While multi-spectral imaging (MSI) can greatly minimize distortions caused by metal implants , it is of primary importance to image the neighborhood around an implant or device; spatially displaced regions from the device are more efficiently imaged with standard acquisition methods. In this work, a method for reduced field of view multi-spectral imaging is demonstrated in a phantom and in a subject with bilateral hip replacements. This method can, in turn, be used to achieve effective “accelerated” full field of view images as has been proposed elsewhere¹.

A two dimensional MSI acquisition² was performed in cases of a water phantom with an included Co/Cr orthopedic device (GE Discovery MR750 3T) and a patient with bilateral hip replacements (GE Discovery MR450 1.5T). In the case of the phantom, imaging parameters included: 256 matrix, TE 7.1 ms, TR 5000 ms, 16 frequency bins, 1 kHz bin width, 600 Hz bin separation, and a 32-channel array. In the case of the patient, the imaging parameters were identical matching parameters, except 24 frequency bins were employed.

Data were retrospectively decimated to obtain reduction factors of 2, 4, and 8 in the phase encoding direction to simulate half, quarter, and one eighth field of views. For calibration, the central 32 x 32 matrix of k-space was extracted from each coil and each bin before decimation. Alternatively, a separate low-resolution calibration acquisition can be used in place of this method³. A standard SENSE algorithm⁴ was applied in which each coil sensitivity signal was replaced by a coil-bin combination. Thus, in the phantom case with a 32-channel coil and 16 bins, “sensitivity encoding” was effectively achieved with 512 elements.

Figures 1 illustrates phantom images including the ideal sum of squares combined image (A), calibration combined image (B), the aliased and half field of view images (C, D), aliased and quarter field of view images (E, F), and aliased and eighth field of view images (G, H). Figure 2 includes the same ideal, calibration, aliased and reduced field of view images for the patient. Figure 3 illustrates final half field of view images for the human demonstration before and after processing. It is clear that images were reconstructed with an effective reduction factor of 4 with minimal residual aliasing artifact, even outside of the reduced field of view. Interestingly, the reduced field of view neighborhood of the implant can be effectively reconstructed without artifact with even higher reduction factors because of the smaller spatial extent of the “sensitivity profiles” from the frequency bins in that area. This is particularly evident in the case of the 8x field of view reduction with the phantom. The small spatial extent of those bins greatly reduces the available signal to alias with the reduced field of view arising from the sub-sampled k-space.In cases where multi-spectral imaging is employed, it is often desirable to image a reduced field of view around a region where an implant is located. The extent of anatomy in the imaging plane, however, effectively limits the field of view that can be achieved without aliasing artifact. However, frequency offset bins, acquired in a reference scan, can offer a level of spatial encoding along with that of individual receiver array elements. This further spatial encoding is responsible for the retained signal-to-noise ratio in areas around the device with higher acceleration factors, further enabling greater reduction in the reconstructed field of view. These reduced fields of view could benefit in creating high-resolution images of pathology that occur closer to the implant boundary.This preliminary work illustrates the use of offset frequency bins as effective sensitivity profiles to utilize in reduced field of view imaging around orthopedic implants. Specifically, high reduction factors in the field of view are appropriate in regions around implants where off-resonance bins are spatially compact.