Healthcare projects
Below is an overview of the current BSc and MSc student projects. For more detailed information, please click on the respective button. Looking for something specific? Please feel free to contact one of the below researchers directly to discuss the possibilities.
Medical ultrasound techniques
Ultrasound is the most widely used diagnostic technique because of its versatility, ease of use, cost effectiveness, repeatability for patients and reproducibility. Until now the ultrasound diagnosis were based in 2D imaging,
but new developments with integrated electronics has open novel three-dimensional (3D) ultrasound imaging methods and systems for real-time imaging of heart, liver, kidney and other organs but also for monitoring the perfusion of the brains of preterm babies. If you like the combination of signal processing, computer simulations and experimental work and like to apply your work in the medical field than medical ultrasound imaging may just be something for you.
-
One in ten children is born prematurely, which amounts to approximately 500,000 preterm infants in Europe each year. Yet, they regularly show neurodevelopmental problems, including cognitive deficits, motor disabilities and psychiatric diseases with ensuing lifelong burdens for the up growing individuals and their families. A major cause of these neurodevelopmental problems is brain injury, linked to inadequate brain perfusion during and after delivery as shown in the figure (Two-dimensional perfusion map: Left, normal perfusion. Right, perfusion defect).
The Maresca Lab: Biomolecular Ultrasound Imaging
We pursue fundamental advances at the interface of ultrasound imaging physics and molecular engineering, taking advantage of the discovery of acoustic biomolecules to interface ultrasound waves with cellular processes. The resulting technologies will be applied in many areas of biology and medicine, and a primary focus of the lab is the non-invasive ultrasound assessment of vascular function at all scales, from arteries down to capillaries and cells.
KDAI Lab: AI for Medical Imaging
Recent advances in AI have led to major breakthroughs in many scientific fields, including medical imaging. Much of the current success can be attributed to the development of powerful deep learning methods that learn from massive data. However, many mysteries of deep learning remain to be unravelled, for example, what is the appropriate inductive bias in specific applications, what leads to vulnerability at test time, and what causes AI uncertainty from a Bayesian point of view. We are a TU Delft AI lab that strives to study such interesting problems, in the domain of medical imaging. We are driven by our curiosity to understand AI and enthusiasm to improve AI for medical imaging applications. Besides, we work on clinical utilization of AI, for various applications such as image quantification, clinical diagnosis, and risk prediction.
-
We have a wide range of research projects open for students at bachelor, master, and PhD level, in both technical and clinical flavors. Examples include generalizability of deep learning in medical image analysis, data visualization, risk stratification, image segmentation and registration.
Advanced ultrasound imaging devices and non-linear wave propagation
The realization of novel imaging devices and techniques poses many challenges. For example, a matrix array for making 3D medical images contains thousands of piezoelectric elements that can each transmit and receive ultrasound pulses. To deal with this large amount of signals, the elements are directly mounted on an electronic chip that performs the first stage of data processing. Typical challenges are: minimization of crosstalk between the elements, development of beamforming strategies for superfast scanning of a volume, and the development of imaging strategies for achieving superior 3D resolution. Another part of my research involves accurate computation of acoustic waves, in particular non-linear waves that form the basis of medical harmonic imaging. Most projects involve both experimental and numerical work, and can have a component that will be performed at Erasmus MC.
-
A real-time 3D ultrasound probe for intra-cardiac echography (ICE) could provide accurate guidance during minimally-invasive procedures for heart valve replacement and treatment of cardiac rhythm disorders. Currently, only two-dimensional visualization is possible. This project involves novel methods and techniques that will enable a real-time 3D ICE catheter. Rather than sequentially scanning the heart with a narrow steered ultrasound beam, the complete volume will be insonified in one shot by transmitting a non-steered, spherically expanding wave, originating from a virtual point source behind the array. This is much simpler and much faster than generating a focused beam. Our part of the project involves the design of the ultrasound array and the beamforming strategy.
Medical image analysis
My research focuses on medical image analysis. I aim to develop new techniques to measure relevant features from medical images (e.g. MRI/CT). These analysis techniques are characterized by clever modelling, incorporating prior knowledge and relying strongly on the physical principles of the image formation. If you like to solve a real clinical problem using fundamental theory than you are very welcome to join my group!
-
Diffusion MRI allows to determine the integrity of white matter tracts, based on diffusion measurements. This is highly relevant for study diseases such as ALS and Alzheimer’s disease.
-
Dynamic contrast enhanced MRI enables measurent of properties of the vasculature in patients. This is important for characterization of pathologies such as Crohn's disease and liver tumors. Previously students have worked on techniques to align imaging data from different patients, modalities and over time, to model diffusional processes and analyze differences between distinct patient groups. When you drop by, then I will inform you about our latest issues we have!
Magnetic Resonance Systems Lab (MARS)
At Mars Lab we work all day and night - mostly day - to advance quantitative Magnetic Resonance Imaging (MRI). We pursue two main veins of research. Firstly, we work on the development of novel biomarkers to enable non-invasive assessment of tissue integrity in greater detail and provide insights beyond conventional MRI contrasts. Secondly, we work on accelerated and motion-insensitive quantification of imaging parameters to facilitate clinical translation of quantitative MRI.
-
MRI is a very powerful and non-invasive technique for medical imaging. However, most patients get annoyed or even scared, only because MRI can be very loud during scans (louder than a jet engine!). The main source of this noise comes from vibrating gradient coils - they are affected by the Lorentz force as the currents in them are rapidly switched. In addition, there are other noise sources in the imaging room, such as helium pump and air handling equipment.
To understand the MRI acoustic noise better, in this project we will use spatial source separation to decompose the noise signal into its main components. This will enable studying the noise caused by the gradient currents directly, and eventually lead to developing quieter MRI.
-
Electrophysiological recordings are the backbone of neuroscience, enabling to listen to the electrical signals that constitute our brain activity at a single cell resolution. Novel hardware developments put ephys-recordings at a brink of revolution, enabling the simultaneous recording from hundreds of neurons rather than less than 10. These developments stir up neuroscience as they potentially enable the detailed understanding of neural circuits by listening to each individual neuron.