Automated Vascular MRI Evaluation

Automated Evaluation of Vascular MR Image data

Dutch Technology Foundation (STW LGT.6454)
Patrick de Koning, MSc
Ronald van 't Klooster, MSc

Background

Magnetic resonance imaging (MRI) is playing an increasingly important clinical role as a non-invasive imaging modality for the evaluation of the vascular system. Magnetic resonance angiography (MRA) provides three-dimensional (3D) images of the vascular lumen, allowing detection of stenoses based on 3D image information. MRI vessel wall imaging (MR-VWI), on the other hand, provides information about the vessel morphology, and also information about the tissue components within the vessel wall based on high-resolution 2D cross-sectional images. While MRA is mainly of value for the detection of vascular stenoses or aneurisms, the strength of MR-VWI is in the early detection of the formation of plaque in the arterial wall and the identification of the type of plaque. Various studies have demonstrated the potential value of MR-VWI for the identification of 'vulnerable' plaque as opposed to stable plaque. Vulnerable plaques are rupture-prone plaques, which are mostly associated with mild to moderate luminal narrowings (20-50% luminal reduction) and exhibit well-defined histological characteristics; these are also often referred to as "thin-cap fibroatheromas". In addition to the presence of a thin fibrous cap, plaques are also thought to be vulnerable when they contain an elevated number of macrophages, and a large lipid/necrotic core. Vulnerable plaque(s) can be identified prospectively by MRI, followed by a proper therapeutic intervention; this would result in the prevention of the serious complications of plaque rupture.
As described above, the introduction of these new non-invasive MR imaging techniques will provides important clinical benefits. However, visual analysis of all the acquired image data is difficult due to the large number of images acquired in a typical MR examination. A typical MRA data set consists of between 500 and 1500 images and a typical VW dataset contains several tens of images. Accurate quantification of vascular stenoses from MRA data alone is hampered by the relatively low image resolution. The current state-of-the-art for the analysis of MRA data is based on visual assessment of Maximum Intensity Projections (MIP renderings) from multiple viewing angles, which result in loss of 3D information and loss in signal-to-noise ratio (SNR). Visual image analysis is also subject to inter- and intra-observer variabilities. Also VWI-MR images are currently mainly analyzed visually, or quantitatively using observer-dependent manual contour tracing tools. When multiple scans are acquired, such as combinations of MRA and VWI, currently no versatile software algorithms are available to link the image data from the MRA scan with the VWI scan.

Goals

In the proposed project the research is focused at developing novel knowledge-guided automated image segmentation methods for quantitative analysis of vascular MR image data by integrating information from both the MRA and MR-VWI techniques. The combination of MRA and MR-VWI is potentially a very powerful combination providing a detailed description of vascular pathologies in a non-invasive manner.

Approach

The MRA image data is used to obtain a 3D segmentation of the vascular lumen. After image registration, the segmented MRA lumen is used as initialization for the lumen and outer wall contour detection in the 2D VWI slices. Automated registration of the various VWI sequences will be performed such that for each position in the vessel wall the intensity values of the obtained VWI sequences can be derived. For each pixel in the vessel wall various features will be computed derived form the image data. Using pattern recognition techniques, each pixel will be classified as either being part of a certain plaque component, or as normal vessel wall.

Processing pipeline
Figure 1
Image processing pipeline

MRApathline
Figure 2 Vessel pathline automatically detected in the MRA volume

MRA segmentation
Figure 3.
Automated segmentation of the lumen boundaries from the MRA volume 

Segmenationsteps
Figure 4
Segmentation steps for the lumen and outer contours in the T1-weighted images

Screenschot of VesselMASS
Figure 5
Screen shot of the VesselMASS software package showing a manually plaque classification result for the common and internal carotid artery of a patient. The upper left panel shows the 6 available sequences for the selected slice level: 1) Time-of-flight, 2) MPRAGE, 3) Contrast-enhanced MRA, 4) T2-weighted, 5) T1-weighted, 6) T1-weighted post-contrast. The bottom panel shows the T1-weighted images of all acquired 18 slices in the current study.

Red=Lumen contour, Green=Outer contour, Yellow=Lipid tissue, Blue=hemorrhagic tissue.

Status

Ongoing

Publications

  1. Guzman R, Remonda L, de Koning PJH, van der Geest RJ, Oswald O, Schroth G. Correlation of quantitative MR angiography of the carotid artery with in-vivo measurement during carotid endarterectemy. In: Computer Assisted Radiology and Surgery - CARS 2002. HU Lemke, MW Vannier, K Inamura, AG Farman, K Doi, JHC Reiber (Eds.). Elsevier Science BV 2002: 917-922.
  2. de Koning PJH, Schaap JA, Janssen JP, Westenberg JJM, van der Geest RJ, Reiber JHC. Automated segmentation and analysis of vascular structures in magnetic resonance angiographic images. Magn Reson Med 2003;50; 1189-1198.
  3. Adame, IM, van der Geest RJ, Mohamed M, Wasserman BA, Reiber JHC, Lelieveldt BPF. Automatic Plaque Characterization and Vessel Wall Segmentation in Magnetic Resonance Images of Atherosclerotic Carotid Arteries, SPIE. Medical Imaging 2004; 5370,:265-273, 2004.
  4. Schaap JA, de Koning PJ, Janssen JP, van der Geest RJ, Reiber JH. Quantitative analysis of vascular images, in particular of abdominal aorta aneurysms from 3D CTA data sets. Stud Health Technol Inform. 2004;103:252-258
  5. Adame IM, van der Geest RJ, Wasserman BA, Mohamed M, Reiber JHC, Lelieveldt BPF. Automatic segmentation and plaque characterization in atherosclerotic carotid artery MR images
    MAGMA (Magnetic Resonance Materials in Physics, Biology and Medicine) 2004;16 (5): 227-234.
  6. de Vries M, de Koning PJ, de Haan MW, Kessels AG, Nelemans PJ, Nijenhuis RJ, Planken RN, Vasbinder GBC, van Engelshoven JMA, van der Geest RJ, Leiner T. Accuracy of Semiautomated Analysis of 3D Contrast-Enhanced Magnetic Resonance Angiography for detection and quantification of aortoiliac stenoses. Invest Radiol 2005;40:495-503.
  7. Adame IM, de Koning PJH, Lelieveldt BPF, Wasserman BA, Reiber JHC, van der Geest RJ. An integrated automated analysis method for quantifying vessel stenosis and plaque burden from carotid MRI images: Combined postprocessing of MRA and vessel wall MR. Stroke 2006;37(8):2162-2164.
  8. Adame IM, van der Geest RJ, Bluemke DA, Lima JA, Reiber JHC, Lelieveldt RBF. Automatic vessel wall contour detection and quantification of wall thickness in in-vivo MR images of the human aorta. J Magn Reson Imaging. 2006;24:595-602
  9. Makowski P, de Koning PJH, Angelié., Westenberg JJM, van der Geest RJ, Reiber JHC. 3D Cylindrical B-Spline Segmentation of Carotid Arteries from MRI Images. In: Lecture Notes in Computer Science, Biomedical Simulation. Springer Berlin/Heidelberg 2006;188-196
  10. Schoonman GG, Bakker D, Schmitz N, van der Geest RJ, van der Grond J, Ferrari MD, van Buchem MA. Magnetic resonance angiography of the human middle meningeal artery: implications for migraine. J Magn Reson Imaging. 2006; 24:918-921.
  11. Zudilova-Seinstra EV, de Koning PJH, Suinesiaputra A, van Schooten BW, van der Geest RJ, Reiber JHC, Sloot PMA. Evaluation of 2D and 3D glove input applied to medical image analysis. International Journal of Human Computer Studies 2009,doi:10.1016/j.ijhcs.2009.08.001.
  12. Zudilova-Seinstra EV, de Koning PJH, Suinesiaputra A, van Schooten BW, van der Geest RJ, Reiber JHC, Sloot PMA. Evaluation of 2D and 3D glove input applied to medical image analysis. INTERNATIONAL JOURNAL OF HUMAN-COMPUTER STUDIES 2010;68 (6) 355-369.
  13. Asghar MS, Hansen AE, Kapijimpanga T, van der Geest RJ, van der Koning P, Larsson HBW, Olesen J, Ashina M. Dilation by CGRP of middle meningeal artery and reversal by sumatriptan in normal volunteers. Neurology 2010;75:1520–1526.
  14. Asghar MS, Hansen AE, Amin FM, van der Geest RJ, Koning PV, Larsson HB, Olesen J, Ashina M. Evidence for a vascular factor in migraine. Ann Neurol. 2011;69(4):653-645.
  15. van 't Klooster R, de Koning PJ, Dehnavi RA, Tamsma JT, de Roos A, Reiber JH, van der Geest RJ. Automatic lumen and outer wall segmentation of the carotid artery using deformable three-dimensional models in MR angiography and vessel wall images. Magn Reson Imaging. 2012;35(1):156-165.
  16. Suinesiaputra A, de Koning PJ, Zudilova-Seinstra E, Reiber JH, van der Geest RJ. Automated quantification of carotid artery stenosis on contrast-enhanced MRA data using a deformable vascular tube model. Int J Cardiovasc Imaging. 2011 Dec 9. [Epub ahead of print].
  17. Amin FM, Asghar MS, Guo S, Hougaard A, Hansen AE, Schytz HW, van der Geest RJ, de Koning PJ, Larsson HB, Olesen J, Ashina M. Headache and prolonged dilatation of the middle meningeal artery by PACAP38 in healthy volunteers. Cephalalgia. 2011 Dec 15. [Epub ahead of print].

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
Fax. +31 (0)71 526 6801
e-mail: R.J.van_der_Geest@lumc.nl