work

Download my curriculum vitae (pdf) here.

Currently I work as a researcher on medical image analysis at the department of Electrical Engineering of the Eindhoven University of Technology (TU/e) (Signal Processing Group).
From 2005 until 2011, I worked as an assistant professor in Cardiac Image Analysis (CIA/e) at the same university. I was part of the Biomedical Image Analysis (BMIA) group within Biomedical Engineering (BME) department of TU/e.


Figure 1. Motion field of the LV extracted with optical flow from MR tagging data

In March 2005 I joined BMIA, after having finished my PhD (download PhD thesis here – pdf, 16MB) (promotor: prof. Hans Reiber) on active shape modelling for cardiac segmentation at LKEB, a medical image processing lab within the radiology department of the Leiden University Medical Center (LUMC). Besides my research within the Molecular Imaging of Ischemic Heart Disease (funded by BSIK), I have set up and developed a cardiac research line, Cardiac Image Analysis Eindhoven (CIA/e). Currently, CIA/e contains two main foci, viz., cardiac motion analysis (see figure 1 (movie) to the right) and cardiac intervention support (see figure 3 below).

Cardiac motion analysis
Curently we investigate the possibilities of extracting motion from cardiac tagged MRI images. The motion of the the cardiac left ventricle (LV) contains much information about cardiac function. Cardiologists evaluate cardiac images (MR, CT, Ultrasound, among others) to diagnose patients. This evaluation is mostly qualitative and depends often on the experience of the clinician. Both systolic and diastolic function are interesting and important for diagnosis.

Figure 2. Circumferential strain in a basal short-axis slice in a high resolution AHA 17-segment equivalent representation

With the help of dedicated medical image analysis methods we design, investigate and develop, we aim at bringing quantification into the evaluation of cardiac function for diagnostic purposes. Important global functional parameters for the cardiac left ventricle are wall thickness and thickening, ejection fraction, stroke volume etc.
Local functional parameters can be obtained from cardiac motion, like the (local) amount of rotation, deformation and strain, and local timing of contraction and relaxation. Figure 2 (movie) shows the circumferential strain over time of a volunteer in a basal short-axis slice through the LV mapped to a high resolution equivalent of the American Heart Association 17-segment model, however for a single slice only.
Our larger aim is the development of a statistical model of normal LV motion, trained on a large database of volunteers.

Cardiac intervention support

The last decade has shown an increasing number of ablation procedures based on anatomic considerations.
The largest part concerns catheter ablation of Atrial Fibrillation (AF).
Knowledge of the true sites of the ablation lesions following ablation therapy for the treatment of atrial
fibrillation is of vital importance to electrophysiologists for the assessment of gaps in the ablation patterns
or large areas of edema that could be causing temporary electrical block.

Segmentation of the cardiac left atrium from high-resolution CT data.

Figure 3. Segmentation of the cardiac left atrium from high-resolution CT data.

The ability to image and quantify atrial wall properties, such as, thickness and acute RF ablation lesions
would be of great benefit in the early determination of gaps in the ablation patterns.
Emerging techniques rely on manual segmentation or thresholding of scar, introducing the potential for high
user error. To obtain information about local left atrial (LA) myocardial wall thickness, a pre-segmentation
tool including both endocardial and epicardial surface detection is required. Acute and chronic scar have to
be quantified for therapy success assessment.
Segmentation of the LA epicardial border and local atrial wall thickness quantification is largely unexplored
territory.
We propose to develop novel techniques for the quantification of LA wall properties of the LA (thickness,
edema, scar) and for segmentation of the esophagus. We propose to use coupled double segmentation
techniques based on either level sets or graph-theory. The image data used as an input will be from both
multi slice CT and high-resolution MR, the latter of which is developed at KCL, one of the partners.
We expect this project to yield these novel techniques, the performance of which will be fully evaluated,
leading to improved ablation results, and better assessment tools for therapy success.

Collaborations
For both topics, many collaborations with other groups, departments and institutes have been set up. The link with the TU/e Mathematical Image Analysis (MIA) group (prof. Luc Florack, dr. Remco Duits) mainly focuses on the cardiac motion part (mainly from MRI) and fiber orientation analysis from confocal and two-photon laser scanning microscopy data in the context of tissue engineering: muscle fibres (prof. Carlijn Bouten) and angiogenesis (dr. Daisy van der Schaft). These are more academically oriented topics. The link with the Biomedical NMR group (dr. Gustav Strijkers) within the BME department also focuses on cardiac motion, i.e. the question “how to improve image acquisition and analysis” to get the best results. To give an example, we have recently studied the effect of tag width on the robustness and accuracy of LV motion analysis. Together with the Cardiovascular Biomechanics group (dr. Peter Bovendeerd) we try to find the relationship between the motion of the LV and its myofiber directions.

Collaboration with Philips Healthcare (prof. Marcel Breeuwer, dr. Peter Rongen, dr. Geert Gijsbers) focuses on the development of new techniques for application in the clinic, and utilization by Philips.

QANU and BME
BMIA (as part of the BME department) was recently evaluated by QANU (Quality Assurance Netherlands Universities). BMIA was awarded 19.5 on a scale of 20. The two main research lines within BMIA (“neuro” and “cardio”) were specifically mentioned in the report being “very convincing”.

IST/e
Early 2010, a joint initiative of MIA, BMIA, biomedical NMR group of BME, and the Visualization group of the Computer Science department at TU/e to set up the Image Sciences and Technology Eindhoven (IST/e) was rewarded with a grant of 1 M Euro by the board of TU/e, with its High Potential Program. The two main lines with IST/e are “neuro” and “cardio”, which are supported by a third, fundamental image processing, research line. I was one of the applicants of this grant.

Grant proposals
Apart from IST/e I have written a number of grant proposals to request funding from the so-called “second funding stream” (NWO Physical Sciences, STW, NWO Innovational Research Incentives Scheme (VENI). The last VENI proposal I wrote was – although considered eligible for funding – not granted unfortunately due to lack of funds at NWO. The decision letter of NWO said about the proposal “your proposal is considered eligible for funding, but based on the ranking and due to limited funds this proposal could not be rewarded.” And about me personally “The committee considers dr. Van Assen a good and productive researcher with a good publication list. […].

STW Perspectief CARISMA
Recently, on August 30 2010, a program for grant proposals of STW called “Cardiovascular Risk Management by Advanced Medical Image Analsis”, acronym CARISMA, has been closed for submission. I have submitted a proposal for “characterisation of atrial wall properties for image guided management of atrial fibrillation”, called CHARIGMA. This proposal is a joint effort of TU/e, Philips Healthcare and King’s College London (prof. Tobias Schaeffter, dr. Kawal Rhode). Furthermore, I participate in another grant proposal, concerning image analysis for percutaneous aortic valve replacement (called Papaver) together with the Biomedical Engineering & Physics group of the Amsterdam Medical Center (AMC) (dr. Henk Marquering). Finally, I participate in a third proposal, called CARESSE, about the classification and visualization of myocardial tissue and and its feature space, initiated by prof. Marcel Breeuwer and dr. Anna Vilanova.