Drug discovery and development pipelines are long, complex and depend on numerous factors. Machine learning (ML) approaches provide a set of tools that can improve discovery and decision making for well-specified questions with abundant, high-quality data. Opportunities to apply ML occur in all stages of drug discovery. Examples include target validation, identification of prognostic biomarkers and analysis of digital pathology data in clinical trials. Applications have ranged in context and methodology, with some approaches yielding accurate predictions and insights. The challenges of applying ML lie primarily with the lack of interpretability and repeatability of ML-generated results, which may limit their application. In all areas, systematic and comprehensive high-dimensional data still need to be generated. With ongoing efforts to tackle these issues, as well as increasing awareness of the factors needed to validate ML approaches, the application of ML can promote data-driven decision making and has the potential to speed up the process and reduce failure rates in drug discovery and development. Machine learning has been applied to numerous stages in the drug discovery pipeline. Here, Vamathevan and colleagues discuss the most useful techniques and how machine learning can promote data-driven decision making in drug discovery and development. They highlight major hurdles in the field, such as the required data characteristics for applying machine learning, which will need to be solved as machine learning matures.

Applications of machine learning in drug discovery and development.

What has happened in machine learning lately, and what does it mean for the future of medical image analysis? Machine learning has witnessed a tremendous amount of attention over the last few years. The current boom started around 2009 when so-called deep artificial neural networks began outperforming other established models on a number of important benchmarks. Deep neural networks are now the state-of-the-art machine learning models across a variety of areas, from image analysis to natural language processing, and widely deployed in academia and industry. These developments have a huge potential for medical imaging technology, medical data analysis, medical diagnostics and healthcare in general, slowly being realized. We provide a short overview of recent advances and some associated challenges in machine learning applied to medical image processing and image analysis. As this has become a very broad and fast expanding field we will not survey the entire landscape of applications, but put particular focus on deep learning in MRI. 
Our aim is threefold: (i) give a brief introduction to deep learning with pointers to core references; (ii) indicate how deep learning has been applied to the entire MRI processing chain, from acquisition to image retrieval, from segmentation to disease prediction; (iii) provide a starting point for people interested in experimenting and perhaps contributing to the field of machine learning for medical imaging by pointing out good educational resources, state-of-the-art open-source code, and interesting sources of data and problems related medical imaging.

An overview of deep learning in medical imaging focusing on MRI

What has happened in machine learning lately, and what does it mean for the future of medical image analysis? Machine learning has witnessed a tremendous amount of attention over the last few years. The current boom started around 2009 when so-called deep artificial neural networks began outperforming other established models on a number of important benchmarks. Deep neural networks are now the state-of-the-art machine learning models across a variety of areas, from image analysis to natural language processing, and widely deployed in academia and industry. These developments have a huge potential for medical imaging technology, medical data analysis, medical diagnostics and healthcare in general, slowly being realized. We provide a short overview of recent advances and some associated challenges in machine learning applied to medical image processing and image analysis. As this has become a very broad and fast expanding field we will not survey the entire landscape of applications, but put particular focus on deep learning in MRI. Our aim is threefold: (i) give a brief introduction to deep learning with pointers to core references; (ii) indicate how deep learning has been applied to the entire MRI processing chain, from acquisition to image retrieval, from segmentation to disease prediction; (iii) provide a starting point for people interested in experimenting and perhaps contributing to the field of deep learning for medical imaging by pointing out good educational resources, state-of-the-art open-source code, and interesting sources of data and problems related medical imaging.

Demonstrate a novel fast method for reconstruction of multi-dimensional MR fingerprinting (MRF) data using deep learning methods.A neural network (NN) is defined using the TensorFlow framework and trained on simulated MRF data computed with the extended phase graph formalism. The NN reconstruction accuracy for noiseless and noisy data is compared to conventional MRF template matching as a function of training data size and is quantified in simulated numerical brain phantom data and International Society for Magnetic Resonance in Medicine/National Institute of Standards and Technology phantom data measured on 1.5T and 3T scanners with an optimized MRF EPI and MRF fast imaging with steady state precession (FISP) sequences with spiral readout. The utility of the method is demonstrated in a healthy subject in vivo at 1.5T.Network training required 10 to 74 minutes; once trained, data reconstruction required approximately 10 ms for the MRF EPI and 76 ms for the MRF FISP sequence. Reconstruction of simulated, noiseless brain data using the NN resulted in a RMS error (RMSE) of 2.6 ms for T1 and 1.9 ms for T2 . The reconstruction error in the presence of noise was less than 10% for both T1 and T2 for SNR greater than 25 dB. Phantom measurements yielded good agreement (R2  = 0.99/0.99 for MRF EPI T1 /T2 and 0.94/0.98 for MRF FISP T1 /T2 ) between the T1 and T2 estimated by the NN and reference values from the International Society for Magnetic Resonance in Medicine/National Institute of Standards and Technology phantom.Reconstruction of MRF data with a NN is accurate, 300- to 5000-fold faster, and more robust to noise and dictionary undersampling than conventional MRF dictionary-matching.

MR fingerprinting Deep RecOnstruction NEtwork (DRONE)

Image reconstruction from undersampled k-space data has been playing an important role in fast magnetic resonance imaging (MRI). Recently, deep learning has demonstrated tremendous success in various fields and also shown potential in significantly accelerating MRI reconstruction with fewer measurements. This article provides an overview of deep-learning-based image reconstruction methods for MRI. Two types of deep-learningbased approaches are reviewed, those that are based on unrolled algorithms and those that are not, and the main structures of both are explained. Several signal processing issues for maximizing the potential of deep reconstruction in fast MRI are discussed, which may facilitate further development of the networks and performance analysis from a theoretical point of view.

Deep Magnetic Resonance Image Reconstruction: Inverse Problems Meet Neural Networks

PURPOSE To develop a fast magnetic resonance fingerprinting (MRF) method for quantitative chemical exchange saturation transfer (CEST) imaging. METHODS We implemented a CEST-MRF method to quantify the chemical exchange rate and volume fraction of the Nα -amine protons of L-arginine (L-Arg) phantoms and the amide and semi-solid exchangeable protons of in vivo rat brain tissue. L-Arg phantoms were made with different concentrations (25-100 mM) and pH (pH 4-6). The MRF acquisition schedule varied the saturation power randomly for 30 iterations (phantom: 0-6 μT; in vivo: 0-4 μT) with a total acquisition time of ≤2 min. The signal trajectories were pattern-matched to a large dictionary of signal trajectories simulated using the Bloch-McConnell equations for different combinations of exchange rate, exchangeable proton volume fraction, and water T1 and T2 relaxation times. RESULTS The chemical exchange rates of the Nα -amine protons of L-Arg were significantly (P < 0.0001) correlated with the rates measured with the quantitation of exchange using saturation power method. Similarly, the L-Arg concentrations determined using MRF were significantly (P < 0.0001) correlated with the known concentrations. The pH dependence of the exchange rate was well fit (R2  = 0.9186) by a base catalyzed exchange model. The amide proton exchange rate measured in rat brain cortex (34.8 ± 11.7 Hz) was in good agreement with that measured previously with the water exchange spectroscopy method (28.6 ± 7.4 Hz). The semi-solid proton volume fraction was elevated in white (12.2 ± 1.7%) compared to gray (8.1 ± 1.1%) matter brain regions in agreement with previous magnetization transfer studies. CONCLUSION CEST-MRF provides a method for fast, quantitative CEST imaging.

/pdf/rapid-and-quantitative-chemical-exchange-saturation-transfer-pl4np6g0pr.pdf

Rapid and quantitative chemical exchange saturation transfer (CEST) imaging with magnetic resonance fingerprinting (MRF).

Algorithm comparison for schedule optimization in MR fingerprinting.

Purpose To understand the influence of various acquisition parameters on the ability of CEST MR-Fingerprinting (MRF) to discriminate different chemical exchange parameters and to provide tools for optimal acquisition schedule design and parameter map reconstruction. Methods Numerical simulations were conducted using a parallel computing implementation of the Bloch-McConnell equations, examining the effect of TR, TE, flip-angle, water T 1 and T 2 , saturation-pulse duration, power, and frequency on the discrimination ability of CEST-MRF. A modified Euclidean distance matching metric was evaluated and compared to traditional dot product matching. L-Arginine phantoms of various concentrations and pH were scanned at 4.7T and the results compared to numerical findings. Results Simulations for dot product matching demonstrated that the optimal flip-angle and saturation times are 30 ∘ and 1100 ms, respectively. The optimal maximal saturation power was 3.4 μT for concentrated solutes with a slow exchange rate, and 5.2 μT for dilute solutes with medium-to-fast exchange rates. Using the Euclidean distance matching metric, much lower maximum saturation powers were required (1.6 and 2.4 μT, respectively), with a slightly longer saturation time (1500 ms) and 90 ∘ flip-angle. For both matching metrics, the discrimination ability increased with the repetition time. The experimental results were in agreement with simulations, demonstrating that more than a 50% reduction in scan-time can be achieved by Euclidean distance-based matching. Conclusions Optimization of the CEST-MRF acquisition schedule is critical for obtaining the best exchange parameter accuracy. The use of Euclidean distance-based matching of signal trajectories simultaneously improved the discrimination ability and reduced the scan time and maximal saturation power required.

CEST MR-Fingerprinting: practical considerations and insights for acquisition schedule design and improved reconstruction

PURPOSE Demonstrate an optimized multi-inversion echo-planar imaging technique to accelerate quantitative T1 mapping by judicious selection of inversion times for each slice. METHODS Slice ordering is optimized to maximize discrimination between tissues with different T1 values. The optimized slice orderings are tested in the International Society for Magnetic Resonance in Medicine/National Institute of Standards and Technology phantom and compared with an unoptimized 21-measurement acquisition. The utility of the method is demonstrated in a healthy subject in vivo at 3 T and validated with a gold-standard inversion-recovery sequence. The in vivo precision of our technique was tested by repeated scans of the same subject within a scan session and across scan sessions, occurring 28 days apart. RESULTS Phantom measurements yielded good agreement (R2  = 0.99) between the T1 estimates from the proposed optimized protocol, reference values from the National Institute of Standards and Technology phantom and gold-standard inversion-recovery values, as well as a negligible estimation bias that was slightly lower than that from the unoptimized 21-measurement protocol (0.74 versus 19 ms). The range of values for the scan-rescan coefficient of variation was 0.86 to 0.93 (within session) and 0.83 to 0.92 (across sessions) across all scan durations tested. CONCLUSIONS Optimized slice orderings allow faster quantitative T1 mapping. The optimized sequence yielded accurate and precise T1 maps. Magn Reson Med 79:2101-2112, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Ouri Cohen

Papers

MR fingerprinting Deep RecOnstruction NEtwork (DRONE)

Rapid and quantitative chemical exchange saturation transfer (CEST) imaging with magnetic resonance fingerprinting (MRF).

Algorithm comparison for schedule optimization in MR fingerprinting.

CEST MR-Fingerprinting: practical considerations and insights for acquisition schedule design and improved reconstruction

Optimized inversion‐time schedules for quantitative T1 measurements based on high‐resolution multi‐inversion EPI