MRI is the modality of choice for evaluating the integrity of the silicone breast implants. Presently, the most widely used method for obtaining silicone-specific images is based on a combination of short-tau inversion recovery (STIR) for suppressing the fat signal and chemical shift selective saturation (CHESS) for suppressing the water signal. In practice, the combined STIR-CHESS method often suffers from the low SNR and long acquisition time of STIR, and from the sensitivity of CHESS to magnetic field inhomogeneity. As an alternative, silicone-specific images can be obtained by using the Dixon methods which are insensitive to magnetic field inhomogeneity. The first attempt in using the Dixon methods for silicone-specific imaging is based on a three-point Dixon technique that requires two in-phase and one 180^o out-of-phase acquisitions and on assuming that the frequency separation between water (W) and fat (F) is twice that between F and silicone (S)¹. This assumption, which is only an approximation to the actual frequency separations, is later removed in a combined STIR-Dixon approach in which STIR is used to remove the fat signal and a three-point Dixon technique is used to separate the remaining water and silicone signals^{2, 3}. In this work, we propose a new unipolar flexible fast spin echo triple echo Dixon (FTED) technique for silicone-specific imaging. The underlying idea is to treat the W and F signals as a single component by acquiring images only when W and F are in-phase, and to use Dixon processing to separate the remaining S signal. A similar idea was demonstrated using a fast 3D gradient echo single-point Dixon technique⁴. Here, we demonstrate the feasibility of using a unipolar flexible FTED sequence with its advantages.

The proposed pulse sequence is shown in Fig. 1. Similar to the original FTED⁵, three raw images are acquired in a single acquisition without interleave. However, the following differences are incorporated. First, while the second echo (gxw) is with all signals in phase, the first echo (gxw-) and the third echo (gxw+) are acquired when W and F signals are in-phase while S signals have a relative phase of θ. A two-point Dixon processing algorithm with flexible echo times⁶ is then applied to generate a silicone-only image and a W+F image. Secondly, the unipolar readouts are used for all three echoes to avoid spatial misregistration along the frequency encode direction due to chemical shifts or other off-resonance that would be present if the more efficient bipolar readouts are used.

We implemented the sequence on a GE 3.0 Tesla whole-body scanner (GE Healthcare, Waukesha, WI). The image reconstruction algorithms were implemented using GE Healthcare’s Orchestra SDK software (Waukesha, WI) that allows generation of DICOM images directly from the acquired raw-data without any user inputs. The sequence was used to image the silicone breast implant of a patient after the IRB approval and informed consent. An eight-channel phased array breast coil was used and the scan parameters were: TR/TE = 5000/66ms, echo train length (ETL) = 12, acquisition matrix = 256x192, FOV = 24x24cm, RBW = ±200kHz, slice-thickness/slice-gap = 4/0mm, signal average (NEX) =2, no phase wrap (NPW) turned on, and no parallel imaging acceleration. A total of 28 slices were acquired in 5:31 minutes.

Fig. 2 a) and b) show the raw magnitude images from the 1st echo and the 2nd echo of a representative slice. For the image of the 2nd echo, all three species of W, F, and S are in-phase. For the image of the 1st echo (as well as the third echo, not shown), W and F are again in-phase, but S has a relative phase of approximately 150^o. Fig. 2 c) and d) show the corresponding W+F image and S image, respectively. These images show clean and uniform silicone separation across the field of view without any misregistration artefacts. In comparison to many techniques that may be used for silicone-specific imaging, the proposed unipolar flexible FTED technique has several important advantages. First, it does not use STIR or CHESS, and therefore maintains high SNR, scan efficiency, and insensitivity to field inhomogeneity. Second, the images are based on a single FSE acquisition with unipolar readouts. Therefore, the silicone images are T2-weighted and do not suffer from misregistration artefacts. Motion artefacts are usually minimal when compared to an interleaved acquisition. Finally, the technique can be extended to 3D or somewhat less-efficient interleaved FSE two-point Dixon technique⁷ when the two images are acquired both at water and fat in-phase.