Development of Efficient Dynamic Magnetic Resonance Imaging Methods with Application to Breast Cancer Detection and Diagnosis.

Abstract

The goal of this predoctoral fellowship research project is to improve the temporal and spatial resolutions in dynamic contrast-enhanced magnetic resonance imaging of the breast by optimizing the Reduced-encoding Imaging by Generalized-series Reconstruction (RIGR) method. Specifically, we investigated the use of non-Fourier encoding for collecting the reduced encoding dynamic data sets. The conclusion from our study was that the current SVD encoding method biases the results towards reproducing the known features in the reference image and, therefore, is not appropriate for dynamic imaging applications. For that reason, we continue to acquire the dynamic data using Fourier encoding. Next, we incorporated dynamic information into the basis functions of the generalized-series model used by the RIGR algorithm. The TRIGR method resulted from incorporating information about the dynamic changes into the basis functions. Explicit edge constraints derived from the reference image were then used along with the contrast information from the dynamic data to inject dynamic information into the basis functions for both RIGR and TRIGR. Of these, the TRIGR method works better for contrast-enhanced imaging because the active reference image can be used for the edge extraction step.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 01, 1996
Accession Number
ADA320354

Entities

People

  • Jill M. Hanson
  • Paul Lauterbur

Organizations

  • University of Illinois Urbana–Champaign

Tags

DTIC Thesaurus Topics

  • Acquisition
  • Algorithms
  • Breast Cancer
  • Coding
  • Contrast
  • Data Acquisition
  • Data Sets
  • Detection
  • Engineering
  • Extraction
  • High Resolution
  • Image Reconstruction
  • Low Resolution
  • Magnetic Resonance
  • Magnetic Resonance Imaging
  • Neoplasms
  • Resonance

Readers

  • Image Processing and Computer Vision.
  • Neural Network Machine Learning.