Objective. To evaluate the feasibility of transspine focused ultrasound using simulation-based phase corrections from a CTderived ray acoustics model. 
Approach. Bilateral transspine focusing was performed in ex vivo human vertebrae with a spine-specific ultrasound array. Ray acoustics-derived phase correction was compared to geometric focusing and a hydrophonecorrected gold standard. Planar hydrophone scans were recorded in the spinal canal and three metrics were calculated: target pressure, coronal and sagittal focal shift, and coronal and sagittal Sørensen-Dice similarity to the free field. Post hoc analysis was performed in silico to assess the impact of windows between vertebrae on focal shift.
Main results. Hydrophone correction reduced mean sagittal plane shift from 1.74 ± 0.82 mm to 1.40 ± 0.82 mm and mean coronal plane shift from 1.07 ± 0.63 mm to 0.54 ± 0.49 mm. Ray acoustics correction reduced mean sagittal plane and coronal plane shift to 1.63 ± 0.83 mm and 0.83 ± 0.60 mm, respectively. Hydrophone correction increased mean sagittal similarity from 0.48 ± 0.22 to 0.68 ± 0.19 and mean coronal similarity from 0.48 ± 0.23 to 0.70 ± 0.19. Ray acoustics correction increased mean sagittal and coronal similarity to 0.53 ± 0.25 and 0.55 ± 0.26, respectively. Target pressure was relatively unchanged across beamforming methods. In silico analysis found that, for some targets, unoccluded paths may have increased focal shift. 
Significance. Gold standard phase correction significantly reduced coronal shift and significantly increased sagittal and coronal Sørensen-Dice similarity (p < 0.05). Ray acoustics-derived phase correction reduced sagittal and coronal shift and increased sagittal and coronal similarity but did not achieve statistical significance. Across beamforming methods, mean focal shift was comparable to MRI resolution, suggesting that transspine focusing is possible with minimal correction in favorable targets. Future work will explore the mitigation of acoustic windows with anti-focus control points.