Abbreviated yet enhanced neuro MRI protocol - Project Summary Given the high cost of MRI, there is a need to maximize the value of these exams in order to make them as worthwhile to society as possible. Value can be increased in a number of ways, such as improving accessibility or increasing the impact on patient management. The present project takes aim at one of the most common types of MRI exams, brain exams. The overall goal is to develop a time-abbreviated exam whereby greater amounts of information are obtained in a smaller amount of time. While faster exams can help increase acces- sibility on a given fleet of scanners, an increased information content can potentially have a positive impact on patient management. Brain exams represent up to 40% of all MRI exams, and typical time slots are 20- to 30-min long. Most brain MRI exams involve a gadolinium (Gd) contrast injection, and much of the time is spent acquiring data with several different MR image contrasts before and/or after the injection. The present project is ambitious in scope; it aims to reduce exam duration by roughly three-fold, while capturing quantitative and qualitative MRI contrasts, as well as the Gd-based enhancement. The duration of the proposed exam is primarily determined by the dynamics of the contrast agent, rather than by the demands of MRI. It takes roughly 5 to 10 minutes for the Gd agent to work its way through normal and abnormal anatomy and, therefore, our proposed exam dura- tion is 8 min. As for contrast types, current clinical exams typically capture only qualitative (i.e., no quantitative) contrasts, and only pre- and post-enhancement phases (i.e., the Gd-based enhancement is not resolved in time). Quantitative contrasts and time-resolved enhancement curves, captured in our proposed approach, will increase the diagnostic information content of individual exams. In contrast with abbreviated protocols that re- duce scan time by cutting sequences and/or increasing slice thickness, the proposed exam adds valuable in- formation over the full-length exam rather than reducing it. We developed a new 3D multi-pathway multi-echo (MPME) MRI sequence and associated machine- learning contrast translation to convert information-rich MPME signals into a variety of contrasts. Signals will be acquired in a continuous manner over the 8-minute exam, generating a series of 3D whole-head images. Re- constructed MRI contrasts include: T1-weighted, T2-weighted, proton density weighted, susceptibility- weighted, T1 maps, T2 maps, T2 FLAIR and MPRAGE images. A total of 200 patients will receive our time- abbreviated multi-contrast exam, both for neural network training/validation and for testing. Qualitative and quantitative quality metrics, as well as diagnostic outcome, will be used for validation. An additional twenty-five healthy volunteers will also be recruited, for a repeatability study. In summary, a fast, comprehensive brain MRI technique is being developed based on our MPME sequence and machine-learning contrast translation. This project fits into a larger body of work aimed at increasing the overall value of MRI as an imaging modality.
