Purpose

MRI has become the imaging modality of choice to monitor thermal therapies such as focused ultrasound (FUS), microwave, and laser induced thermal therapies (LITT) because it combines good contrast anatomical imaging with the ability to monitor temperature change using the proton resonance frequency shift (PRFS) method. Acquiring high spatio-temporal resolution PRFS-maps with real-time image reconstruction remain challenging. Non-Cartesian acquisitions such as radial and spiral imaging are fast and efficient, but reconstruction including gridding and off-resonance correction can be time consuming(1). Iterative and thermal model-based methods have also been described, but iterative methods are computationally heavy by nature, and model based methods require knowledge of tissue and thermal properties(2–5). Parallel imaging can also be employed(6), but for some treatment systems only a single receive coil is available. In this work we present a dynamic anti-aliasing image reconstruction (DAIR) method that is appropriate for subsampled k-space data for localized thermal therapies. A fully sampled reference image acquired before the start of the heating is used to remove the aliasing artifact from dynamic subsampled images acquired during the heating. DAIR does not make use of multiple coils, and image reconstruction is very fast.

Methods

The DAIR algorithm is described in Figure 1. The method is very fast since the aliasing artifact can be computed once a fully sampled reference image is acquired, and the dynamic image reconstruction is then a simple subtraction in image space. All imaging was performed on a 3T scanner (PrismaFit, Siemens, Germany) with a 3D gradient recalled echo segmented echo planar imaging pulse sequence. The signal was detected with an in-house built single-channel RF loop coil. Three different CAIPI-type sampling patterns (7) with subsampling factor of R = 2, 3, and 4 were used. For R = 2 and 3 R_y = 1 and Δ = 1, and for R = 4 R_y = 2 and Δ = 1, as defined in (7). All data was zero-filled interpolated to 1-mm isotropic voxel spacing. FUS sonications were performed with a 1-MHz 256-element phased-array transducer (Imasonic/IGT, France) in a tissue mimicking gelatin phantom. All MR and FUS parameters are listed in Table 1. Three sonications were imaged with each subsampling pattern. For comparison identical sonications were monitored with fully sampled “truth” data acquisitions (at lower temporal resolution).

Results

Figure 2 show a comparison of phase maps for the three subsampling schemes, comparing the aliased phase map (i.e. from the 3D Fourier transform of the subsampled data), to phase map resulting from a complex subtraction between two subsampled frames, phase map of the DAIR reconstructions, and lastly the DAIR phase map after complex baseline subtraction. Figures 3 and 4 show temperature rise versus time and spatial temperature maps at maximum heating, compared to fully sampled “truth”.

Discussion

Phase maps in Figure 2 show that DAIR reconstructions experience less artifacts and a more coherent focal spot than is the result of complex subtraction of two aliased images. Temperature curves and maps in Figures 3 and 4 further show good agreement with fully sampled “truth”. For in-vivo imaging multiple baseline images can be acquired and used to decrease sensitivity to motion. In this work only single RF-coil data was used, but DAIR extends naturally to multi-coil data. The fact that the dynamic reconstruction is a simple subtraction means it lends it self well for real-time applications. The limiting factor in DAIR is the spacing of the delta functions in the point-spread function (which depends on R), Figure 1d, with respect to the size of the heating pattern. If the heating pattern is larger than FOV/R the aliasing subtraction might not be clean. Because of this limitation DAIR might be most useful for LITT and microwave heating, and less useful for FUS where the fully insonified FOV is ideally monitored for off-focal heating.

Conclusions

A novel image reconstruction algorithm for undersampled MRI monitoring of localized heating has been presented. Accurate PRFS temperature maps monitoring FUS heatings in a phantom was achieved. Future work will investigate SNR behavior and constraints, combining DAIR with parallel imaging methods, and investigate reconstruction of non-Cartesian data.

Acknowledgements

This work was supported by a University of Utah seed grant, and NIH grants R01s EB013433, and CA172787.

References

1. Marx M, Butts Pauly K. Improved MRI thermometry with multiple-echo spirals. Magn. Reson. Med. 2016;76:747–756. 2. Todd N, Adluru G, Payne A, DiBella EVR, Parker D. Temporally constrained reconstruction applied to MRI temperature data. Magn. Reson. Med. 2009;62:406–19. 3. Gaur P, Grissom WA. Accelerated MRI thermometry by direct estimation of temperature from undersampled k-space data. Magn. Reson. Med. 2015;73:1914–25. 4. Todd N, Payne A, Parker DL. Model predictive filtering for improved temporal resolution in MRI temperature imaging. Magn. Reson. Med. 2010;63:1269–1279. 5. Denis de Senneville B, Roujol S, Hey S, Moonen C, Ries M, Roujol E, Hey S, Moonen C, Ries M. Extended Kalman filtering for continuous volumetric MR-temperature imaging. IEEE Trans. Med. Imaging 2013;32:711–8. 6. Mei CS, Panych LP, Yuan J, McDannold NJ, Treat LH, Jing Y, Madore B. Combining two-dimensional spatially selective RF excitation, parallel imaging, and UNFOLD for accelerated MR thermometry imaging. Magn. Reson. Med. 2011;66:112–22. 7. Breuer FA, Blaimer M, Mueller MF, Seiberlich N, Heidemann RM, Griswold MA, Jakob PM. Controlled aliasing in volumetric parallel imaging (2D CAIPIRINHA). Magn. Reson. Med. 2006;55:549–556.