Computer-aided identification of coronary artery disease

The main aim of the this project is to develop image analysis algorithms to provide automatic, reproducible, and quantitative analysis of non-invasively acquired CT images of the heart to identify patients who need to undergo percutaneous coronary treatment. The overall goal is to replace unnecessarily performed high-cost and invasive X-ray coronary catheterization and invasive measurement of coronary fractional flow reserve (FFR) by simple and fast non-invasive alternative.

Coronary artery disease (CAD) is the first cause of morbidity and mortality in the Western world and rising one in the developing countries. In clinical routine, patients with CAD are increasingly identified using non-invasive coronary CT angiography (CCTA), a non-invasive imaging tool for detection and exclusion of the obstructive coronary artery stenosis. Despite its high sensitivity CCTA is currently not capable of determining the functional significance of the detected stenosis. Therefore, after undergoing CCTA many patients are referred for additional testing with other imaging modalities to establish the functional significance of coronary stenoses. The goal of this project is to better use the information present in the CT data to extract information about the hemodynamic significance of coronary stenoses. This is important as clinically only treatment of functionally significant stenosis has proven to reduce CAD morbidity and mortality. Conversely, treatment of stenotic coronary arteries without hemodynamically significant stenosis is associated with more harm than benefit. Currently, many patients undergo invasive coronary angiography (ICA) prior to treatment, 30-40% unnecessary. During ICA, FFR, a quantitative marker of functional stenosis significance is determined. FFR is currently considered the reference standard for determination of the significance of coronary stenoses and is clinically used to guide treatment.

The main challenge in this project is to design an accurate, reproducible and quantitative method to determine which coronary artery stenoses as seen on CCTA images are functionally significant, and thereby to identify patients who need to undergo invasive coronary catheterization and spare those who do not. More specific, the challenges of this project are:

1. to define imaging protocols that combine CCTA and cardiac CT perfusion (CCTP), thereby exploiting the trend that acquisition of CT images already is becoming standard practice;
2. to develop automatic and objective image processing and analysis methods that are able to extract data relevant for diagnosis of CAD;
3. to extract quantitative markers indicative for the relevance of coronary artery obstruction and thus for selecting the preferred therapy or intervention
4. to exploit the combination of CCTA and CCTP data to enhance FFRCT assessment

The proposed method will exist of three main parts:

1. information about the geometry and morphology of the coronary arteries will be analyzed using CCTA images. Therefore, automated segmentation of the coronary arteries, detection and quantification of atherosclerotic plaque, detection and quantification of stenoses, and analysis of contrast attenuation in the coronary artery lumen will be performed. Furthermore, CCTP images provide information about the presence of ischemia in the myocardium. By relating the myocardial territory exhibiting perfusion deficiency with the coronary artery or its segment perfusing that territory, information about the location of the stenosis causing the ischemia will be obtained. For this purpose, segmentation and parcellation of the myocardium will be performed and subsequently, areas of hypo-perfusion in the myocardium will be delineated, quantified and related in CCTA images acquired at cardiac stress and rest.
2. analysis of CCTA and CCTP flow dynamics will be carried out to identify regions with impaired myocardial perfusion. Patient specific fluid dynamics modeling to analyze and quantify myocardial perfusion to identify alternative markers is currently under development but is still at its infancy and not yet available for direct clinical use;
3. the prognostic value of these two approaches will be compared.

In addition, we will investigate whether a novel approach, based on a combination of these two different approaches, leads to an improvement in performance. To ensure robustness of the designed method and enable subsequent commercialization, each marker will be developed with and evaluated using images from multiple hospitals with CT scanners of different vendors.

When successful, our approach will lead to reduction of healthcare costs by decreasing catheterizations carried out solely for diagnostic purposes. This means that, many patients will be spared a costly procedure that is still associated with non-negligible risks of morbidity and mortality as invasive and expensive coronary angiography could be replaced by simple and fast non-invasive imaging test.

Ivana Išgum graduated in Mathematics at the University of Zagreb, Croatia in 1999. The same year she was employed as a scientific software engineer at Silicon Biomedical Instruments BV, The Netherlands. In 2001 she became a PhD student at the Image Sciences Institute. Her PhD degree was obtained in 2007 with a thesis entitled Computer-aided detection and quantification of arterial calcifications with CT. She then worked for a year as a PostDoc at the Laboratory for Clinical and Experimental Image Processing in Leiden University Medical Center on the detection of atherosclerotic carotid plaque from combined magnetic resonance angiography and vessel wall images. Ivana is currently an Associate Professor at UMC Utrecht where she is developing methods for automatic calcium scoring and their application to large scale screening trials. She is also working on projects related to automatic segmentation of the developing neonatal brain with MRI.