An In Silico Exploration of Aberration-Corrected Transspinal Focused Ultrasound Using Zero Echo Time MRI

This study numerically assesses the feasibility of magnetic resonance imaging (MRI)-corrected transspinal focused ultrasound (FUS). Eleven ex vivo thoracic vertebrae were imaged with computed tomography (CT) and zero echo time (ZTE) magnetic resonance imaging. ZTE images were converted to pseudo-CT (pCT) with a linear fit from the literature (-2085×ZTE+2329). Transspinal focusing was simulated with a 256-element array (400 kHz, focal full-width-at-half-maximum ∼20 mm axial, 4 mm lateral) using five correction methods: no correction; pCT-corrected, homogeneous acoustic properties; pCT-corrected, heterogeneous properties; CT-corrected, homogeneous properties; and CT-corrected, heterogeneous properties (gold standard). Pressure fields generated using each correction method were evaluated against the gold standard. Compared to no correction, heterogeneous pCT-corrected focusing reduced mean spatial maximum shift (1.5 ± 2.0 mm vs. 3.5 ± 4.1 mm), increased Sørensen-Dice similarity of the 50% contour (0.75 ± 0.10 vs. 0.57 ± 0.18), and improved the change in driving efficiency relative to the gold standard (-12% ± 11% vs. -21% ± 25%). pCT-corrected with homogeneous properties performed similarly to the heterogeneous case. CT-corrected with homogeneous properties approached the gold standard, with a mean spatial shift of 0.8 ± 1.7 mm, Dice coefficient of 0.91 ± 0.06, and little change in driving efficiency (1% ± 3%). MR-corrected focusing using ZTE-derived acoustic properties outperformed uncorrected focusing but did not match gold-standard CT-corrected focusing. Similarities in the performance of heterogeneous versus homogeneous pCT correction, along with the strong performance of homogeneous CT-corrected focusing, suggest that accurate morphological representation of the spine is key to accurate MR-corrected transspinal focused ultrasound.