Access Type

Open Access Dissertation

Date of Award

January 2010

Degree Type

Dissertation

Degree Name

Ph.D.

Department

Biomedical Engineering

First Advisor

Ewart M. Haacke

Abstract

Modern magnetic resonance imaging sequences allow detailed non-invasive imaging of both the arteries and veins. This work is divided into four sections that examine different applications and analysis of these sequences.

Susceptibility weighted imaging (SWI) typically generates excellent negative venous contrast. Techniques to generate positive arterial contrast in SWI images without degrading the venous contrast with a single echo are examined. By using high isotropic resolution and high readout bandwidth flow losses can be minimized (generating good arterial contrast) even at the long echo times required for good venous contrast. A downsampling filter is then used to restore lost venous phase contrast caused by isotropic resolution. Using these techniques good arterial contrast distal to the middle cerebral artery (MCA) and good venous contrast can be obtained from a single SWI scan.

Traditional magnetic resonance angiography (MRA) is unable to image arteries smaller than approximately 500 µm. The role of flip angle, echo time, image subtraction, and resolution were examined to increase contrast and make much smaller arteries in the 100-200 µm range visible. While hints of these vessels appear, calculations based on current optimized signal levels show that a resolution of 100 µm is required to reliably image 100 µm vessels. Currently these resolutions are not attainable with clinical human scanners limiting our ability to image smaller vessels.

Cerebral microbleeds (CMB) are increasingly being recognized as an important biomarker for other neurovascular diseases. A technique is presented that semi-automatically identifies CMBs in SWI. This will both reduce the processing time and increase the consistency over manual methods. This technique relies on a statistical thresholding algorithm to identify hypointensities within the image. A support vector machine (SVM) supervised learning classifier is then used to separate true CMB from other marked hypointensities. The classifier relies on identifying features such as shape and signal intensity to identify true CMBs. The results from the automated section are then subject to manual review to remove false positives. This technique is able to achieve a sensitivity of 81.7% compared to the gold standard of manual review and consensus by multiple reviewers. This presents a much faster alternative to current manual techniques at the cost of some lost sensitivity.

The settling properties of venous blood in the peripheral vasculature during periods of immobility were examined. SWI images were collected for nine subjects at two time points: within ten minutes of entering the magnet and after 40 minutes spent stationary in the magnet. Changes in the phase and in the distribution of phase of the veins were used to draw conclusions about the separation of red blood cells from plasma over time. Settling was observed to occur in eight of the nine subjects, the only exception being the youngest subject (18 years). The bottom half of some veins were seen to darken while the top half showed little change often with a clear dividing line between the two. Phase values measured in the bottom layer were consistent with the layer consisting entirely of red blood cells. Settling was seen to increase with time spent stationary and to correlate with the size of veins in the calf. Older subjects tended to have larger veins and consequently more settling of the red blood cells. Our results show that even 40 minutes of rest can easily lead to settling of the blood depending on the position of the leg.

Download

Share

COinS