MR imaging of bilateral brachial plexus has been challenging, primarily due to the following reasons:

1) Fat suppression failures. The shape of the anatomy in head and neck is complex, and uniform B₀ field cannot be achieved by shimming gradients.

2) For patient comforts, arms-down scanning is preferred. However, this will cause arms-wrapping artifacts in coronal imaging, which is also the preferred scan orientation for this application.

3) 3D imaging with high spatial resolution for the large FOV needed in this application is time consuming.

In this work, these aforementioned challenges were addressed by combining and utilizing novel MR imaging techniques, and a fast and robust protocol for bilateral brachial plexus MR imaging is proposed, and demonstrated at both 1.5T and 3.0T.

1) To address the fat suppression challenge, a Dixon-based water-fat separation method was used [1]. Because Dixon-based techniques are less sensitive to B₀ inhomegeneity, uniform fat suppression can be achieved.

2) To address arms wrapping, cosine modulated very selective saturation pulses [2] that employed outer volume suppression (Focus) on either side of the phase field-of-view [3] is used. Since the arms are not excited in the first place, there is no signal aliasing from the arms.

3) MR images are typically compressible, a property that is well-suited for Compressed Sensing (CS) [4,5]. Therefore, to address the long scan time associated with 3D acquisition, CS is used to reduce total scan time.

Two healthy volunteers were scanned at 1.5T and 3.0T to demonstrate the feasibility of the combined technique (Optima 450w and Discovery MR750w, GE Healthcare, Waukesha, WI, U.S.A.). The pulse sequence is based on a 3D fast spin echo (FSE) sequence (Cube). For 1.5T scanning, FOV = 36 × 32.4 cm², TR = 2250 ms, 100 slices with thickness of 1.4 mm. Acquisition matrix was 256 × 230, ETL = 100, BW = ±62.5 kHz. Imaging parameters at 3.0T were similar to 1.5T, expect BW = ±200 kHz.

Figure 1 and Figure 2 shows a single slice from different scans on the 1.5T and 3.0T, respectively. As shown in the figure, conventional spectrally-selective fat saturation (Chem FatSat) failed in the head and neck region due to B₀ inhomogeneity. With the Dixon-based FTED technique, fat suppression was significantly improved, but signal aliasing from arms are present in the image and obscuring the visualization of the nerves. Then Focus was able to remove the aliased signal. Finally, CS was used to reduce scan time by 1 min (~27%), and without any noticeable difference. Limited Maximum Intensity Projection (MIP) images on the reformatted images are shown in Figure 3, with well visualization of the entire brachial plexus nerves.The combination of these methods allows high resolution volumetric imaging of the brachial plexus with robust fat suppression in a short time, allowing for reformats in oblique planes of the nerves. Current state of the art 2D protocols with similar resolution may take up to 7~8 minutes. A diffusion preparation pulse could be further used to suppress blood vessels, and improve conspicuity of nerves as well as compressibility of the images allowing for higher CS acceleration.In this work, a fast and robust protocol for bilateral brachial plexus MR imaging is proposed, and preliminary results from both 1.5T and 3.0T have been demonstrated.