Transcranial ultrasound tomography for brain imaging: Ex vivo results and potential for stroke imaging

Abstract

Background

Stroke is a leading cause of death and disability in America and around the world. Due to the differences in underlying cause and clinical treatment between ischemic and hemorrhagic stroke, it is critical to perform imaging before administration of treatment. The current clinical standard is to image patients with computed tomography or magnetic resonance imaging, which creates a major delay in stroke treatment because paramedics have to transport patients to a major medical center before imaging can even begin. Therefore, there is an unmet need to provide imaging at the point of injury for stroke patients, so that therapeutic intervention can occur faster.

Purpose

This work investigates the feasibility of using ultrasound tomography (UST) with full waveform inversion (FWI) image reconstruction as a point-of-injury tool to help provide faster stroke triage, using in silico, in vitro, and ex vivo imaging phantoms

Methods

In silico image data were simulated for three datasets, which each include skull, brain, and a unique hemorrhage. These data were reconstructed with two different frequency ranges; one (100–700 kHz) to simulate an ideal hardware setup, and the other (300–700 kHz) to represent the limitations of our current image acquisition system. Next, a replica skull was filled with a brain-mimicking gelatin phantom and three regions of blood to simulate hemorrhage prior to imaging and reconstruction to visualize the blood contrast in a realistic environment. Then, a preserved macaque brain was imaged both outside of and within the replica skull to demonstrate our ability to visualize anatomical landmarks and the effects generated by the skull. Finally, an intact human cadaveric brain was imaged to demonstrate our ability to resolve important anatomical landmarks in a relevant model.

Results

In silico experiments show that UST is capable of imaging anatomical landmarks and hemorrhage pathologies through the skull, despite artifacts when starting FWI reconstruction at 300 kHz. The in vitro hemorrhage phantom also demonstrated that hemorrhages as small as 0.7 cm in diameter can be visualized through the replica skull using UST. Multiple features, such as the interhemispheric fissure, sylvian fissures, ventricles, and the brain stem could be visualized in the macaque brain when imaged in a water bath, and these features remained visible even when the brain was placed within the replica skull despite additional artifacts. Finally, the human brain was visualized with UST, showing high-resolution images with significant anatomical detail. Ultimately, the images presented in this work demonstrate the level of detail which can ultimately be achieved using this technique in the absence of the skull, which will guide future development.

Conclusion

This work shows that UST imaging of the brain is feasible through a skull-mimicking phantom for a variety of targets. In silico, in vitro and ex vivo targets could all be visualized in sound speed images with a skull phantom present despite the presence of cycle skipping artifacts. Additionally, an intact human brain sample was imaged to demonstrate our current ability to visualize anatomical features, and therefore guide future development of this work.