Rapid 3D Ultrasound Tomography Reconstruction Methods for Guided Interventions

Michigan State University, East Lansing MI

Investigators

Linked publications & trials

Abstract

Project Summary/Abstract Background: As a portable non-radioactive modality, ultrasound has been increasingly used in guided interven- tions such as biopsy and surgery procedure in breast, prostate, brain, face and neck. Current 2D handheld ultra- sound and 3D Automated Breast Ultrasound System (ABUS) both use only ultrasound reflection data to generate images. 3D Ultrasound Computed Tomography (USCT) was developed to use both reflection and transmission data to provide improved image quality and potentially better diagnostic value. Challenge: USCT image reconstruction presents a historical challenge of heavy computational complexity, due to its non-linear, non-convex nature. To our best knowledge, all existing USCT algorithms with high fidelity are iterative, optimization-based, and thus suffer a heavy computation load. This computation load is especially cumbersome when higher number of transducers are added to the system to obtain more anatomical information. Therefore, there is an urgent need to develop an USCT imaging method to provide high fidelity and high speed at the same time to satisfy the requirement of guided intervention. Method: We hypothesize that boundary control method will achieve non-iterative USCT image reconstruction, leading to significant increase in computational efficiency while warranting fidelity and robustness to noise. This idea has been mathematically proven and validated by our preliminary research with a 10-fold increase in com- putation speed while maintaining high fidelity level. The developed method will serve as an ideal non-radioactive intra-operative imaging guide, and bring new perspective to USCT imaging reconstruction algorithm research. Al- though this project is not intended for clinical use, we will perform a virtual clinical trial to systematically evaluate the developed system with computationally simulated phantoms, 3D-printed phantoms, and digital patient-based phantoms. Impact: Upon completion, this work will have achieved a computationally light, high-fidelity, near real-time 3D USCT imaging system, ideal for guided intervention. This system has the potential to: · Efficiently support considerably more transducers without compromising image quality. · Enable portable USCT to provide powerful tools for guided intervention and other clinical applications. · Lead to genuine real-time 3D USCT imaging and inspire new applications of USCT.