Prospective Deployment of Deep Learning in MRI: A Framework for Important Considerations, Challenges, and Recommendations for Best Practices

Abstract

Artificial intelligence algorithms based on principles of deep learning (DL) have made a large impact on the acquisition, reconstruction, and interpretation of MRI data. Despite the large number of retrospective studies using DL, there are fewer applications of DL in the clinic on a routine basis. To address this large translational gap, we review the recent publications to determine three major use cases that DL can have in MRI, namely, that of model-free image synthesis, model-based image reconstruction, and image or pixel-level classification. For each of these three areas, we provide a framework for important considerations that consist of appropriate model training paradigms, evaluation of model robustness, downstream clinical utility, opportunities for future advances, as well recommendations for best current practices. We draw inspiration for this framework from advances in computer vision in natural imaging as well as additional healthcare fields. We further emphasize the need for reproducibility of research studies through the sharing of datasets and software. Level of Evidence: 5 Technical Efficacy Stage: 2

Low-count whole-body PET with deep learning in a multicenter and externally validated study

“…Maintaining comparable SUV quantitation between the low-count-enhanced and standard PET images is important to assess tumor avidity and response to therapy. Coupled with the high concordance in the depiction of lesions and the SUV quantification between the standard and low-count-enhanced scans, the proposed method maintained pixelwise accuracy and imagewise quality, two criteria important for prospective image-acquisition studies 34 . Overall, this shows that the proposed method can be used for accurate staging and prognostication.…”

Diagnostic Accuracy of Quantitative Multicontrast 5-Minute Knee MRI Using Prospective Artificial Intelligence Image Quality Enhancement

“…The novelty of this work compared with previous studies of rapid imaging protocols is the simultaneous acquisition of anatomic and quantitative imaging along with the incorporation of quantitative T2 mapping into the MRI interpretation. Additionally, our study presents a prospective implementation of deep learning to enhance image quality from lower-resolution acquired data as opposed to a simply retrospective undersampling of high-resolution data, as is more commonly performed [28]. To our knowledge, this is the first study to use both simultaneous T2 map generation and analysis for evaluating knee internal derangement and prospective deep learning for 3D knee MRI.…”

Identification of Tumor-Specific MRI Biomarkers Using Machine Learning (ML)

“…Despite the large number of retrospective studies ( Figure 2 ), there are fewer applications of deep learning in the clinic on a routine basis [ 127 ]. The three major use cases that deep learning can have in MRI diagnostics: (1) model-free image synthesis, (2) model-based image reconstruction, and (3) image or pixel-level classification [ 127 ].…”

“…Despite the large number of retrospective studies ( Figure 2 ), there are fewer applications of deep learning in the clinic on a routine basis [ 127 ]. The three major use cases that deep learning can have in MRI diagnostics: (1) model-free image synthesis, (2) model-based image reconstruction, and (3) image or pixel-level classification [ 127 ]. Hence, deep learning has the potential to improve every step of the MRI diagnostic workflow and to provide value for every user, from the technologists performing the scan, the physicians ordering the imaging, the radiologists providing the interpretation, and most importantly, the patients who are receiving health care.…”