PROJECT SUMMARY
The overall goal of this proposal is to improve histotripsy treatments of liver tumors by developing an accurate,
clinically translatable workflow for targeting liver tumors transcostally. Globally, there are ~780,000 deaths from
liver cancer each year and in the US, this is expected to rise by more than 150% in the next 30 years. Histotripsy
is an emerging noninvasive, nonthermal and nonionizing focal tumor treatment that has recently undergone a
successful Phase I human clinical trial for treatment of liver tumors. It uses focused, short duration ultrasound
pulses to produce tissue cavitation at a focal point. Cavitation occurs from nanometer scale gas pockets present
in tissues, which rapidly expand and collapse, causing a high stress and strain that mechanically disrupts
adjacent tissues and cells. Histotripsy damage is a binary process where tissue is either destroyed or it isn’t,
providing many benefits over current thermal ablation techniques. A critical limitation of histotripsy is the sole
use of ultrasound to visualize and target the tumors. Targeting with diagnostic US limits treatments to a subcostal
approach due acoustic blockage from the rib cage, leaving an estimated ~50% of liver tumors non-visualizable
and therefore, untreatable. With the recent development of cone-beam CT-guided histotripsy, tumors can be
visualized throughout the liver, including beneath ribs, but cannot be accurately targeted due to acoustic
aberrations from intervening ribs potentially altering treatment size and location. In this proposal, we will develop
an accurate, clinically translatable workflow for targeting liver tumors for transcostal histotripsy. This will be
accomplished by first quantifying the effect intervening ribs have on targeting accuracy through ex-vivo and in-
vivo experiments (porcine model) (Aim 1). Secondly, a robust machine learning model will be developed to
predict targeting offsets, which will be trained and validated with ex-vivo, in-vivo, and in-silico datasets (Aim 2).
Finally, the model will be incorporated and tested to deliver patient-specific treatment plans in rabbit VX2 liver
tumors (Aim 3). This proposal includes radiographic image analysis of ex-vivo and in-vivo data, histopathologic
assessment of in-vivo treatments (Aims 1 and 3) and development and validation of a robust machine learning
algorithm to create a multi-dimensional analysis of a novel targeting method for histotripsy treatments. An
accurate and robust targeting method for performing transcostal histotripsy of liver tumors is crucial to expanding
the cohort of patients eligible for the therapy. It has the potential to almost double the number of patients with
liver tumors eligible for the treatment and the technique is translatable to tumors in other organs (eg, kidney and
pancreas) as well.