructed knee MRI. Anomaly detection may serve as a complementary tool alongside other reconstruction metrics to assess the preservation of clinically important features in reconstructed images, warranting further investigation.

Validation of a Deep Learning-Based Method for Accelerating Susceptibility-Weighted Imaging in Clinical Settings

Susceptibility-weighted imaging (SWI) has been widely used in clinical contexts, in which the speed of acquisition is frequently a critical issue. In this study, we aim to test the feasibility of a deep learning (DL)-based reconstruction method for accelerating SWI acquisition in clinical settings. A total of 61 subjects were consecutively enrolled. SWI scans using prospective under-sampling and DL-based reconstruction method (acceleration factor = 5, acquisition time: 1:46) and parallel imaging (PI, acceleration factor = 2, acquisition time: 4:45) were acquired from each subject. The DL-based method utilizes a cascaded convolutional neural network, namely, ReconNet3D, to perform k-space to image reconstruction with the capacity of single-channel input and single-channel output. The DL-SWI and PI-SWI results were compared quantitatively using structural similarity index (SSIM) and peak signal-to-noise ratio (PSNR). Two raters independently assessed image quality from five aspects: artifacts, noise level, sharpness, lesion conspicuity, and overall image quality. The numbers of microbleeds were also counted. Finally, both sets of images of the same subject were reviewed side-by-side for noninferiority assessments. The DL-SWI images showed good similarity in terms of SSIM (mean ± SD: 0.89 ± 0.02) and PSNR (mean ± SD: 36.91 ± 2.41) with PI-SWI. In comparison, DL-SWI images had significantly better scores in terms of artifacts, noise, and overall image quality score (4.15 vs. 3.33; 3.85 vs. 3.16; 3.85 vs. 3.44, all with p < 0.001). DL-SWI showed reduced image sharpness (p = 0.031), but no significant difference regarding lesion conspicuity. The number of microbleeds identified from the DL-SWI images completely matched the PI-SWI results. We had not observed any false-negative or false-positive readings on DL-SWI images. The DL-based SWI acceleration method could significantly reduce scan time and maintain image quality, suggesting its great potential in clinical applications.

AI-Assisted Compressed Sensing Enables Faster Brain MRI for the Elderly: Image Quality and Diagnostic Equivalence with Conventional Imaging

Purpose

Conventional brain MRI protocols are time-consuming, which can lead to patient discomfort and inefficiency in clinical settings. This study aims to assess the feasibility of using artificial intelligence-assisted compressed sensing (ACS) to reduce brain MRI scan time while maintaining image quality and diagnostic accuracy compared to a conventional imaging protocol.

Patients and Methods

Seventy patients from the department of neurology underwent brain MRI scans using both conventional and ACS protocols, including axial and sagittal T2-weighted fast spin-echo sequences and T2-fluid attenuated inversion recovery (FLAIR) sequence. Two radiologists independently evaluated image quality based on avoidance of artifacts, boundary sharpness, visibility of lesions, and overall image quality using a 5-point Likert scale. Pathological features, including white matter hyperintensities, lacunar infarcts, and enlarged perivascular spaces, were also assessed. The interchangeability of the two protocols was determined by calculating the 95% confidence interval (CI) for the individual equivalence index. Additionally, Cohen's weighted kappa statistic was used to assess inter-protocol intra-observer agreement.

Results

The ACS images demonstrated superior quality across all qualitative features compared to the conventional ones. Both protocols showed no significant difference in detecting pathological conditions. The 95% CI for the individual equivalence index was below 5% for all variables except enlarged perivascular spaces, indicating the interchangeability of the conventional and ACS protocols in most cases. The inter-rater reliability between the two radiologists was strong, with kappa values of 0.78, 0.74, 0.70 and 0.86 for image quality evaluation and 0.74, 0.80 and 0.70 for diagnostic performance, indicating good-to-excellent agreement in their evaluations.

Conclusion

The ACS technique reduces brain MRI scan time by 29.2% while achieving higher image quality and equivalent diagnostic accuracy compared to the conventional protocol. This suggests that ACS could be potentially adopted for routine clinical use in brain MRI.

Validation of deep-learning accelerated quantitative susceptibility mapping for deep brain nuclei

Purpose

To test the feasibility and consistency of a deep-learning (DL) accelerated QSM method for deep brain nuclei evaluation.

Methods

Participants were scanned with both parallel imaging (PI)-QSM and DL-QSM methods. The PI- and DL-QSM scans had identical imaging parameters other than acceleration factors (AF). The DL-QSM employed Poisson disk style under-sampling scheme and a previously developed cascaded CNN based reconstruction model, with acquisition time of 4:35, 3:15, and 2:11 for AF of 3, 4, and 5, respectively. For PI-QSM acquisition, the AF was 2 and the acquisition time was 6:46. The overall image similarity was assessed between PI- and DL-QSM images using the structural similarity index (SSIM) and peak signal-to-noise ratio (PSNR). QSM values from 7 deep brain nuclei were extracted and agreements between images with different Afs were assessed. Finally, the correlations between age and QSM values in the selected deep brain nuclei were evaluated.

Results

59 participants were recruited. Compared to PI-QSM images, the mean SSIM of DL images were 0.87, 0.86, and 0.85 for AF of 3, 4, and 5. The mean PSNR were 44.56, 44.53, and 44.23. Susceptibility values from DL-QSM were highly consistent with routine PI-QSM images, with differences of less than 5% at the group level. Furthermore, the associations between age and QSM values could be consistently revealed.

Conclusion

DL-QSM could be used for measuring susceptibility values of deep brain nucleus. An AF up to 5 did not significantly impact the correlation between age and susceptibility in deep brain nuclei.

Fast and Reliable Score-Based Generative Model for Parallel MRI

The score-based generative model (SGM) can generate high-quality samples, which have been successfully adopted for magnetic resonance imaging (MRI) reconstruction. However, the recent SGMs may take thousands of steps to generate a high-quality image. Besides, SGMs neglect to exploit the redundancy in space. To overcome the above two drawbacks, in this article, we propose a fast and reliable SGM (FRSGM). First, we propose deep ensemble denoisers (DEDs) consisting of SGM and the deep denoiser, which are used to solve the proximal problem of the implicit regularization term. Second, we propose a spatially adaptive self-consistency (SASC) term as the regularization term of the -space data. We use the alternating direction method of multipliers (ADMM) algorithm to solve the minimization model of compressed sensing (CS)-MRI incorporating the image prior term and the SASC term, which is significantly faster than the related works based on SGM. Meanwhile, we can prove that the iterating sequence of the proposed algorithm has a unique fixed point. In addition, the DED and the SASC term can significantly improve the generalization ability of the algorithm. The features mentioned above make our algorithm reliable, including the fixed-point convergence guarantee, the exploitation of the space, and the powerful generalization ability.

Deep Learning Superresolution for Simultaneous Multislice Parallel Imaging-Accelerated Knee MRI Using Arthroscopy Validation

Background Deep learning (DL) methods can improve accelerated MRI but require validation against an independent reference standard to ensure robustness and accuracy. Purpose To validate the diagnostic performance of twofold-simultaneous-multislice (SMSx2) twofold-parallel-imaging (PIx2)-accelerated DL superresolution MRI in the knee against conventional SMSx2-PIx2-accelerated MRI using arthroscopy as the reference standard. Materials and Methods Adults with painful knee conditions were prospectively enrolled from December 2021 to October 2022. Participants underwent fourfold SMSx2-PIx2-accelerated standard-of-care and investigational DL superresolution MRI at 3 T. Seven radiologists independently evaluated the MRI examinations for overall image quality (using Likert scale scores: 1, very bad, to 5, very good) and the presence or absence of meniscus and ligament tears. Articular cartilage was categorized as intact, or partial or full-thickness defects. Statistical analyses included interreader agreements (Cohen κ and Gwet AC2) and diagnostic performance testing used area under the receiver operating characteristic curve (AUC) values. Results A total of 116 adults (mean age, 45 years ± 15 [SD]; 74 men) who underwent arthroscopic surgery within 38 days ± 22 were evaluated. Overall image quality was better for DL superresolution MRI (median Likert score, 5; range, 3-5) than conventional MRI (median Likert score, 4; range, 3-5) (P < .001). Diagnostic performances of conventional versus DL superresolution MRI were similar for medial meniscus tears (AUC, 0.94 [95% CI: 0.89, 0.97] vs 0.94 [95% CI: 0.90, 0.98], respectively; P > .99), lateral meniscus tears (AUC, 0.85 [95% CI: 0.78, 0.91] vs 0.87 [95% CI: 0.81, 0.94], respectively; P = .96), and anterior cruciate ligament tears (AUC, 0.98 [95% CI: 0.93, >0.99] vs 0.98 [95% CI: 0.93, >0.99], respectively; P > .99). DL superresolution MRI (AUC, 0.78; 95% CI: 0.75, 0.81) had higher diagnostic performance than conventional MRI (AUC, 0.71; 95% CI: 0.67, 0.74; P = .002) for articular cartilage lesions. DL superresolution MRI did not introduce hallucinations or erroneously omit abnormalities. Conclusion Compared with conventional SMSx2-PIx2-accelerated MRI, fourfold SMSx2-PIx2-accelerated DL superresolution MRI in the knee provided better image quality, similar performance for detecting meniscus and ligament tears, and improved performance for depicting articular cartilage lesions. © RSNA, 2025 Supplemental material is available for this article. See also the editorial by Nevalainen in this issue.

Agreement of Image Quality Metrics with Radiological Evaluation in the Presence of Motion Artifacts

Purpose

Reliable image quality assessment is crucial for evaluating new motion correction methods for magnetic resonance imaging. In this work, we compare the performance of commonly used reference-based and reference-free image quality metrics on a unique dataset with real motion artifacts. We further analyze the image quality metrics' robustness to typical pre-processing techniques.

Methods

We compared five reference-based and five reference-free image quality metrics on data acquired with and without intentional motion (2D and 3D sequences). The metrics were recalculated seven times with varying pre-processing steps. The anonymized images were rated by radiologists and radiographers on a 1-5 Likert scale. Spearman correlation coefficients were computed to assess the relationship between image quality metrics and observer scores.

Results

All reference-based image quality metrics showed strong correlation with observer assessments, with minor performance variations across sequences. Among reference-free metrics, Average Edge Strength offers the most promising results, as it consistently displayed stronger correlations across all sequences compared to the other reference-free metrics. Overall, the strongest correlation was achieved with percentile normalization and restricting the metric values to the skull-stripped brain region. In contrast, correlations were weaker when not applying any brain mask and using min-max or no normalization.

Conclusion

Reference-based metrics reliably correlate with radiological evaluation across different sequences and datasets. Pre-processing steps, particularly normalization and brain masking, significantly influence the correlation values. Future research should focus on refining pre-processing techniques and exploring machine learning approaches for automated image quality evaluation.

Motion-Guided Deep Image Prior for Cardiac MRI

Cardiovascular magnetic resonance imaging is a powerful diagnostic tool for assessing cardiac structure and function. Traditional breath-held imaging protocols, however, pose challenges for patients with arrhythmias or limited breath-holding capacity. We introduce Motion-Guided Deep Image prior (M-DIP), a novel unsupervised reconstruction framework for accelerated real-time cardiac MRI. M-DIP employs a spatial dictionary to synthesize a time-dependent template image, which is further refined using time-dependent deformation fields that model cardiac and respiratory motion. Unlike prior DIP-based methods, M-DIP simultaneously captures physiological motion and frame-to-frame content variations, making it applicable to a wide range of dynamic applications. We validate M-DIP using simulated MRXCAT cine phantom data as well as free-breathing real-time cine and single-shot late gadolinium enhancement data from clinical patients. Comparative analyses against state-of-the-art supervised and unsupervised approaches demonstrate M-DIP's performance and versatility. M-DIP achieved better image quality metrics on phantom data, as well as higher reader scores for in-vivo patient data.

Architecture-Agnostic Untrained Network Priors for Image Reconstruction with Frequency Regularization

Untrained networks inspired by deep image priors have shown promising capabilities in recovering high-quality images from noisy or partial measurements without requiring training sets. Their success is widely attributed to implicit regularization due to the spectral bias of suitable network architectures. However, the application of such network-based priors often entails superfluous architectural decisions, risks of overfitting, and lengthy optimization processes, all of which hinder their practicality. To address these challenges, we propose efficient architecture-agnostic techniques to directly modulate the spectral bias of network priors: 1) bandwidth-constrained input, 2) bandwidth-controllable upsamplers, and 3) Lipschitz-regularized convolutional layers. We show that, with just a few lines of code, we can reduce overfitting in underperforming architectures and close performance gaps with high-performing counterparts, minimizing the need for extensive architecture tuning. This makes it possible to employ a more compact model to achieve performance similar or superior to larger models while reducing runtime. Demonstrated on inpainting-like MRI reconstruction task, our results signify for the first time that architectural biases, overfitting, and runtime issues of untrained network priors can be simultaneously addressed without architectural modifications. Our code is publicly available .

Improved Resolution and Image Quality of Musculoskeletal Magnetic Resonance Imaging using Deep Learning-based Denoising Reconstruction: A Prospective Clinical Study

OBJECTIVE

To prospectively evaluate a deep learning-based denoising reconstruction (DLR) for improved resolution and image quality in musculoskeletal (MSK) magnetic resonance imaging (MRI).

METHODS

Images from 137 contrast-weighted sequences in 40 MSK patients were evaluated. Each sequence was performed twice, first with the routine parameters and reconstructed with a routine reconstruction filter (REF), then with higher resolution and reconstructed with DLR, and with three conventional reconstruction filters (NL2, GA43, GA53). The five reconstructions (REF, DLR, NL2, GA43, and GA53) were de-identified, randomized, and blindly reviewed by three MSK radiologists using eight scoring criteria and a forced ranking. Quantitative SNR, CNR, and structure's full width at half maximum (FWHM) for resolution assessment were measured and compared. To account for repeated measures, Generalized Estimating Equations (GEE) with Bonferroni adjustment was used to compare the reader's scores, SNR, CNR, and FWHM between DLR vs. NL2, GA43, GA53, and REF.

RESULTS

Compared to the routine REF images, the resolution was improved by 47.61% with DLR from 0.39 ± 0.15 mm2 to 0.20 ± 0.06 mm2 (p < 0.001). Per-sequence average scan time was shortened by 7.93% with DLR from 165.58 ± 21.86 s to 152.45 ± 25.65 s (p < 0.001). Based on the average scores, DLR images were rated significantly higher in all image quality criteria and the forced ranking (p < 0.001).

CONCLUSION

This prospective clinical evaluation demonstrated that DLR allows approximately two times finer resolution and improved image quality compared to the standard-of-care images.

Deep-learning-based image reconstruction with limited data: generating synthetic raw data using deep learning

OBJECT

Deep learning has shown great promise for fast reconstruction of accelerated MRI acquisitions by learning from large amounts of raw data. However, raw data is not always available in sufficient quantities. This study investigates synthetic data generation to complement small datasets and improve reconstruction quality.

MATERIALS AND METHODS

An adversarial auto-encoder was trained to generate phase and coil sensitivity maps from magnitude images, which were combined into synthetic raw data. On a fourfold accelerated MR reconstruction task, deep-learning-based reconstruction networks were trained with varying amounts of training data (20 to 160 scans). Test set performance was compared between baseline experiments and experiments that incorporated synthetic training data.

RESULTS

Training with synthetic raw data showed decreasing reconstruction errors with increasing amounts of training data, but importantly this was magnitude-only data, rather than real raw data. For small training sets, training with synthetic data decreased the mean absolute error (MAE) by up to 7.5%, whereas for larger training sets the MAE increased by up to 2.6%.

DISCUSSION

Synthetic raw data generation improved reconstruction quality in scenarios with limited training data. A major advantage of synthetic data generation is that it allows for the reuse of magnitude-only datasets, which are more readily available than raw datasets.

Evaluation of 3D T1-weighted spoiled gradient echo MR image quality using artificial intelligence image reconstruction techniques in the pediatric brain

PURPOSE

To assess image quality and diagnostic confidence of 3D T1-weighted spoiled gradient echo (SPGR) MRI using artificial intelligence (AI) reconstruction.

MATERIALS AND METHODS

This prospective, IRB-approved study enrolled 50 pediatric patients (mean age = 11.8 ± 3.1 years) undergoing clinical brain MRI. In addition to standard of care (SOC) compressed SENSE (CS = 2.5), 3D T1-weighted SPGR images were obtained with higher CS acceleration factors (5 and 8) to evaluate the ability of AI reconstruction to improve image quality and reduce scan time. Images were reviewed independently on dedicated research PACS workstations by two neuroradiologists. Quantitative analysis of signal intensities to calculate apparent grey and white matter signal to noise (aSNR) and grey-white matter apparent contrast to noise ratios (aCNR) was performed.

RESULTS

AI improved overall image quality compared to standard CS reconstruction in 35% (35/100) of evaluations in CS = 2.5 (average scan time = 221 ± 6.9 s), 100% (46/46) of CS = 5 (average scan time = 113.3 ± 4.6 s) and 94% (47/50) of CS = 8 (average scan time = 74.1 ± 0.01 s). Quantitative analysis revealed significantly higher grey matter aSNR, white matter aSNR and grey-white matter aCNR with AI reconstruction compared to standard reconstruction for CS 5 and 8 (all p-values < 0.001), however not for CS 2.5.

CONCLUSIONS

AI reconstruction improved overall image quality and gray-white matter qualitative and quantitative aSNR and aCNR in highly accelerated (CS = 5 and 8) 3D T1W SPGR images in the majority of pediatric patients.

The STOIC2021 COVID-19 AI challenge: Applying reusable training methodologies to private data

Challenges drive the state-of-the-art of automated medical image analysis. The quantity of public training data that they provide can limit the performance of their solutions. Public access to the training methodology for these solutions remains absent. This study implements the Type Three (T3) challenge format, which allows for training solutions on private data and guarantees reusable training methodologies. With T3, challenge organizers train a codebase provided by the participants on sequestered training data. T3 was implemented in the STOIC2021 challenge, with the goal of predicting from a computed tomography (CT) scan whether subjects had a severe COVID-19 infection, defined as intubation or death within one month. STOIC2021 consisted of a Qualification phase, where participants developed challenge solutions using 2000 publicly available CT scans, and a Final phase, where participants submitted their training methodologies with which solutions were trained on CT scans of 9724 subjects. The organizers successfully trained six of the eight Final phase submissions. The submitted codebases for training and running inference were released publicly. The winning solution obtained an area under the receiver operating characteristic curve for discerning between severe and non-severe COVID-19 of 0.815. The Final phase solutions of all finalists improved upon their Qualification phase solutions.

Deep learning and AI in reducing magnetic resonance imaging scanning time: advantages and pitfalls in clinical practice

Magnetic resonance imaging (MRI) is a powerful imaging modality, but one of its drawbacks is its relatively long scanning time to acquire high-resolution images. Reducing the scanning time has become a critical area of focus in MRI, aiming to enhance patient comfort, reduce motion artifacts, and increase MRI throughput. In the past 5 years, artificial intelligence (AI)-based algorithms, particularly deep learning models, have been developed to reconstruct high-resolution images from significantly fewer data points. These new techniques significantly enhance MRI efficiency, improve patient comfort and lower patient motion artifacts. Improving MRI throughput with lower scanning duration increases accessibility, potentially reducing the need for additional MRI machines and associated costs. Several fields can benefit from shortened protocols, especially for routine exams. In oncologic imaging, faster MRI scans can facilitate more regular monitoring of cancer patients. In patients suffering from neurological disorders, rapid brain imaging can aid in the quick assessment of conditions like stroke, multiple sclerosis, and epilepsy, improving patient outcomes. In chronic inflammatory disease, faster imaging may help in reducing the interval between imaging to better check therapy outcomes. Additionally, reducing scanning time could effectively help MRI to play a role in emergency medicine and acute conditions such as trauma or acute ischaemic stroke. The purpose of this paper is to describe and discuss the advantages and disadvantages of introducing deep learning reconstruction techniques to reduce MRI scanning times in clinical practice.

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 6 min 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.

Screening for pancreatic cancer in high-risk individuals using MRI: optimization of scan techniques to detect small lesions

Pancreatic cancer has a dismal prognosis in the general population. However, early detection and treatment of disease in high-risk individuals can improve survival, as patients with localized disease and especially patients with lesions smaller than 10 mm show greatly improved 5-year survival rates. To achieve early detection through MRI surveillance programs, optimization of imaging is required. Advances in MRI technologies in both hardware and software over the years have enabled reliable detection of pancreatic cancer at a small size and early stage. Standardization of dedicated imaging protocols for the pancreas are still lacking. In this review we discuss state of the art scan techniques, sequences, reduction of artifacts and imaging strategies that enable early detection of lesions. Furthermore, we present the imaging features of small pancreatic cancers from a large cohort of high-risk individuals. Refinement of MRI techniques, increased scan quality and the use of artificial intelligence may further improve early detection and the prognosis of pancreatic cancer in a screening setting.

Rapid 2D 23Na MRI of the calf using a denoising convolutional neural network

PURPOSE

23Na MRI can be used to quantify in-vivo tissue sodium concentration (TSC), but the inherently low 23Na signal leads to long scan times and/or noisy or low-resolution images. Reconstruction algorithms such as compressed sensing (CS) have been proposed to mitigate low signal-to-noise ratio (SNR); although, these can result in unnatural images, suboptimal denoising and long processing times. Recently, machine learning has been increasingly used to denoise 1H MRI acquisitions; however, this approach typically requires large volumes of high-quality training data, which is not readily available for 23Na MRI. Here, we propose using 1H data to train a denoising convolutional neural network (CNN), which we subsequently demonstrate on prospective 23Na images of the calf.

METHODS

1893 1H fat-saturated transverse slices of the knee from the open-source fastMRI dataset were used to train denoising CNNs for different levels of noise. Synthetic low SNR images were generated by adding gaussian noise to the high-quality 1H k-space data before reconstruction to create paired training data. For prospective testing, 23Na images of the calf were acquired in 10 healthy volunteers with a total of 150 averages over ten minutes, which were used as a reference throughout the study. From this data, images with fewer averages were retrospectively reconstructed using a non-uniform fast Fourier transform (NUFFT) as well as CS, with the NUFFT images subsequently denoised using the trained CNN.

RESULTS

CNNs were successfully applied to 23Na images reconstructed with 50, 40 and 30 averages. Muscle and skin apparent TSC quantification from CNN-denoised images were equivalent to those from CS images, with <0.9 mM bias compared to reference values. Estimated SNR was significantly higher in CNN-denoised images compared to NUFFT, CS and reference images. Quantitative edge sharpness was equivalent for all images. For subjective image quality ranking, CNN-denoised images ranked equally best with reference images and significantly better than NUFFT and CS images.

CONCLUSION

Denoising CNNs trained on 1H data can be successfully applied to 23Na images of the calf; thus, allowing scan time to be reduced from ten minutes to two minutes with little impact on image quality or apparent TSC quantification accuracy.

Deep learning for accelerated and robust MRI reconstruction

Deep learning (DL) has recently emerged as a pivotal technology for enhancing magnetic resonance imaging (MRI), a critical tool in diagnostic radiology. This review paper provides a comprehensive overview of recent advances in DL for MRI reconstruction, and focuses on various DL approaches and architectures designed to improve image quality, accelerate scans, and address data-related challenges. It explores end-to-end neural networks, pre-trained and generative models, and self-supervised methods, and highlights their contributions to overcoming traditional MRI limitations. It also discusses the role of DL in optimizing acquisition protocols, enhancing robustness against distribution shifts, and tackling biases. Drawing on the extensive literature and practical insights, it outlines current successes, limitations, and future directions for leveraging DL in MRI reconstruction, while emphasizing the potential of DL to significantly impact clinical imaging practices.

Evaluation of T2W FLAIR MR image quality using artificial intelligence image reconstruction techniques in the pediatric brain

BACKGROUND

Artificial intelligence (AI) reconstruction techniques have the potential to improve image quality and decrease imaging time. However, these techniques must be assessed for safe and effective use in clinical practice.

OBJECTIVE

To assess image quality and diagnostic confidence of AI reconstruction in the pediatric brain on fluid-attenuated inversion recovery (FLAIR) imaging.

MATERIALS AND METHODS

This prospective, institutional review board (IRB)-approved study enrolled 50 pediatric patients (median age=12 years, Q1=10 years, Q3=14 years) undergoing clinical brain MRI. T2-weighted (T2W) FLAIR images were reconstructed by both standard clinical and AI reconstruction algorithms (strong denoising). Images were independently rated by two neuroradiologists on a dedicated research picture archiving and communication system (PACS) to indicate whether AI increased, decreased, or had no effect on image quality compared to standard reconstruction. Quantitative analysis of signal intensities was also performed to calculate apparent signal to noise (aSNR) and apparent contrast to noise (aCNR) ratios.

RESULTS

AI reconstruction was better than standard in 99% (reader 1, 49/50; reader 2, 50/50) for overall image quality, 99% (reader 1, 49/50; reader 2, 50/50) for subjective SNR, and 98% (reader 1, 49/50; reader 2, 49/50) for diagnostic preference. Quantitative analysis revealed significantly higher gray matter aSNR (30.6±6.5), white matter aSNR (21.4±5.6), and gray-white matter aCNR (7.1±1.6) in AI-reconstructed images compared to standard reconstruction (18±2.7, 14.2±2.8, 4.4±0.8, p<0.001) respectively.

CONCLUSION

We conclude that AI reconstruction improved T2W FLAIR image quality in most patients when compared with standard reconstruction in pediatric patients.

Deep Learning-based Image Enhancement Techniques for Fast MRI in Neuroimaging

Despite its superior soft tissue contrast and non-invasive nature, MRI requires long scan times due to its intrinsic signal acquisition principles, a main drawback which technological advancements in MRI have been focused on. In particular, scan time reduction is a natural requirement in neuroimaging due to detailed structures requiring high resolution imaging and often volumetric (3D) acquisitions, and numerous studies have recently attempted to harness deep learning (DL) technology in enabling scan time reduction and image quality improvement. Various DL-based image reconstruction products allow for additional scan time reduction on top of existing accelerated acquisition methods without compromising the image quality.

Improvement of image quality in diffusion-weighted imaging with model-based deep learning reconstruction for evaluations of the head and neck

OBJECTIVES

To investigate the utility of deep learning (DL)-based image reconstruction using a model-based approach in head and neck diffusion-weighted imaging (DWI).

MATERIALS AND METHODS

We retrospectively analyzed the cases of 41 patients who underwent head/neck DWI. The DWI in 25 patients demonstrated an untreated lesion. We performed qualitative and quantitative assessments in the DWI analyses with both deep learning (DL)- and conventional parallel imaging (PI)-based reconstructions. For the qualitative assessment, we visually evaluated the overall image quality, soft tissue conspicuity, degree of artifact(s), and lesion conspicuity based on a five-point system. In the quantitative assessment, we measured the signal-to-noise ratio (SNR) of the bilateral parotid glands, submandibular gland, the posterior muscle, and the lesion. We then calculated the contrast-to-noise ratio (CNR) between the lesion and the adjacent muscle.

RESULTS

Significant differences were observed in the qualitative analysis between the DWI with PI-based and DL-based reconstructions for all of the evaluation items (p < 0.001). In the quantitative analysis, significant differences in the SNR and CNR between the DWI with PI-based and DL-based reconstructions were observed for all of the evaluation items (p = 0.002 ~ p < 0.001).

DISCUSSION

DL-based image reconstruction with the model-based technique effectively provided sufficient image quality in head/neck DWI.

[Ethics and artificial intelligence]

The introduction of artificial intelligence (AI) into radiology promises to enhance efficiency and improve diagnostic accuracy, yet it also raises manifold ethical questions. These include data protection issues, the future role of radiologists, liability when using AI systems, and the avoidance of bias. To prevent data bias, the datasets need to be compiled carefully and to be representative of the target population. Accordingly, the upcoming European Union AI act sets particularly high requirements for the datasets used in training medical AI systems. Cognitive bias occurs when radiologists place too much trust in the results provided by AI systems (overreliance). So far, diagnostic AI systems are used almost exclusively as "second look" systems. If diagnostic AI systems are to be used in the future as "first look" systems or even as autonomous AI systems in order to enhance efficiency in radiology, the question of liability needs to be addressed, comparable to liability for autonomous driving. Such use of AI would also significantly change the role of radiologists.

Accelerated MRI reconstructions via variational network and feature domain learning

We introduce three architecture modifications to enhance the performance of the end-to-end (E2E) variational network (VarNet) for undersampled MRI reconstructions. We first implemented the Feature VarNet, which propagates information throughout the cascades of the network in an N-channel feature-space instead of a 2-channel feature-space. Then, we add an attention layer that utilizes the spatial locations of Cartesian undersampling artifacts to further improve performance. Lastly, we combined the Feature and E2E VarNets into the Feature-Image (FI) VarNet, to facilitate cross-domain learning and boost accuracy. Reconstructions were evaluated on the fastMRI dataset using standard metrics and clinical scoring by three neuroradiologists. Feature and FI VarNets outperformed the E2E VarNet for 4 × , 5 × and 8 × Cartesian undersampling in all studied metrics. FI VarNet secured second place in the public fastMRI leaderboard for 4 × Cartesian undersampling, outperforming all open-source models in the leaderboard. Radiologists rated FI VarNet brain reconstructions with higher quality and sharpness than the E2E VarNet reconstructions. FI VarNet excelled in preserving anatomical details, including blood vessels, whereas E2E VarNet discarded or blurred them in some cases. The proposed FI VarNet enhances the reconstruction quality of undersampled MRI and could enable clinically acceptable reconstructions at higher acceleration factors than currently possible.

AI-based motion artifact severity estimation in undersampled MRI allowing for selection of appropriate reconstruction models

BACKGROUND

Magnetic Resonance acquisition is a time consuming process, making it susceptible to patient motion during scanning. Even motion in the order of a millimeter can introduce severe blurring and ghosting artifacts, potentially necessitating re-acquisition. Magnetic Resonance Imaging (MRI) can be accelerated by acquiring only a fraction of k-space, combined with advanced reconstruction techniques leveraging coil sensitivity profiles and prior knowledge. Artificial intelligence (AI)-based reconstruction techniques have recently been popularized, but generally assume an ideal setting without intra-scan motion.

PURPOSE

To retrospectively detect and quantify the severity of motion artifacts in undersampled MRI data. This may prove valuable as a safety mechanism for AI-based approaches, provide useful information to the reconstruction method, or prompt for re-acquisition while the patient is still in the scanner.

METHODS

We developed a deep learning approach that detects and quantifies motion artifacts in undersampled brain MRI. We demonstrate that synthetically motion-corrupted data can be leveraged to train the convolutional neural network (CNN)-based motion artifact estimator, generalizing well to real-world data. Additionally, we leverage the motion artifact estimator by using it as a selector for a motion-robust reconstruction model in case a considerable amount of motion was detected, and a high data consistency model otherwise.

RESULTS

Training and validation were performed on 4387 and 1304 synthetically motion-corrupted images and their uncorrupted counterparts, respectively. Testing was performed on undersampled in vivo motion-corrupted data from 28 volunteers, where our model distinguished head motion from motion-free scans with 91% and 96% accuracy when trained on synthetic and on real data, respectively. It predicted a manually defined quality label ('Good', 'Medium' or 'Bad' quality) correctly in 76% and 85% of the time when trained on synthetic and real data, respectively. When used as a selector it selected the appropriate reconstruction network 93% of the time, achieving near optimal SSIM values.

CONCLUSIONS

The proposed method quantified motion artifact severity in undersampled MRI data with high accuracy, enabling real-time motion artifact detection that can help improve the safety and quality of AI-based reconstructions.

Predicting dynamic, motion-related changes in B0 field in the brain at a 7T MRI using a subject-specific fine-trained U-net

PURPOSE

Subject movement during the MR examination is inevitable and causes not only image artifacts but also deteriorates the homogeneity of the main magnetic field (B0 ), which is a prerequisite for high quality data. Thus, characterization of changes to B0 , for example induced by patient movement, is important for MR applications that are prone to B0 inhomogeneities.

METHODS

We propose a deep learning based method to predict such changes within the brain from the change of the head position to facilitate retrospective or even real-time correction. A 3D U-net was trained on in vivo gradient-echo brain 7T MRI data. The input consisted of B0 maps and anatomical images at an initial position, and anatomical images at a different head position (obtained by applying a rigid-body transformation on the initial anatomical image). The output consisted of B0 maps at the new head positions. We further fine-trained the network weights to each subject by measuring a limited number of head positions of the given subject, and trained the U-net with these data.

RESULTS

Our approach was compared to established dynamic B0 field mapping via interleaved navigators, which suffer from limited spatial resolution and the need for undesirable sequence modifications. Qualitative and quantitative comparison showed similar performance between an interleaved navigator-equivalent method and proposed method.

CONCLUSION

It is feasible to predict B0 maps from rigid subject movement and, when combined with external tracking hardware, this information could be used to improve the quality of MR acquisitions without the use of navigators.

FastMRI Prostate: A public, biparametric MRI dataset to advance machine learning for prostate cancer imaging

Magnetic resonance imaging (MRI) has experienced remarkable advancements in the integration of artificial intelligence (AI) for image acquisition and reconstruction. The availability of raw k-space data is crucial for training AI models in such tasks, but public MRI datasets are mostly restricted to DICOM images only. To address this limitation, the fastMRI initiative released brain and knee k-space datasets, which have since seen vigorous use. In May 2023, fastMRI was expanded to include biparametric (T2- and diffusion-weighted) prostate MRI data from a clinical population. Biparametric MRI plays a vital role in the diagnosis and management of prostate cancer. Advances in imaging methods, such as reconstructing under-sampled data from accelerated acquisitions, can improve cost-effectiveness and accessibility of prostate MRI. Raw k-space data, reconstructed images and slice, volume and exam level annotations for likelihood of prostate cancer are provided in this dataset for 47468 slices corresponding to 1560 volumes from 312 patients. This dataset facilitates AI and algorithm development for prostate image reconstruction, with the ultimate goal of enhancing prostate cancer diagnosis.

Compressed Sensitivity Encoding Artificial Intelligence Accelerates Brain Metastasis Imaging by Optimizing Image Quality and Reducing Scan Time

BACKGROUND AND PURPOSE

Accelerating the image acquisition speed of MR imaging without compromising the image quality is challenging. This study aimed to evaluate the feasibility of contrast-enhanced (CE) 3D T1WI and CE 3D-FLAIR sequences reconstructed with compressed sensitivity encoding artificial intelligence (CS-AI) for detecting brain metastases (BM) and explore the optimal acceleration factor (AF) for clinical BM imaging.

MATERIALS AND METHODS

Fifty-one patients with cancer with suspected BM were included. Fifty participants underwent different customized CE 3D-T1WI or CE 3D-FLAIR sequence scans. Compressed SENSE encoding acceleration 6 (CS6), a commercially available standard sequence, was used as the reference standard. Quantitative and qualitative methods were used to evaluate image quality. The SNR and contrast-to-noise ratio (CNR) were calculated, and qualitative evaluations were independently conducted by 2 neuroradiologists. After exploring the optimal AF, sample images were obtained from 1 patient by using both optimized sequences.

RESULTS

Quantitatively, the CNR of the CS-AI protocol for CE 3D-T1WI and CE 3D-FLAIR sequences was superior to that of the CS protocol under the same AF (P < .05). Compared with reference CS6, the CS-AI groups had higher CNR values (all P < .05), with the CS-AI10 scan having the highest value. The SNR of the CS-AI group was better than that of the reference for both CE 3D-T1WI and CE 3D-FLAIR sequences (all P < .05). Qualitatively, the CS-AI protocol produced higher image quality scores than did the CS protocol with the same AF (all P < .05). In contrast to the reference CS6, the CS-AI group showed good image quality scores until an AF of up to 10 (all P < .05). The CS-AI10 scan provided the optimal images, improving the delineation of normal gray-white matter boundaries and lesion areas (P < .05). Compared with the reference, CS-AI10 showed reductions in scan time of 39.25% and 39.93% for CE 3D-T1WI and CE 3D-FLAIR sequences, respectively.

CONCLUSIONS

CE 3D-T1WI and CE 3D-FLAIR sequences reconstructed with CS-AI for the detection of BM may provide a more effective alternative reconstruction approach than CS. CS-AI10 is suitable for clinical applications, providing optimal image quality and a shortened scan time.

Five-minute knee MRI: An AI-based super resolution reconstruction approach for compressed sensing. A validation study on healthy volunteers

PURPOSE

To investigate the potential of combining Compressed Sensing (CS) and a newly developed AI-based super resolution reconstruction prototype consisting of a series of convolutional neural networks (CNN) for a complete five-minute 2D knee MRI protocol.

METHODS

In this prospective study, 20 volunteers were examined using a 3T-MRI-scanner (Ingenia Elition X, Philips). Similar to clinical practice, the protocol consists of a fat-saturated 2D-proton-density-sequence in coronal, sagittal and transversal orientation as well as a sagittal T1-weighted sequence. The sequences were acquired with two different resolutions (standard and low resolution) and the raw data reconstructed with two different reconstruction algorithms: a conventional Compressed SENSE (CS) and a new CNN-based algorithm for denoising and subsequently to interpolate and therewith increase the sharpness of the image (CS-SuperRes). Subjective image quality was evaluated by two blinded radiologists reviewing 8 criteria on a 5-point Likert scale and signal-to-noise ratio calculated as an objective parameter.

RESULTS

The protocol reconstructed with CS-SuperRes received higher ratings than the time-equivalent CS reconstructions, statistically significant especially for low resolution acquisitions (e.g., overall image impression: 4.3 ± 0.4 vs. 3.4 ± 0.4, p < 0.05). CS-SuperRes reconstructions for the low resolution acquisition were comparable to traditional CS reconstructions with standard resolution for all parameters, achieving a scan time reduction from 11:01 min to 4:46 min (57 %) for the complete protocol (e.g. overall image impression: 4.3 ± 0.4 vs. 4.0 ± 0.5, p < 0.05).

CONCLUSION

The newly-developed AI-based reconstruction algorithm CS-SuperRes allows to reduce scan time by 57% while maintaining unchanged image quality compared to the conventional CS reconstruction.

Deep-learning-based reconstruction of undersampled MRI to reduce scan times: a multicentre, retrospective, cohort study

BACKGROUND

The extended acquisition times required for MRI limit its availability in resource-constrained settings. Consequently, accelerating MRI by undersampling k-space data, which is necessary to reconstruct an image, has been a long-standing but important challenge. We aimed to develop a deep convolutional neural network (dCNN) optimisation method for MRI reconstruction and to reduce scan times and evaluate its effect on image quality and accuracy of oncological imaging biomarkers.

METHODS

In this multicentre, retrospective, cohort study, MRI data from patients with glioblastoma treated at Heidelberg University Hospital (775 patients and 775 examinations) and from the phase 2 CORE trial (260 patients, 1083 examinations, and 58 institutions) and the phase 3 CENTRIC trial (505 patients, 3147 examinations, and 139 institutions) were used to develop, train, and test dCNN for reconstructing MRI from highly undersampled single-coil k-space data with various acceleration rates (R=2, 4, 6, 8, 10, and 15). Independent testing was performed with MRIs from the phase 2/3 EORTC-26101 trial (528 patients with glioblastoma, 1974 examinations, and 32 institutions). The similarity between undersampled dCNN-reconstructed and original MRIs was quantified with various image quality metrics, including structural similarity index measure (SSIM) and the accuracy of undersampled dCNN-reconstructed MRI on downstream radiological assessment of imaging biomarkers in oncology (automated artificial intelligence-based quantification of tumour burden and treatment response) was performed in the EORTC-26101 test dataset. The public NYU Langone Health fastMRI brain test dataset (558 patients and 558 examinations) was used to validate the generalisability and robustness of the dCNN for reconstructing MRIs from available multi-coil (parallel imaging) k-space data.

FINDINGS

In the EORTC-26101 test dataset, the median SSIM of undersampled dCNN-reconstructed MRI ranged from 0·88 to 0·99 across different acceleration rates, with 0·92 (95% CI 0·92-0·93) for 10-times acceleration (R=10). The 10-times undersampled dCNN-reconstructed MRI yielded excellent agreement with original MRI when assessing volumes of contrast-enhancing tumour (median DICE for spatial agreement of 0·89 [95% CI 0·88 to 0·89]; median volume difference of 0·01 cm3 [95% CI 0·00 to 0·03] equalling 0·21%; p=0·0036 for equivalence) or non-enhancing tumour or oedema (median DICE of 0·94 [95% CI 0·94 to 0·95]; median volume difference of -0·79 cm3 [95% CI -0·87 to -0·72] equalling -1·77%; p=0·023 for equivalence) in the EORTC-26101 test dataset. Automated volumetric tumour response assessment in the EORTC-26101 test dataset yielded an identical median time to progression of 4·27 months (95% CI 4·14 to 4·57) when using 10-times-undersampled dCNN-reconstructed or original MRI (log-rank p=0·80) and agreement in the time to progression in 374 (95·2%) of 393 patients with data. The dCNN generalised well to the fastMRI brain dataset, with significant improvements in the median SSIM when using multi-coil compared with single-coil k-space data (p<0·0001).

INTERPRETATION

Deep-learning-based reconstruction of undersampled MRI allows for a substantial reduction of scan times, with a 10-times acceleration demonstrating excellent image quality while preserving the accuracy of derived imaging biomarkers for the assessment of oncological treatment response. Our developments are available as open source software and hold considerable promise for increasing the accessibility to MRI, pending further prospective validation.

FUNDING

Deutsche Forschungsgemeinschaft (German Research Foundation) and an Else Kröner Clinician Scientist Endowed Professorship by the Else Kröner Fresenius Foundation.

Domain transformation learning for MR image reconstruction from dual domain input

Medical images are acquired through diverse imaging systems, with each system employing specific image reconstruction techniques to transform sensor data into images. In MRI, sensor data (i.e., k-space data) is encoded in the frequency domain, and fully sampled k-space data is transformed into an image using the inverse Fourier Transform. However, in efforts to reduce acquisition time, k-space is often subsampled, necessitating a sophisticated image reconstruction method beyond a simple transform. The proposed approach addresses this challenge by training a model to learn domain transform, generating the final image directly from undersampled k-space input. Significantly, to improve the stability of reconstruction from randomly subsampled k-space data, folded images are incorporated as supplementary inputs in the dual-input ETER-net. Moreover, modifications are made to the formation of inputs for the bi-RNN stages to accommodate non-fixed k-space trajectories. Experimental validation, encompassing both regular and irregular sampling trajectories, validates the method's effectiveness. The results demonstrated superior performance, measured by PSNR, SSIM, and VIF, across acceleration factors of 4 and 8. In summary, the dual-input ETER-net emerges as an effective both regular and irregular sampling trajectories, and accommodating diverse acceleration factors.

HyperSLICE: HyperBand optimized spiral for low-latency interactive cardiac examination

PURPOSE

Interactive cardiac MRI is used for fast scan planning and MR-guided interventions. However, the requirement for real-time acquisition and near-real-time visualization constrains the achievable spatio-temporal resolution. This study aims to improve interactive imaging resolution through optimization of undersampled spiral sampling and leveraging of deep learning for low-latency reconstruction (deep artifact suppression).

METHODS

A variable density spiral trajectory was parametrized and optimized via HyperBand to provide the best candidate trajectory for rapid deep artifact suppression. Training data consisted of 692 breath-held CINEs. The developed interactive sequence was tested in simulations and prospectively in 13 subjects (10 for image evaluation, 2 during catheterization, 1 during exercise). In the prospective study, the optimized framework-HyperSLICE- was compared with conventional Cartesian real-time and breath-hold CINE imaging in terms quantitative and qualitative image metrics. Statistical differences were tested using Friedman chi-squared tests with post hoc Nemenyi test (p < 0.05).

RESULTS

In simulations the normalized RMS error, peak SNR, structural similarity, and Laplacian energy were all statistically significantly higher using optimized spiral compared to radial and uniform spiral sampling, particularly after scan plan changes (structural similarity: 0.71 vs. 0.45 and 0.43). Prospectively, HyperSLICE enabled a higher spatial and temporal resolution than conventional Cartesian real-time imaging. The pipeline was demonstrated in patients during catheter pull back, showing sufficiently fast reconstruction for interactive imaging.

CONCLUSION

HyperSLICE enables high spatial and temporal resolution interactive imaging. Optimizing the spiral sampling enabled better overall image quality and superior handling of image transitions compared with radial and uniform spiral trajectories.

How AI May Transform Musculoskeletal Imaging

While musculoskeletal imaging volumes are increasing, there is a relative shortage of subspecialized musculoskeletal radiologists to interpret the studies. Will artificial intelligence (AI) be the solution? For AI to be the solution, the wide implementation of AI-supported data acquisition methods in clinical practice requires establishing trusted and reliable results. This implementation will demand close collaboration between core AI researchers and clinical radiologists. Upon successful clinical implementation, a wide variety of AI-based tools can improve the musculoskeletal radiologist's workflow by triaging imaging examinations, helping with image interpretation, and decreasing the reporting time. Additional AI applications may also be helpful for business, education, and research purposes if successfully integrated into the daily practice of musculoskeletal radiology. The question is not whether AI will replace radiologists, but rather how musculoskeletal radiologists can take advantage of AI to enhance their expert capabilities.

Sharing Data Is Essential for the Future of AI in Medical Imaging

If we want artificial intelligence to succeed in radiology, we must share data and learn how to share data.

Feasibility of AI-assisted compressed sensing protocols in knee MR imaging: a prospective multi-reader study

OBJECTIVES

To evaluate the image quality and diagnostic performance of AI-assisted compressed sensing (ACS) accelerated two-dimensional fast spin-echo MRI compared with standard parallel imaging (PI) in clinical 3.0T rapid knee scans.

METHODS

This prospective study enrolled 130 consecutive participants between March and September 2022. The MRI scan procedure included one 8.0-min PI protocol and two ACS protocols (3.5 min and 2.0 min). Quantitative image quality assessments were performed by evaluating edge rise distance (ERD) and signal-to-noise ratio (SNR). Shapiro-Wilk tests were performed and investigated by the Friedman test and post hoc analyses. Three radiologists independently evaluated structural disorders for each participant. Fleiss κ analysis was used to compare inter-reader and inter-protocol agreements. The diagnostic performance of each protocol was investigated and compared by DeLong's test. The threshold for statistical significance was set at p  < 0.05.

RESULTS

A total of 150 knee MRI examinations constituted the study cohort. For the quantitative assessment of four conventional sequences with ACS protocols, SNR improved significantly (p < 0.001), and ERD was significantly reduced or equivalent to the PI protocol. For the abnormality evaluated, the intraclass correlation coefficient ranged from moderate to substantial between readers (κ = 0.75-0.98) and between protocols (κ = 0.73-0.98). For meniscal tears, cruciate ligament tears, and cartilage defects, the diagnostic performance of ACS protocols was considered equivalent to PI protocol (Delong test, p > 0.05).

CONCLUSIONS

Compared with the conventional PI acquisition, the novel ACS protocol demonstrated superior image quality and was feasible for achieving equivalent detection of structural abnormalities while reducing acquisition time by half.

CLINICAL RELEVANCE STATEMENT

Artificial intelligence-assisted compressed sensing (ACS) providing excellent quality and a 75% reduction in scanning time presents significant clinical advantages in improving the efficiency and accessibility of knee MRI for more patients.

KEY POINTS

• The prospective multi-reader study showed no difference in diagnostic performance between parallel imaging and AI-assisted compression sensing (ACS) was found. • Reduced scan time, sharper delineation, and less noise with ACS reconstruction. • Improved efficiency of the clinical knee MRI examination by the ACS acceleration.

Morphological assessment of cartilage and osteoarthritis in clinical practice and research: Intermediate-weighted fat-suppressed sequences and beyond

Magnetic resonance imaging (MRI) is widely regarded as the primary modality for the morphological assessment of cartilage and all other joint tissues involved in osteoarthritis. 2D fast spin echo fat-suppressed intermediate-weighted (FSE FS IW) sequences with a TE between 30 and 40ms have stood the test of time and are considered the cornerstone of MRI protocols for clinical practice and trials. These sequences offer a good balance between sensitivity and specificity and provide appropriate contrast and signal within the cartilage as well as between cartilage, articular fluid, and subchondral bone. Additionally, FS IW sequences enable the evaluation of menisci, ligaments, synovitis/effusion, and bone marrow edema-like signal changes. This review article provides a rationale for the use of FSE FS IW sequences in the morphological assessment of cartilage and osteoarthritis, along with a brief overview of other clinically available sequences for this indication. Additionally, the article highlights ongoing research efforts aimed at improving FSE FS IW sequences through 3D acquisitions with enhanced resolution, shortened examination times, and exploring the potential benefits of different magnetic field strengths. While most of the literature on cartilage imaging focuses on the knee, the concepts presented here are applicable to all joints. KEY POINTS: 1. MRI is currently considered the modality of reference for a "whole-joint" morphological assessment of osteoarthritis. 2. Fat-suppressed intermediate-weighted sequences remain the keystone of MRI protocols for the assessment of cartilage morphology, as well as other structures involved in osteoarthritis. 3. Trends for further development in the field of cartilage and joint imaging include 3D FSE imaging, faster acquisition including AI-based acceleration, and synthetic imaging providing multi-contrast sequences.

Optimal MRI undersampling patterns for ultimate benefit of medical vision tasks

Compressed sensing is commonly concerned with optimizing the image quality after a partial undersampling of the measurable k-space to accelerate MRI. In this article, we propose to change the focus from the quality of the reconstructed image to the quality of the downstream image analysis outcome. Specifically, we propose to optimize the patterns according to how well a sought-after pathology could be detected or localized in the reconstructed images. We find the optimal undersampling patterns in k-space that maximize target value functions of interest in commonplace medical vision problems (reconstruction, segmentation, and classification) and propose a new iterative gradient sampling routine universally suitable for these tasks. We validate the proposed MRI acceleration paradigm on three classical medical datasets, demonstrating a noticeable improvement of the target metrics at the high acceleration factors (for the segmentation problem at ×16 acceleration, we report up to 12% improvement in Dice score over the other undersampling patterns).

Evaluation of a Deep Learning Reconstruction for High-Quality T2-Weighted Breast Magnetic Resonance Imaging

Deep learning (DL) reconstruction techniques to improve MR image quality are becoming commercially available with the hope that they will be applicable to multiple imaging application sites and acquisition protocols. However, before clinical implementation, these methods must be validated for specific use cases. In this work, the quality of standard-of-care (SOC) T2w and a high-spatial-resolution (HR) imaging of the breast were assessed both with and without prototype DL reconstruction. Studies were performed using data collected from phantoms, 20 retrospectively collected SOC patient exams, and 56 prospectively acquired SOC and HR patient exams. Image quality was quantitatively assessed via signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and edge sharpness. Qualitatively, all in vivo images were scored by either two or four radiologist readers using 5-point Likert scales in the following categories: artifacts, perceived sharpness, perceived SNR, and overall quality. Differences in reader scores were tested for significance. Reader preference and perception of signal intensity changes were also assessed. Application of the DL resulted in higher average SNR (1.2-2.8 times), CNR (1.0-1.8 times), and image sharpness (1.2-1.7 times). Qualitatively, the SOC acquisition with DL resulted in significantly improved image quality scores in all categories compared to non-DL images. HR acquisition with DL significantly increased SNR, sharpness, and overall quality compared to both the non-DL SOC and the non-DL HR images. The acquisition time for the HR data only required a 20% increase compared to the SOC acquisition and readers typically preferred DL images over non-DL counterparts. Overall, the DL reconstruction demonstrated improved T2w image quality in clinical breast MRI.

Image Quality and Diagnostic Performance of Accelerated 2D Hip MRI with Deep Learning Reconstruction Based on a Deep Iterative Hierarchical Network

OBJECTIVES

Hip MRI using standard multiplanar sequences requires long scan times. Accelerating MRI is accompanied by reduced image quality. This study aimed to compare standard two-dimensional (2D) turbo spin echo (TSE) sequences with accelerated 2D TSE sequences with deep learning (DL) reconstruction (TSEDL) for routine clinical hip MRI at 1.5 and 3 T in terms of feasibility, image quality, and diagnostic performance.

MATERIAL AND METHODS

In this prospective, monocentric study, TSEDL was implemented clinically and evaluated in 14 prospectively enrolled patients undergoing a clinically indicated hip MRI at 1.5 and 3T between October 2020 and May 2021. Each patient underwent two examinations: For the first exam, we used standard sequences with generalized autocalibrating partial parallel acquisition reconstruction (TSES). For the second exam, we implemented prospectively undersampled TSE sequences with DL reconstruction (TSEDL). Two radiologists assessed the TSEDL and TSES regarding image quality, artifacts, noise, edge sharpness, diagnostic confidence, and delineation of anatomical structures using an ordinal five-point Likert scale (1 = non-diagnostic; 2 = poor; 3 = moderate; 4 = good; 5 = excellent). Both sequences were compared regarding the detection of common pathologies of the hip. Comparative analyses were conducted to assess the differences between TSEDL and TSES.

RESULTS

Compared with TSES, TSEDL was rated to be significantly superior in terms of image quality (p ≤ 0.020) with significantly reduced noise (p ≤ 0.001) and significantly improved edge sharpness (p = 0.003). No difference was found between TSES and TSEDL concerning the extent of artifacts, diagnostic confidence, or the delineation of anatomical structures (p > 0.05). Example acquisition time reductions for the TSE sequences of 52% at 3 Tesla and 70% at 1.5 Tesla were achieved.

CONCLUSION

TSEDL of the hip is clinically feasible, showing excellent image quality and equivalent diagnostic performance compared with TSES, reducing the acquisition time significantly.

Complexities of deep learning-based undersampled MR image reconstruction

Artificial intelligence has opened a new path of innovation in magnetic resonance (MR) image reconstruction of undersampled k-space acquisitions. This review offers readers an analysis of the current deep learning-based MR image reconstruction methods. The literature in this field shows exponential growth, both in volume and complexity, as the capabilities of machine learning in solving inverse problems such as image reconstruction are explored. We review the latest developments, aiming to assist researchers and radiologists who are developing new methods or seeking to provide valuable feedback. We shed light on key concepts by exploring the technical intricacies of MR image reconstruction, highlighting the importance of raw datasets and the difficulty of evaluating diagnostic value using standard metrics.Relevance statement Increasingly complex algorithms output reconstructed images that are difficult to assess for robustness and diagnostic quality, necessitating high-quality datasets and collaboration with radiologists.Key points• Deep learning-based image reconstruction algorithms are increasing both in complexity and performance.• The evaluation of reconstructed images may mistake perceived image quality for diagnostic value.• Collaboration with radiologists is crucial for advancing deep learning technology.

Artificial Intelligence in Breast Cancer Diagnosis and Personalized Medicine

Breast cancer is a significant cause of cancer-related mortality in women worldwide. Early and precise diagnosis is crucial, and clinical outcomes can be markedly enhanced. The rise of artificial intelligence (AI) has ushered in a new era, notably in image analysis, paving the way for major advancements in breast cancer diagnosis and individualized treatment regimens. In the diagnostic workflow for patients with breast cancer, the role of AI encompasses screening, diagnosis, staging, biomarker evaluation, prognostication, and therapeutic response prediction. Although its potential is immense, its complete integration into clinical practice is challenging. Particularly, these challenges include the imperatives for extensive clinical validation, model generalizability, navigating the "black-box" conundrum, and pragmatic considerations of embedding AI into everyday clinical environments. In this review, we comprehensively explored the diverse applications of AI in breast cancer care, underlining its transformative promise and existing impediments. In radiology, we specifically address AI in mammography, tomosynthesis, risk prediction models, and supplementary imaging methods, including magnetic resonance imaging and ultrasound. In pathology, our focus is on AI applications for pathologic diagnosis, evaluation of biomarkers, and predictions related to genetic alterations, treatment response, and prognosis in the context of breast cancer diagnosis and treatment. Our discussion underscores the transformative potential of AI in breast cancer management and emphasizes the importance of focused research to realize the full spectrum of benefits of AI in patient care.

Snapshot CEST++: Advancing rapid whole-brain APTw-CEST MRI at 3 T

APTw CEST MRI suffers from long preparation times and consequently long acquisition times (~5 min). Recently, a consensus on the preparation module for clinical APTw CEST at 3 T was found in the community, and we present a fast whole-brain APTw CEST MRI sequence following this consensus preparation of pulsed RF irradiation of 2 s duration at 90% RF duty-cycle and a B1,rms of 2 μT. After optimization of the snapshot CEST approach for APTw imaging regarding flip angle, voxel size and frequency offset sampling, we extend it by undersampled GRE acquisition and compressed sensing reconstruction. This allows 2 mm isotropic whole-brain APTw imaging for clinical research at 3 T below 2 min. With this sequence, a fast snapshot APTw imaging method is now available for larger clinical studies of brain tumors.

Deep learning enabled fast 3D brain MRI at 0.055 tesla

In recent years, there has been an intensive development of portable ultralow-field magnetic resonance imaging (MRI) for low-cost, shielding-free, and point-of-care applications. However, its quality is poor and scan time is long. We propose a fast acquisition and deep learning reconstruction framework to accelerate brain MRI at 0.055 tesla. The acquisition consists of a single average three-dimensional (3D) encoding with 2D partial Fourier sampling, reducing the scan time of T1- and T2-weighted imaging protocols to 2.5 and 3.2 minutes, respectively. The 3D deep learning leverages the homogeneous brain anatomy available in high-field human brain data to enhance image quality, reduce artifacts and noise, and improve spatial resolution to synthetic 1.5-mm isotropic resolution. Our method successfully overcomes low-signal barrier, reconstructing fine anatomical structures that are reproducible within subjects and consistent across two protocols. It enables fast and quality whole-brain MRI at 0.055 tesla, with potential for widespread biomedical applications.

An open competition involving thousands of competitors failed to construct useful abstract classifiers for new diagnostic test accuracy systematic reviews

There are currently no abstract classifiers, which can be used for new diagnostic test accuracy (DTA) systematic reviews to select primary DTA study abstracts from database searches. Our goal was to develop machine-learning-based abstract classifiers for new DTA systematic reviews through an open competition. We prepared a dataset of abstracts obtained through database searches from 11 reviews in different clinical areas. As the reference standard, we used the abstract lists that required manual full-text review. We randomly splitted the datasets into a train set, a public test set, and a private test set. Competition participants used the training set to develop classifiers and validated their classifiers using the public test set. The classifiers were refined based on the performance of the public test set. They could submit as many times as they wanted during the competition. Finally, we used the private test set to rank the submitted classifiers. To reduce false exclusions, we used the Fbeta measure with a beta set to seven for evaluating classifiers. After the competition, we conducted the external validation using a dataset from a cardiology DTA review. We received 13,774 submissions from 1429 teams or persons over 4 months. The top-honored classifier achieved a Fbeta score of 0.4036 and a recall of 0.2352 in the external validation. In conclusion, we were unable to develop an abstract classifier with sufficient recall for immediate application to new DTA systematic reviews. Further studies are needed to update and validate classifiers with datasets from other clinical areas.