Incorporating time-resolved three-dimensional phase contrast (4D flow) MRI in clinical workflow: initial experiences at a large tertiary care medical center

Abstract

Background: Time-resolved three-dimensional phase contrast (4D flow) MRI allows for visualization of three-dimensional cardiovascular anatomy and pulsatile flow with full volumetric coverage in a single, easy to prescribe 3D acquisition. The technique provides comprehensive flow visualization and permits retrospective flow quantification at any userdefined region of interest. [1] To our knowledge, no center has incorporated 4D flow MRI as a part of standard clinical cardiovascular MRI (CMR). The goals of this study include: 1) reporting on the incorporation of 4D flow MRI acquisition and processing as part of clinical CMR workflow and 2) better understanding the clinical impact of 3D flow visualization and retrospective flow quantification derived from 4D flow MRI in CMR. Methods: Patients referred to Northwestern Memorial Hospital for CMR with relevant clinical indications as judged by an attending radiologist were selected to have 4D flow MRI included in accordance with an IRBapproved protocol. Images were processed using in-house software for noise reduction, anti-aliasing, and eddy current correction. Flow visualization and quantification were performed using EnSight (CEI, Apex, NC). Processing time was recorded. Resulting 4D flow pathline cine images and flow quantification results were converted to DICOM format and transferred to the local PACS server to be reviewed as part of the patient's clinical images. Clinician-requested quantitative data was compared between 4D flow and two-dimensional phase contrast (2D PC) MRI techniques. Each case was retrospectively reviewed by an attending radiologist who assigned a qualitative measure of the 4D flow analysis' impact on the case 1) excluding 2D PC and 2) including 4D flow and 2D PC together (criteria listed in Table 1). Results: Sixteen patients had clinical 4D flow MRI over 10 weeks. Clinical indications are reviewed in Table 1. The average 4D flow impact factor (scale 1-4) excluding and including 2D PC was 2.93 +/- 0.77 and 2.80 +/- 0.75, respectively. The average percent difference in quantitative data was 35 +/- 28%. One patient could not be evaluated secondary to aliasing. Average time for 4D flow post-processing was 88.5 +/- 22.5 min. Three clinical cases are shown in Figure 1. (Table presented) Conclusions: Including 4D flow MRI as part of clinical CMR workflow is feasible and has the potential to impact clinical assessment in multiple cardiovascular pathologies. The ability to evaluate flow throughout the acquired 3D volume retrospectively may reduce dependence on timeintensive 2D PC acquisitions while yielding accurate and efficient flow quantification.