Image analysis for 4D MR velocity mapping

Dutch Technology Foundation (STW)

Jos J. M. Westenberg PhD
Patrick J.H. de Koning, MSc
Mohammed S. Elbas, MSc

Background

Magnetic resonance imaging (MRI) allows acquisition of three dimensional (3D) image volumes with encoding of blood flow or tissue velocities in three orthogonal directions throughout the cardiac cycle. When applied to the heart, this MRI technique (4D VENC MRI) can be used to depict the blood flow through all the cardiac valves simultaneously, making it a highly promising tool to assess valvular disease. In addition, 4D VENC MRI enables the visualization of the 3D time-varying intra-cardiac blood flow providing relevant diagnostic information on underlying cardiac pathologies such as heart failure, congenital heart disease and valvular disease. Despite its tremendous potential, the clinical usage of 4D VENC MRI to asses valvular and intra-cardiac flow abnormalities in one scan is hampered by the amount and complexity of the data per patient study (5,000-10,000 images). To make 4D VENC MRI suitable for clinical application, dedicated image analysis techniques need to be developed for feature extraction, quantification and visualization that enable a time-efficient and intuitive analysis of the enormously large image data sets. What the radiologist / cardiologist needs is a "single analysis button" that executes the automated extraction of clinically relevant information about the function of the heart, all within a matter of minutes.

Goals

The goal of this project is to develop and validate novel image processing and visualization methods for extraction of clinically relevant information from three-dimensional phase-contrast MRI image data with velocity encoding in three directions (4D VENC MRI). The specific focus is on development of methods for quantitative and visual analysis of blood flow patterns in the heart chambers, around the valves and through the valves.

The project has the following sub-goals:

  • Development of optimized protocols for MR imaging and analysis for whole-heart 4D VENC MR, and specialized sub-protocols for specific clinical questions.
  • Development of methods for time-resolved 3D visualization of blood flow patterns in relation to anatomical information derived from additionally available MR scans.
  • Development of methods for automated detection and tracking of heart valves in the acquired 3D data sets for subsequent quantification of trans-valvular flow including correction for motion of the valve attachment.
  • Development of methods for detection and visualization of blood-flow abnormalities in the heart or around the valves.
  • Development of methods for automated detection of ventricular function abnormalities, such as systolic LV dyssynchrony by combining intra-cardiac flow patterns and ventricular wall motion patterns
  • Technical and clinical validation of the abovementioned methods in selected patients.

Contact

Rob J. van der Geest, PhD.
Division of Image Processing
Department of Radiology, 1-C2S
Leiden University Medical Center
P.O. Box 9600
2300 RC Leiden, The Netherlands
Tel. +31 (0)71 526 2138
e-mail: R.J.van_der_Geest@lumc.nl 

B.P.F. Lelieveldt PhD,
Division of Image Processing
Department of Radiology, 1-C2S
Leiden University Medical Center
P.O. Box 9600
2300 RC Leiden, The Netherlands
Tel. +31 (0)71 526 1882
email: B.P.F.Lelieveldt@lumc.nl