25 research outputs found
The cause and effect of an MR image:Robustness and generalizability
No full textThis chapter discusses issues regarding the robustness and generalizability of AI models applied to medical imaging, specifically magnetic resonance imaging (MRI) data. We consider the implications of domain shifts on machine learning models and uncover challenges with respect to fairness, algorithmic bias, and shortcut learning. We also address possible mitigation techniques and solutions to make AI models more robust and generalizable. The notion of model invariance with respect to nuisance factors, such as the scanner model used, is introduced, and its connections to causal modeling and statistical independence statements are elucidated. Throughout the chapter, we give examples from the medical imaging and MRI literature that highlight the problems as well as point toward potential solutions. Our findings emphasize the substantial risk of generalization problems in this domain
CS_LAB: Initial release
No full text<p>Acquisition:</p>
<ul>
<li>subsampling class v1.0</li>
<li>CS-FLASH (VB20P) v1.2.1</li>
</ul>
<p>Reconstruction:</p>
<ul>
<li>c++: Gadget for Gadgetron v1.0 + interface to matlab (via mex) v2.0</li>
<li>matlab: v1.0.1</li>
</ul>https://sites.google.com/site/kspaceastronauts/compressed-sensing/csla
Proline-rich tyrosine kinase 2 mediates gonadotropin-releasing hormone signaling to a specific extracellularly regulated kinase-sensitive transcriptional locus in the luteinizing hormone beta-subunit gene
No full textG protein-coupled receptor regulation of gene transcription primarily occurs through the phosphorylation of transcription factors by MAPKs. This requires transduction of an activating signal via scaffold proteins that can ultimately determine the outcome by binding signaling kinases and adapter proteins with effects on the target transcription factor and locus of activation. By investigating these mechanisms, we have elucidated how pituitary gonadotrope cells decode an input GnRH signal into coherent transcriptional output from the LH beta-subunit gene promoter. We show that GnRH activates c-Src and multiple members of the MAPK family, c-Jun NH2-terminal kinase 1/2, p38MAPK, and ERK1/2. Using dominant-negative point mutations and chemical inhibitors, we identified that calcium-dependent proline-rich tyrosine kinase 2 specifically acts as a scaffold for a focal adhesion/cytoskeleton-dependent complex comprised of c-Src, Grb2, and mSos that translocates an ERK-activating signal to the nucleus. The locus of action of ERK was specifically mapped to early growth response-1 (Egr-1) DNA binding sites within the LH beta-subunit gene proximal promoter, which was also activated by p38MAPK, but not c-Jun NH2-terminal kinase 1/2. Egr-1 was confirmed as the transcription factor target of ERK and p38MAPK by blockade of protein expression, transcriptional activity, and DNA binding. We have identified a novel GnRH-activated proline-rich tyrosine kinase 2-dependent ERK-mediated signal transduction pathway that specifically regulates Egr-1 activation of the LH beta-subunit proximal gene promoter, and thus provide insight into the molecular mechanisms required for differential regulation of gonadotropin gene expression
Accelerated 4D Respiratory Motion-Resolved Cardiac MRI with a Model-Based Variational Network
No full textRespiratory motion and long scan times remain major challenges in free-breathing 3D cardiac MRI. Respiratory motion-resolved approaches have been proposed by binning the acquired data to different respiratory motion states. After inter-bin motion estimation, motion-compensated reconstruction can be obtained. However, respiratory bins from accelerated acquisitions are highly undersampled and have different undersampling patterns depending on the subject-specific respiratory motion. Remaining undersampling artifacts in the bin images can influence the accuracy of the motion estimation. We propose a model-based variational network (VN) which reconstructs motion-resolved images jointly by exploiting shared information between respiratory bins. In each stage of VN, conjugate gradient is adopted to enforce data-consistency (CG-VN), achieving better enforcement of data consistency per stage than the classic VN with proximal gradient descent step (GD-VN), translating to faster convergence and better reconstruction performance. We compare the performance of CG-VN and GD-VN for reconstruction of respiratory motion-resolved images for two different cardiac MR sequences. Our results show that CG-VN with less stages outperforms GD-VN by achieving higher PSNR and better generalization on prospectively undersampled data. The proposed motion-resolved CG-VN provides consistently good reconstruction quality for all motion states with varying undersampling patterns by taking advantage of redundancies among motion bins.</p
Acceleration of Magnetic Resonance Cholangiopancreatography Using Compressed Sensing at 1.5 and 3 T
No full textAttention Incorporated Network for Sharing Low-rank, Image and K-space Information during MR Image Reconstruction to Achieve Single Breath-hold Cardiac Cine Imaging
No full textCardiac Cine Magnetic Resonance Imaging (MRI) provides an accurate assessment of heart morphology and function in clinical practice. However, MRI requires long acquisition times, with recent deep learning-based methods showing great promise to accelerate imaging and enhance reconstruction quality. Existing networks exhibit some common limitations that constrain further acceleration possibilities, including single-domain learning, reliance on a single regularization term, and equal feature contribution. To address these limitations, we propose to embed information from multiple domains, including low-rank, image, and k-space, in a novel deep learning network for MRI reconstruction, which we denote as A-LIKNet. A-LIKNet adopts a parallel-branch structure, enabling independent learning in the k-space and image domain. Coupled information sharing layers realize the information exchange between domains. Furthermore, we introduce attention mechanisms into the network to assign greater weights to more critical coils or important temporal frames. Training and testing were conducted on an in-house dataset, including 91 cardiovascular patients and 38 healthy subjects scanned with 2D cardiac Cine using retrospective undersampling. Additionally, we evaluated A-LIKNet on the real-time 8x prospectively undersampled data from the OCMR dataset. The results demonstrate that our proposed A-LIKNet outperforms existing methods and provides high-quality reconstructions. The network can effectively reconstruct highly retrospectively undersampled dynamic MR images up to 24x accelerations, indicating its potential for single breath-hold imaging
Self-supervised learning-based diffeomorphic non-rigid motion estimation for fast motion-compensated coronary MR angiography
No full textPurpose: To accelerate non-rigid motion corrected coronary MR angiography (CMRA) reconstruction by developing a deep learning based non-rigid motion estimation network and combining this with an efficient implementation of the undersampled motion corrected reconstruction.Methods: Undersampled and respiratory motion corrected CMRA with overall short scans of 5 to 10 min have been recently proposed. However, image reconstruction with this approach remains lengthy, since it relies on several non-rigid image registrations to estimate the respiratory motion and on a subsequent iterative optimization to correct for motion during the undersampled reconstruction. Here we introduce a self-supervised diffeomorphic non-rigid respiratory motion estimation network, DiRespME-net, to speed up respiratory motion estimation. We couple this with an efficient GPU-based implementation of the subsequent motion-corrected iterative reconstruction. DiRespME-net is based on a U-Net architecture, and is trained in a self-supervised fashion, with a loss enforcing image similarity and spatial smoothness of the motion fields. Motion predicted by DiRespME-net was used for GPU-based motion-corrected CMRA in 12 test subjects and final images were compared to those produced by state-of-the-art reconstruction. Vessel sharpness and visible length of the right coronary artery (RCA) and the left anterior descending (LAD) coronary artery were used as metrics of image quality for comparison.Results: No statistically significant difference in image quality was found between images reconstructed with the proposed approach (MC:DiRespME-net) and a motion-corrected reconstruction using cubic B-splines (MC:Niftyreg). Visible vessel length was not significantly different between methods (RCA: MC:Nifty-reg 5.7 +/- 1.7 cm vs MC:DiRespME-net 5.8 +/- 1.7 cm, P = 0.32; LAD: MC:Nifty-reg 7.0 +/- 2.6 cm vs MC:DiRespME-net 6.9 +/- 2.7 cm, P = 0.81). Similarly, no statistically significant difference between methods was observed in terms of vessel sharpness (RCA: MC:Nifty-reg 60.3 +/- 7.2% vs MC:DiRespME-net 61.0 +/- 6.8%, P = 0.19; LAD: MC:Nifty-reg 57.4 +/- 7.9% vs MC:DiRespME-net 58.1 +/- 7.5%, P = 0.27). The proposed approach achieved a 50-fold reduction in computation time, resulting in a total reconstruction time of approximately 20 s.Conclusions: The proposed self-supervised learning-based motion corrected reconstruction enables fast motion corrected CMRA image reconstruction, holding promise for integration in clinical routine
ESMRMB 2024 focus topic: MR beyond trends—fact-checking MR
No full textMagnetic resonance imaging (MRI) plays a key role in modern radiology with at least 100 million diagnostic scans performed annually across the globe. However, besides its undeniable value, the (multivariate) biophysical mechanisms underlying MRI contrast generation are rather unclear and subject of current research. Furthermore, with the advent of AI into all areas of medical imaging, MRI users now tend to employ black-box models in image reconstruction, segmentation, and disease classification, which in turn introduces an additional challenge in trust and interpretation of the results. Therefore, as part of this focus topic, we believe it is paramount to also shed light on the application of AI for MRI data interpretation, emphasizing the importance of explainable AI (xAI) in validating and understanding the obtained results. In addition, the educational sessions will cover the use of postmortem MRI studies to validate tissue models and the innovative approaches in biomarker discovery that rely on accurate and validated quantitative MRI techniques. Moreover, dedicated sessions will discuss the emerging applications of low-field MRI, and the vital role of validation
Generalized low‐rank nonrigid motion‐corrected reconstruction for MR fingerprinting
No full textPurpose: Develop a novel low-rank motion-corrected (LRMC) reconstruction for nonrigid motion-corrected MR fingerprinting (MRF). Methods: Generalized motion-corrected (MC) reconstructions have been developed for steady-state imaging. Here we extend this framework to enable nonrigid MC for transient imaging applications with varying contrast, such as MRF. This is achieved by integrating low-rank dictionary-based compression into the generalized MC model to reconstruct MC singular images, reducing motion artifacts in the resulting parametric maps. The proposed LRMC reconstruction was applied for cardiac motion correction in 2D myocardial MRF (T1 and T2) with extended cardiac acquisition window (~450 ms) and for respiratory MC in free-breathing 3D myocardial and 3D liver MRF. Experiments were performed in phantom and 22 healthy subjects. The proposed approach was compared with reference spin echo (phantom) and with 2D electrocardiogram-triggered/breath-hold MOLLI and T2 gradient-and–spin echo conventional maps (in vivo 2D and 3D myocardial MRF). Results: Phantom results were in general agreement with reference spin-echo measurements, presenting relative errors of approximately 5.4% and 5.5% for T1 and short T2 (<100 ms), respectively. The proposed LRMC MRF reduced residual blurring artifacts with respect to no MC for cardiac or respiratory motion in all cases (2D and 3D myocardial, 3D abdominal). In 2D myocardial MRF, left-ventricle T1 values were 1150 ± 41 ms for LRMC MRF and 1010 ± 56 ms for MOLLI; T2 values were 43.8 ± 2.3 ms for LRMC MRF and 49.5 ± 4.5 ms for T2 gradient and spin echo. Corresponding measurements for 3D myocardial MRF were 1085 ± 30 ms and 1062 ± 29 ms for T1, and 43.5 ± 1.9 ms and 51.7 ± 1.7 ms for T2. For 3D liver, LRMC MRF measured liver T1 at 565 ± 44 ms and liver T2 at 35.4 ± 2.4 ms. Conclusion: The proposed LRMC reconstruction enabled generalized (nonrigid) MC for 2D and 3D MRF, both for cardiac and respiratory motion. The proposed approach reduced motion artifacts in the MRF maps with respect to no motion compensation and achieved good agreement with reference measurements.</p
Scan time reduction in diffusion-weighted imaging of the pancreas using a simultaneous multislice technique with different acceleration factors: How fast can we go?
No full text
