Quantitative characterization of atherosclerotic plaque based on T2 mapping plays a vital role in the diagnosis of cerebrovascular events. T2 mapping has been shown to reliably characterize plaque components [1] and is a significant improvement over qualitative multi contrast imaging. The double-inversion recovery (DIR) fast spin-echo (FSE) sequence can provide excellent contrast between vessel wall and blood and is suitable for assessing plaque build up. The DIR module uses a non-selective (NS) inversion pulse, to suppress the signal from blood within the volume excited by the RF coil, followed by a slice-selective inversion pulse which re-inverts the signal within the slice of interest. There is only a single point in time during the recovery of the inverted magnetization at which the blood signal can be completely nulled. However, due to the NS inversion pulse only one slice can be acquired after each DIR preparatory module preventing the interleaving of multiple slices in a single TR.

Techniques to improve slice coverage by interleaving multiple slices within the same DIR block or across R-R intervals come at the cost of imperfect blood nulling or lower SNR efficiency [2,3,4]. In this work we present a variant of DIR-FSE that uses multi-band (MB) excitation to generate data from multiple slices simultaneously. As data are acquired from all the slices simultaneously, the proposed sequence ensures that all the slices are affected by the same inversion time. The use of MB excitation results in improved SNR efficiency compared to standard 2D imaging. We use a radial version of the MB-DIR-FSE sequence that has the added advantage of yielding both high-resolution T2 maps and black blood TE images from a single acquisition.

MB pulses that can excite four slices simultaneously were designed by summing single band pulses with appropriate frequency offsets and phase modulations and used as the excite pulse in the FSE imaging sequence. The phase profiles of the underlying bands were modulated according to the entries of the rows of the matrix in Fig. 1b [7]. Our current implementation excites four 3 mm slices within a 16.5 mm volume while we refocus a 19.8 mm slab. The selective inversion in the DIR module is also made slab selective and is roughly of the same size as the refocussing slab. An illustration of MB excitation/encoding, slab selective refocussing and uniform blood nulling is given in Fig. 1. Data were acquired on a healthy volunteer on 3T Siemens scanner using a pulse gated MB-DIR-RADFSE sequence with ETL = 16, echo spacing = 11 ms, FOV = 16 cm, readout bandwidth = 300 Hz/pixel, TR = 2 R-R and the MB pulses are interleaved across TR’s.

The acquired data can be decoded into the k-space data of the underlying slices by applying the inverse of the encoding matrix in Fig. 1(b). The decoded k-space data for each slice is then separated into k-space datasets corresponding to each of the 16 TE’s. TE images are reconstructed from these k-space datasets using an iterative T2 mapping technique [5].

Figure 2(a) shows representative black-blood carotid images acquired using the proposed technique. All the four slices show excellent blood suppression as the data are acquired at the same inversion time. Figure 2(b) shows colorized T2 maps.The proposed technique can allow the acquisition of multiple slices at the exact null point of blood and at a better SNR efficiency than 2D DIR-FSE due to MB excitation making it an attractive alternative to full 3D scanning. As the signal in the slice dimension is resolved via RF phase encoding instead of gradient encoding as is the case in 3D MRI, MB encoding does not suffer from truncation and ringing artifacts that affect 3D scans especially when very few slices are encoded in the slab. Another advantage of MB excitation is that it gives us flexibility in positioning the slices within the slab. We also use potentially thinner slabs for refocussing and inversion leading to lower flow related artifacts that can affect 3D scans. The need to encode fewer slices along with the radial nature of k-space sampling also allows better motion robustness than full 3D scanning. The better SNR efficiency of the proposed sequence can also be used to increase in-plane resolution or reduce overall scan time. Our current implementation of the sequence, MB-DIR-RADFSE, can generate black-blood images from four slices, up to 16 TE images/slice and the associated T2 maps at a better SNR efficiency than standard 2D DIR-FSE and with excellent flow suppression.