Model-Based Deep Learning

Shlezinger, Nir; Whang, Jay; Eldar, Yonina C.; Dimakis, Alexandros G.

doi:10.1109/jproc.2023.3247480

Cited by 96 publications

(38 citation statements)

References 85 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Reliable and broadly applicable uncertainty measures for ML prediction are therefore crucial for clinical applicability of ML in MRS. Additionally, deploying hybrid models (combined model-based and data-driven systems) can allow ML contributions to be leveraged by physics-informed models that behave unbiased and have guarantees on their estimates. 98 The clinical utility of ML applications in MRS and MRSI is one of the most important aspects of this research field. Attempts to decrease human-expert involvement, decrease acquisition time, and increase robustness and generalizability of existing MRS tools are therefore essential.…”

Section: Discussionmentioning

confidence: 99%

“…In a clinical setting, such behaviors need to be detected and removed. Reliable and broadly applicable uncertainty measures for ML prediction are therefore crucial for clinical applicability of ML in MRS. Additionally, deploying hybrid models (combined model‐based and data‐driven systems) can allow ML contributions to be leveraged by physics‐informed models that behave unbiased and have guarantees on their estimates 98 …”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

A review of machine learning applications for the proton MR spectroscopy workflow

van de Sande,

Merkofer,

Amirrajab

et al. 2023

Magnetic Resonance in Med

View full text Add to dashboard Cite

This literature review presents a comprehensive overview of machine learning (ML) applications in proton MR spectroscopy (MRS). As the use of ML techniques in MRS continues to grow, this review aims to provide the MRS community with a structured overview of the state‐of‐the‐art methods. Specifically, we examine and summarize studies published between 2017 and 2023 from major journals in the MR field. We categorize these studies based on a typical MRS workflow, including data acquisition, processing, analysis, and artificial data generation. Our review reveals that ML in MRS is still in its early stages, with a primary focus on processing and analysis techniques, and less attention given to data acquisition. We also found that many studies use similar model architectures, with little comparison to alternative architectures. Additionally, the generation of artificial data is a crucial topic, with no consistent method for its generation. Furthermore, many studies demonstrate that artificial data suffers from generalization issues when tested on in vivo data. We also conclude that risks related to ML models should be addressed, particularly for clinical applications. Therefore, output uncertainty measures and model biases are critical to investigate. Nonetheless, the rapid development of ML in MRS and the promising results from the reviewed studies justify further research in this field.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

A review of machine learning applications for the proton MR spectroscopy workflow

van de Sande,

Merkofer,

Amirrajab

et al. 2023

Magnetic Resonance in Med

View full text Add to dashboard Cite

show abstract

“…End-to-end unrolled models are a family of powerful, data-driven recovery algorithms that have been successfully applied to medical imaging, as well as to other areas of computational science [ 6 ]. In the centralized setting, a large corpus of training samples are available at a central location, and an unrolled model is trained end to end.…”

Section: Theorymentioning

confidence: 99%

“…Given that sampling below the Nyquist rate gives rise to an ill-posed inverse problem, and hence can produce artifacts in reconstructed images, a wide area of research has been dedicated to image reconstruction algorithms for accelerated MRI. Recently, deep learning has achieved substantial improvements in reconstruction quality compared to classical methods [ 4 , 5 , 6 ], and is rapidly heading toward clinical validation [ 7 ].…”

Section: Introductionmentioning

confidence: 99%

Federated End-to-End Unrolled Models for Magnetic Resonance Image Reconstruction

2023

View full text Add to dashboard Cite

Image reconstruction is the process of recovering an image from raw, under-sampled signal measurements, and is a critical step in diagnostic medical imaging, such as magnetic resonance imaging (MRI). Recently, data-driven methods have led to improved image quality in MRI reconstruction using a limited number of measurements, but these methods typically rely on the existence of a large, centralized database of fully sampled scans for training. In this work, we investigate federated learning for MRI reconstruction using end-to-end unrolled deep learning models as a means of training global models across multiple clients (data sites), while keeping individual scans local. We empirically identify a low-data regime across a large number of heterogeneous scans, where a small number of training samples per client are available and non-collaborative models lead to performance drops. In this regime, we investigate the performance of adaptive federated optimization algorithms as a function of client data distribution and communication budget. Experimental results show that adaptive optimization algorithms are well suited for the federated learning of unrolled models, even in a limited-data regime (50 slices per data site), and that client-sided personalization can improve reconstruction quality for clients that did not participate in training.

show abstract

“…Unfortunately, DNNs are commonly utilized as a black box and data-driven deep learning does not yet offer the interpretability and reliability of model-based methods [34]. As an alternate that benefits from the advantages of both model-driven and data-driven paradigms, model-based deep learning methods [35] have attracted the attention of the MIMO communication research, which generally incorporate an internal or external DNN into an iterative algorithm. Compared with traditional algorithms implemented for an individual sample, they benefit from learning the domain knowledge and have shown a performance improvement while keeping relative interpretability [36]- [38].…”

Section: Mimo Systems Algorithmsmentioning

confidence: 99%

An ESPRIT-Based Supervised Channel Estimation Method Using Tensor Train Decomposition for mmWave 3-D MIMO-OFDM Systems

Xiao

Chen

Sun

et al. 2023

IEEE Trans. Signal Process.

View full text Add to dashboard Cite

CANDECOMP/PARAFAC (CP) decomposition is the mostly used model to formulate the received tensor signal in a multi-domain massive multiple-input multiple-output (MIMO) system, as the receiver generally sums the components from different paths or users. To achieve accurate and low-latency channel estimation, good and fast CP decomposition algorithms are desired. The CP alternating least squares (CPALS) is the workhorse algorithm for calculating the CP decomposition. However, its performance depends on the initializations, and good starting values can lead to more efficient solutions. Existing initialization strategies are decoupled from the CPALS and are not necessarily favorable for solving the CP decomposition. To enhance the algorithm's speed and accuracy, this paper proposes a deep-learning-aided CPALS (DL-CPALS) method that uses a deep neural network (DNN) to generate favorable initializations. The proposed DL-CPALS integrates the DNN and CPALS to a model-based deep learning paradigm, where it trains the DNN to generate an initialization that facilitates fast and accurate CP decomposition. Moreover, benefiting from the CP low-rankness, the proposed method is trained using noisy data and does not require paired clean data. The proposed DL-CPALS is applied to millimeter wave MIMO orthogonal frequency division multiplexing (mmWave MIMO-OFDM) channel estimation. Experimental results demonstrate the significant improvements of the proposed method in terms of both speed and accuracy for CP decomposition and channel estimation.

show abstract

Model-Based Deep Learning

Cited by 96 publications

References 85 publications

A review of machine learning applications for the proton MR spectroscopy workflow

A review of machine learning applications for the proton MR spectroscopy workflow

Federated End-to-End Unrolled Models for Magnetic Resonance Image Reconstruction

An ESPRIT-Based Supervised Channel Estimation Method Using Tensor Train Decomposition for mmWave 3-D MIMO-OFDM Systems

Contact Info

Product

Resources

About