Towards distinguishing intra-canal and paraspinal cavitation activity during focused ultrasound exposures in the spine

<i>Objective.</i>Although less established than transcranial focused ultrasound (FUS), transvertebral FUS is being developed to treat spinal cord pathologies. Transvertebral sonication of the spinal cord for microbubble-mediated drug delivery generates cavitation at the target in the spinal canal, and outside the spinal canal due to reflection off the posterior surface of the spinal column. In these two regions, circulating microbubbles are excited by local foci to generate acoustic emissions that are used to monitor FUS treatments. When trying to localize acoustic emissions generated from cavitation in the spinal cord, prefocal cavitation emissions emanating from paraspinal regions can dwarf signals originating in the canal and compromising monitoring capabilities. This paper evaluates alternative reconstruction algorithms to delay-sum-and-integrate (DAS)<i>in-silico</i>and<i>ex-vivo</i>to more reliably map intra-spinal canal sources in the face of interference.<i>Approach.</i>A proof-of-concept 400/800 kHz (transmit/receive) spine-specific array prototype was used to generate intracanal cavitation through intact human vertebrae and passively monitor the corresponding acoustic emissions. Delay-multiply-sum-and-integrate (DMAS) beamforming was compared to DAS in two different implementations, full array (DMAS) and half-array multiplicative compounding (DMASMu), in the modeled cavitation scenarios where paraspinal cavitation is present.<i>Main results.</i>Both DMAS and DMASMu improved image quality by reducing peak sidelobes and increasing image signal-to-noise ratio. Aberration corrections further improved image quality metrics and, when applied selectively to voxels co-registered to the canal, assisted localization when prefocal sources were present<i>in-silico</i>. When localizing canal sources in the presence of paraspinal cavitation, a switch to DMAS/DMASMu offered a more consistent localization rate<i>in-silico</i>and<i>ex-vivo</i>, though<i>ex-vivo</i>phase and amplitude corrections failed to replicate<i>in-silico</i>findings.<i>Significance.</i>DMAS or DMASMu reconstruction with multiple dynamic ranges and sub-image integration timings can provide more reliable mapping of cavitation in the canal in the presence of interference from paraspinal cavitation.