A new approach to ultrasonic elasticity imaging

Cameron Hoerig, Jamshid Ghaboussi, Mostafa Fatemi, Michael F. Insana

Research output: Chapter in Book/Report/Conference proceedingConference contribution

5 Scopus citations

Abstract

Biomechanical properties of soft tissues can provide information regarding the local health status. Often the cells in pathological tissues can be found to form a stiff extracellular environment, which is a sensitive, early diagnostic indicator of disease. Quasi-static ultrasonic elasticity imaging provides a way to image the mechanical properties of tissues. Strain images provide a map of the relative tissue stiffness, but ambiguities and artifacts limit its diagnostic value. Accurately mapping intrinsic mechanical parameters of a region may increase diagnostic specificity. However, the inverse problem, whereby force and displacement estimates are used to estimate a constitutive matrix, is ill conditioned. Our method avoids many of the issues involved with solving the inverse problem, such as unknown boundary conditions and incomplete information about the stress field, by building an empirical model directly from measured data. Surface force and volumetric displacement data gathered during imaging are used in conjunction with the AutoProgressive method to teach artificial neural networks the stress-strain relationship of tissues. The Autoprogressive algorithm has been successfully used in many civil engineering applications and to estimate ocular pressure and corneal stiffness; here, we are expanding its use to any tissues imaged ultrasonically. We show that force-displacement data recorded with an ultrasound probe and displacements estimated at a few points in the imaged region can be used to estimate the full stress and strain vectors throughout an entire model while only assuming conservation laws. We will also demonstrate methods to parameterize the mechanical properties based on the stress-strain response of trained neural networks. This method is a fundamentally new approach to medical elasticity imaging that for the first time provides full stress and strain vectors from one set of observation data.

Original languageEnglish (US)
Title of host publicationMedical Imaging 2016: Ultrasonic Imaging and Tomography
PublisherSPIE
Volume9790
ISBN (Electronic)9781510600256
DOIs
StatePublished - 2016
EventMedical Imaging 2016: Ultrasonic Imaging and Tomography - San Diego, United States
Duration: Feb 28 2016Feb 29 2016

Other

OtherMedical Imaging 2016: Ultrasonic Imaging and Tomography
CountryUnited States
CitySan Diego
Period2/28/162/29/16

    Fingerprint

Keywords

  • Autoprogressive Method
  • Finite Element Analysis
  • Machine Learning
  • Neural Network Constitutive Model

ASJC Scopus subject areas

  • Atomic and Molecular Physics, and Optics
  • Electronic, Optical and Magnetic Materials
  • Biomaterials
  • Radiology Nuclear Medicine and imaging

Cite this

Hoerig, C., Ghaboussi, J., Fatemi, M., & Insana, M. F. (2016). A new approach to ultrasonic elasticity imaging. In Medical Imaging 2016: Ultrasonic Imaging and Tomography (Vol. 9790). [97900G] SPIE. https://doi.org/10.1117/12.2216549