The effects of image homogenisation on simulated transcranial ultrasound propagation

Abstract

Transcranial transmission of ultrasound is increasingly used in a variety of clinical and research applications, including high intensity ablation, opening the blood brain barrier, and neural stimulation. Numerical simulations of ultrasound propagation in the head are used to enable effective transcranial focusing and predict intracranial fields. Such simulations require maps of the acoustic properties of the head, which can be derived from clinical CT images. However, the spatial resolution of these images is typically coarser than the scale of heterogeneities within the skull bone, which are known to exert a major influence on ultrasound propagation. In the present work, the impact of image related homogenisation on transcranial transmission from a single element transducer is examined using a dataset of co-registered clinical resolution CT and micro-CT images of skull sections. Reference acoustic property maps are derived from micro-CT images of cortical bone tissue. The influence of imaging resolution is examined by progressively downsampling the segmented acoustic property maps, and through comparison with maps derived from co-registered clinical CT images. The influence of different methods of segmenting the bone volume from the clinical CT images, and for resampling the clinical and micro-CT data are also examined. Image related homogenisation is demonstrated to have a substantial effect on the transcranial transmission of ultrasound, resulting in underestimations of simulated transmission loss and time-of flight. Effects on time-of flight are due to the loss of the internal scattering microstructure of the skull, while changes in transmitted ultrasound amplitude are due to both loss of microstructure and other smoothing effects. Inflating the simulated attenuation coefficient of the skull layer reduces the error in transmitted pressure amplitude to around 40%, however this is unable to correct fully for errors in time of flight and the pressure distribution of the transmitted field.

Transcranial ultrasound simulations: A review

“…This is in agreement with the Robertson et al. study, 67 who showed that the homogenization of acoustic property maps leads to an overestimation in the amplitude of ultrasound and an underestimation of time‐of‐flight. However, smarter homogeneous models can be used to improve computation time without losing too much accuracy.…”

Systematic examination of low-intensity ultrasound parameters on human motor cortex excitability and behavior

“…To model acoustic propagation of applied transcranial ultrasound, each participant's coronal MRI slice corresponding to the neuronavigation-captured transducer position was manually segmented into scalp, skull, and brain tissue planes (Hynynen and Sun, 1999;Rosnitskiy et al, 2019) (Appendix 1). The segmented tissue layers were treated as homogenous tissue masks with material properties such as attenuation coefficient, sound velocity, and density derived from the literature (Mueller et al, 2017;Robertson et al, 2018Robertson et al, , 2017. An acoustic simulation toolbox, k-Wave (Treeby and Cox, 2010), was used to generate simulations of the acoustic focus for each participant based on individualized transducer positions acquired via neuronavigation ( Figure 2B-C) and the resulting pressure field was mapped back onto the original MRI images ( Figure 2-Figure supplement 1).…”

Analysis of Multifrequency and Phase Keying Strategies for Focusing Ultrasound to the Human Vertebral Canal

“…The assumption of bone homogeneity may be insufficient to capture the complexity of ultrasound propagation through vertebral bone, which has been shown to have varying density and complex trabecular structures [44]. Recently, image homogenization has been shown to contribute to the underestimation of transmitted amplitude and time-of-flight in transcranial simulations [45], and is likely to under-predict focal distortions. Additionally, the acoustic properties used in these simulations were based on skull bone, and these may not be transferrable to vertebral bone.…”