PLAQMR

PLAQMR: Automated Carotid PLaque Analysis from Quantitative MR Imaging

Summary

Rupture of the atherosclerotic plaque in the carotid artery is considered a major cause of ischemic stroke. For risk assessment of carotid atherosclerotic plaque rupture, accurate measurements of the spatial distribution of calcium, hemorrhage, lipids and fibrous tissue within the plaque are necessary. Magnetic resonance (MR) imaging offers good soft tissue contrast, and several studies have shown that the different tissue types in carotid plaque can visually be distinguished using the conventional T1-weighted, T2-weighted and PD-weighted images (1). The prognostic value of volumetric plaque measurements solely, e.g. as determined by ultrasound, is much lower. Reproducible, automated, and quantitative analysis based on MR data is still challenging though (2). Complicating factors for automatic plaque segmentation are the lack of a normalized intensity unit for MRI, motion artefacts, and the rather low (anisotropic) resolution. In this research project, we will develop image analysis tools for automated quantitative plaque composition measurements using MRI. In contrast to previous research, we consider the entire pipeline: from acquisition to image analysis. Traditionally, the scanning protocols are optimized for visual assessment by the radiologist, and, given these protocols, image analysis methods are developed. We propose an integrated approach, in which MR protocols and quantitative image analysis methods are concurrently optimized, with the ultimate aim to maximize the precision and accuracy of the plaque segmentation. Major research directions within this project are 1) the use of quantitative MRI techniques (using MR relaxometry) to improve the reproducibility of the segmentation, 2) the development of multispectral segmentation based on absolute T1, T2, and PD tissue properties 3) integrated motion compensation, and 4) determination of the optimal combination of high resolution T1/T2/PD weighted MRI and low resolution quantitative MRI. For validation, we will develop and use a phantom and perform volunteer and patient studies. The plaque segmentations will be compared to histology of excised atherosclerotic specimens. An important aspect throughout the project will be the validation on two different MRI scanners (Philips and General Electric). The image analysis approaches will be designed such that the dependence of the resulting quantitative plaque characteristics on the scanner type is as limited as possible. This aspect has often been ignored in previous studies. The main result of this project will be an automated image analysis tool for carotid plaque segmentation, in combination with an MR scanning protocol optimized for this specific segmentation method.

Project Team

  • Aart J. Nederveen PhD, a.j.nederveen@amc.nl – Project Leader
  • Stefan Klein PhD, s.klein@erasmusmc.nl  - Project Leader
  • Rob J. van der Geest PhD, r.j.van_der_geest@lumc.nl - Project Leader
  • Bram F. Coolen PhD, b.f.coolen@amc.nl – PostDoc
  •  Shan Gao, s.gao@lumc.nl – PhD student
  • Henk Smit  h.smit88@gmail.com – PhD student

 

Publications and Abstracts