Medical Imaging Systems Techniques and Applications

Cardiovascular Systems

Cornelius T Leondes

• Publisher: Routledge
• Year: 2014
• Print ISBN: 9789056995096
• eBook ISBN: 9781134383979
• Edition: 1
• Topic: Medicine
• Subtopic: Health Care Delivery

First Published in 2004. This is Volume I of six of a series on medical imaging systems techniques and applications. This subject area exemplifies a meaningful manifestation of the power of the technologies of the second industrial revolution. The first chapter in this volume on cardiovascular systems emphasizes the importance of accurate measurements of cardiac shape and dynamics as they reflect the scope of cardiac diseases, the major cause of mortality in developed countries today. Cardiac imaging plays an important role in this regard, and almost the only one in this clinical context.

3 TECHNIQUES IN IMAGE SEQUENCE FILTERING FOR CLINICAL ANGIOGRAPHY*

CHEUK L. CHAN,¹,^† A.K. KATSAGGELOS² and A.V. SAHAKIAN²

¹PAR Government Systems Corporation, 1010 Prospect Street, La Jolla, CA, 92037-4146, USA

²Department of Electrical Engineering and Computer Science, Northwestern University, Evanston, Illinois 60208-3118, USA

3.1. INTRODUCTION

The assessment of coronary diseases through the evaluation of contrast imaging has been in existence for nearly a quarter century and has unquestionably aided physicians in their clinical exams. Of equal importance is a conscientious effort to reduce the risk factors of high X-ray exposure to both the patient and the medical staff. These two conflicting demands have often resulted in the degradation of image quality caused by the acquisition of radiographic images at low radiation levels. It is this noise that is treated here, since it has the greatest bearing on the tradeoff between radiation exposure and image quality. The aim of this chapter is to address this problem by first describing the process of radiation transport as a stochastic processs. Based on this description, a noise model is proposed from which a 3-D linear filter is derived to remove this noise. Its recursive counterpart is also presented here as a means of an efficient implementation. Finally, experimental results will be shown on clinical fluoroscopic images with both simulated and actual quantum mottle.

While both of these linear filters have given favorable results in terms of improving the image quality in mottled sequences, recent work in nonlinear filtering and simultaneous motion estimation and filtering may change the impetus of research in this area of clinical angiography. Some preliminary topics and results as they apply to this application will be pointed out in this chapter as well.

3.2. BACKGROUND

Historically, in digital subtraction angiography (DSA), vessels of interest were imaged by acquiring a series of X-ray images containing iodine contrast injected peripherally and subtracting them from a “mask” image of the same area obtained before opacification.⁷⁹ This noninvasive angiographic technique produces images suitable for diagnostic purposes if the structures of the mask image and contrast images are registered such as those found in renal or intracranial vasculatures. In addition, relatively low frame rate (1–2 frames/sec) image acquisitions (with a corresponding increase in exposure time) are possible for noncardiac applications if patient motion is limited. This increased radiation exposure (1 mR per frame) allows for improved contrast resolution and thus greater visibility of the area of interest.

More recently, particularly in cardiac applications, interest has focused on the use of selective intraarterial contrast administration (through the use of catheters) for the assessment of coronary disorders and visualization of higher order coronary branches. In these applications, the images are of sufficient high enough quality that image subtraction is not necessary. Correspondingly, frame rates are also higher for cardiac applications (30 frames/sec) since a longer pulse width such as that used in DSA would cause motion blurring of the vessels thus reducing the spatial resolution of the image.

Typically, for these cardiac applications, 100 to 200 frames are acquired at full radiographic levels for diagnosis (about 25 μR at the input of the image intensifier) integrating to a cumulative exposure of 3.8 mR to the image intensifier for 150 frames.⁷⁶ In addition, the guidance of the catheter to the coronary region requires occasional images to be acquired at fluoroscopic levels in those parts of the body where manipulation of the catheter is difficult. At these levels, quantum mottle is easily visible. Goldberg et al.,³⁵ have also considered the use of “boosted fluoroscopy” levels (i.e., tube current between fluoroscopic levels and full radiographic levels). These images may be sufficient for the location of stenotic regions as opposed to images acquired for quantitative coronary arteriography where full radiation levels are desirable. These different radiation levels have been developed for different tasks and the decision of which level to use has always been heavily weighed by the tradeoff of image quality versus patient safety. In addition, lower doses than these conventional levels may be desirable if post-processing of the images is possible.

Though the radiation can be lowered by acquiring frames at a lower rate (same pulse width and tube current as those in 30 frames/sec applications, but using a longer separation between pulses), objectionable flickering may result due to the jerkiness of the motion.²⁹ In addition, cardiologists often prefer a higher frame rate in order to...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Table of Contents
• Series Description and Motivation
• Series Preface
• Preface
• Three-Dimensional Analysis of Heart Geometry and Function
• Techniques in the Assessment of the Left-Ventricular Function
• Techniques in Image Sequence Filtering for Clinical Angiography
• Automated Analysis of Coronary Angiograms
• Computer Analysis of Intravascular Ultrasound Images