Their projects are supported by various hospitals, the industry, the Danish Research Foundation, private foundations and DTU and comprise research within biomechanics and biomedicine, magnetic resonance imaging (MR), signal processing, and ultrasound.
Currently, our group includes 17 PhD students and though projects are individual and multi-disciplinary, we enjoy many opportunities to discuss ideas, approaches and results within the group.
The PhD students are encouraged to present their work to the faculty on a regular basis and to submit a minimum of two research papers a year to recognized journals. Participation in international conferences is also highly encouraged.
Current Phd projects
An Pham
In-situ Identification of Marine Organisms using High-Frequency Wideband Ultrasound
In his project, An is focusing on methods to identify properties like size and species of fish in the sea using ultrasound. These methods may improve the accuracy of the acoustic surveys that monitor the size and distribution of pelagic fish stocks.
Such information is important to ICES (The International Council for Exploration of the Sea) when they advice the European Union and European governments on fishing quotas.
To be completed: 2012
Mads Møller Pedersen
Vector flow with a clinical perspective
A new vector flow technique developed at CFU has been implemented in a commercial ultrasound scanner. Unlike experimental scanners built at CFU, this scanner can be used at hospitals.
With the new vector flow technique, it is possible to measure the movement of the blood, flowing in all directions, in contrast to conventional Doppler ultrasound. With vector flow, the movements along the surface of the ultrasound transducer can be detected and complex flow patterns become visible.
To be completed: 2011
Anetta Clausen
The main focus in 2009 has been on the production of signal molecules in quorum sensing bacterial cells. These are very important for the bacterial system since the amount produced works as an indicator of how many bacteria are present. If a threshold value is met the bacterial cells attack its host system. Thus if the amount of signal molecules can be limited the bacterial cells are harmless.
Two strategies have been followed in the last year. The decay of the signal molecule N-butanoyl-L-homoserine lactone (C4-HSL) has been monitored and the results published in an article and a poster as co-author. Furthermore, the spectroscopic Raman technique, Surface Enhanced Raman Scattering (SERS), has been applied to the detection of the signal molecule N-Dodecanoyl-DL-homoserine lactone (C12-HSL) with success.
Concentrations below nano Molar was observed. This was presented as a poster at the ASM Biofilm conference in Cancun, Mexico. The results are in preparation to be published.
To be completed: 2011
Jonas Henriksen
Prediction and detection of epileptic seizures
Of the world’s population, 1% suffers from epilepsy. Most of these patients can be treated out of their reoccurring seizures with existing medication or epilepsy surgery, but approximately 25% will continue to experience epileptic attacks. These unforeseen seizures are a cause of social stigma for the patients, and cause considerable demoralization, frustration and anxiety for themselves as well as family and friends.
Using the electrical signals from the brain, we look at the possibility of detecting, and perhaps even predicting, seizures. Instead of measuring at a hospital, we have developed a small portable device following the patient in daily life. The challenge is now to develop a dedicated automatic seizure detection algorithm, which uses a reduced number of electrodes compared to the hospitals’ systems, which often uses 16 or more.
In 2010 the first clinical trials with absence epilepsy will start. Design and test of an algorithm for this patient group will be followed by additional work related to other patient groups.
To be completed: 2012
Sandro Bottaro
Protein dynamics simulations at millisecond time scales
Many proteins are highly dynamic and adopt different states under physiological conditions. These conformational isomers hold the key to understand important biological processes (enzymatic catalysis, signal transduction, regulation and protein misfolding - Alzheimers, Parkinsons disease and type II diabetes). Standard computational tools result in inaccuracies of the force fields employed and sampling inefficiencies.
The overall aim of the project is to develop a general computational method to elucidate the molecular mechanism of conformational switches. At the heart of the project is the use of a novel probabilistic, kinetic method, developed in our group at DTU Elektro overcoming the deficiencies of current methodologies to protein dynamics. Our results will be corroborated against Nuclear Magnetic Resonance data, and possible discrepancies used to refine the force field.
The method will be applied to understand fundamental mechanisms in enzymatic catalysis, and to identify the conformational states that trigger the protein aggregation process associated with misfolding diseases.
To be completed: 2012
Jens Munk Hansen
Synthetic Aperture Compound Imaging
Medical ultrasound imaging is used for many purposes, e.g. for localizing and classifying cysts and tumors. A major drawback of ultrasound imaging is the low contrast due to the presence of speckle artifacts. A successful approach to remedy this is spatial compounding, where images are acquired from different directions and combined. The resulting images have a reduced speckle appearance, but at the cost of a reduced frame rate.
This project develops methods to perform spatial compounding using synthetic aperture data acquired with the experimental ultrasound scanner SARUS. It is investigated how to obtain a reduction in speckle appearance without a reduction in frame rate as well as how to increase the resolution using motion compensation. The methods are evaluated through phantom studies and in pre-clinical trials.
Synthetic aperture compounding should be able to obtain high contrast images allowing for an early detection of cysts and tumors.
To be completed: 2012
Ye Li
Synthetic aperture flow imaging using a dual beam former approach
Synthetic Aperture (SA) flow estimation has many advantages compared to conventional ultrasound imaging in as data can be acquired continuously for the full imaging region. The accuracy of estimates can therefore be very high and the frame rate can be increased because of the continuous data acquisition. It is also possible to find the velocity in all directions and map out complex flow phenomena.
SA flow imaging requires a large number of calculations to be performed. The aim of this project is to develop a computationally efficient method for synthetic aperture velocity estimation based on a dual beam former approach. The challenge is to lower the number of emissions and still maintain the quality sufficient for flow estimation. The approach will be developed and tested through simulations, phantom studies and an implementation for clinical tests on a commercial scanner.
Commercial implementation of a full SA ultrasound scanner capable of both anatomic and flow imaging, thus increasing the diagnostic value of ultrasound imaging by a higher accuracy and frame rate of velocity estimation systems.
To be completed: 2012
Michael Johannes Pihl
Medical ultrasound imaging is widely used for studying blood flow dynamics in the human circulatory system. However, blood velocity estimates using conventional techniques are angle dependent. This strongly limits the possibility of visualizing complicated flow patterns and obtaining the true velocity. This problem has been remedied by methods developed at Center for Fast Ultrasound Imaging (CFU) at DTU Elektro for 2D velocity estimation.
The aim of this project is to develop a method for full 3D vector velocity imaging to correctly show the velocity vector. Methods for acquiring data and estimating the velocity are developed based on 2D techniques already developed at CFU with main focus on the transverse oscillation method. The methods will then be implemented on the experimental ultrasound scanner SARUS at CFU and evaluated on flow phantoms and human volunteers.
With 3D vector flow imaging, complex flow patterns can be visualized and, potentially, pathological flow patterns around occlusions, valves, and bifurcations can be identified.
To be completed: 2012
Sara Matteoli
Diagnosis of heel pad injuries
Heel pad injuries, caused by fall accidents, marathon running, falanga torture etc., may cause the destruction of the heel pad intricate septation leading to permanent damage of its shock absorbency capability, and causing great pain for the patient, when trying to walk. Examinations are difficult to carry out, since there are a number of confounding factors influencing the result. One such factor is the natural variation in skin to heel-bone distance.
The objective is to obtain a quantitative evaluation of the heel pad tissue damage, and to gather information on the complete deformation of the heel pad tissues. A device based on an indentation experiment is to be developed. In order to achieve a description of the tissue damage at a microscopic level a computational simulation (FEM) of the heel pad is to be done. Such simulations will be validated by clinical studies.
Quantitative validated mechanical examination methods constitute pre-requisites for diagnosis and cure of heel pad diseases as well as medico-legal assessment for falanga torture.
To be completed: 2011
Isa Conradsen
Detection and surveillance of epileptic seizures with automatic multi-modal signal analysis
Today 25% of the patients suffering from epilepsy cannot get free of seizures. The unawareness of when the next seizure sets in makes the patient very insecure. As the seizures mostly appear without premonition, the patients are unable to bring themselves to safety. An alarm system able to detect epileptic seizures would therefore be of great help. Whenever a seizure sets in, the patient could immediately get help from family or hospital staff.
The objective is to design medical signal processing algorithms that are able to distinguish between epileptic seizures and normal movements. These algorithms should be based on data from Electromyography (EMG) and movement sensors containing 3D accelerometers, 3D gyroscopes and 3D magnetometers. The alarm system is thereby proposed to be a multi modal system making the seizure detection more reliable.
If the detection modalities respond with a high sensitivity and few false alarms, there will be a basis for developing an alarm system for the detection of epileptic seizures.
Jacob Kempfner
Early Diagnosis of Neurodegenerative Diseases
Research has provided a unique insight into mechanisms, which regulate sleep and wakefulness, including the neurotransmitters.
Certain diseases related to sleep express the early manifestations of other more serious diseases and establish a window for understanding fundamental disease processes. REM sleep behavior syndrome is an early manifestation of developing severe neurodegenerative diseases, such as Parkinsonism and dementia.
In Parkinsonism and similar diseases, there is a destruction of the brain's basic structures years before actual disease manifestations are evident. Once the disease is manifested, the possibility of actual treatment is limited. Early disease recognition is important as it allows an early intervention.
This project addresses some of the earliest disease manifestations in the brain stem and hypothalamus focusing on Parkinsonism and dementia.
The project's aim is to achieve better diagnosis and early disease insights into the mechanisms leading to destruction of the brain's basic functions.
Design of methods for automatic signal description and signal interpretation of the micro-and macro-structural changes occurring in waking and sleep electroencephalogram will form a basis for increased understanding and will provide the opportunity for earlier intervention.
To be completed: 2013
Joachim Rasmussen
The non-linear behavior of human tissue causes the speed of sound in an ultrasound wave to vary. Thus, a sinusoidal waveform gets distorted and changed into a sawtooth waveform consisting of higher order harmonics. These harmonics can be used for high contrast images. In synthetic aperture imaging, sound is emitted from the ultrasound transducer in all directions at once enabling improved contrast, resolution, and frame rate for e.g. fast 3D imaging.
The objective is to create an improved ultrasound imaging method by combining synthetic aperture imaging with non-linear ultrasound imaging. This method will be implemented on the Synthetic Aperture Real Time Ultrasound Scanner (SARUS) and used for pre-clinical trials.
The perspective is to provide high resolution, high contrast images without losing frame rate by exploiting the advantages that lie in non-linear and synthetic aperture imaging.
To be completed: 2013
Pengfei Tian
Computational protein aggregation
Many human diseases with a high social impact are characterized by the presence of highly structured protein deposits (amyloid fibrils) in the involved tissues, and in several fatal diseases the deposits are found in the brain, e.g. in Parkinsonism, Alzheimer, cystic fibrosis, diabetes Type II, and spongiform encephalopathies.
The project focuses on developing efficient computational and statistical methods to simulate and assess the process of protein aggregation. The aim is to establish detailed physical models of the fibrillation process and mechanisms based on recent theoretical and computational developments carried out within our group and collaborators.
Simulation results from the project will be tested against data. In particular, the relation between structural mechanical and kinetic properties of the various steps in amolyid formation will be elucidated by analysis of data from confocal microscopy, optical tweezers, atomic force microscopy, and quartz crystal microbalance techniques. The results of our computational work will be corrobated against these experimental data.
We believe that these methodological advances will make it possible to elucidate the protein aggregation process at a molecular level. A successful outcome of this project will pave way for rational strategies for developing new drugs to combat these protein misfolding diseases.
To be completed: 2013
Gertrud Laura Sørensen
Characterization of early and mature biomarkers for Alzheimer’s and Parkinson’s patients
Alzheimer’s disease (AD) and Parkinson’s disease (PD) are highly debilitating diseases with serious detrimental impact on work, social-, and family life. The disorders cause significant morbidity and mortality and consequently significant direct and indirect costs. These disorders will affect more than 20% of all adults. With the changing demographic composition of Western populations, the prevalence of these disorders is expected to rise. The current treatment of PD includes symptom modifying drugs such as dopaminergic treatment; however these agents do not alter the underlying disease progression. In AD there is treatment available, which delays/reduces disease progression, but disease severity is not improved.
The aim of the project is to use brainwaves (EEG) and associated modalities to characterize patients with PD and AD by identification of biomarkers at early and mature stages of disease progression. EEG recordings and appropriate biomedical signal analysis algorithms present an attractive measure to determine normal and abnormal patterns in the neurophysiologic activity in patients.
The characterization of biomarkers will serve as scientific basis for future studies of correlates between specific patient medication and potential efficacious treatment regimes. It is expected to yield earlier diagnosis and better treatment.
To be completed: 2014
Peter Møller Hansen
Pre-clinical studies of synthetic aperture
The purpose of this research project is to do pre-clinical studies of new digital techniques to improve image quality and and frequency in ultrasound scanning.
The project is a joint venture between the DTU, the Department of Radiology at Copenhagen University Hospital and the private company BK Medical LLC.
More specifically, Peter will study the synthetic aperture technique in combination with three individual techniques to improve image quality in ultrasound scanning of organs. The pictures will be evaluated in terms of resolution, contrast, depth, noise, frame-rate and artefacts.
Other pre-clinical studies will investigate two techniques for improving the visualization of blood flow and synthetic aperture sequentiel beamforming for 3D imaging.
To be completed: 2014
Morten Fischer Rasmussen
Two Stage Beam Forming Methods for 3D Imaging
In conventional ultrasound imaging, sound is emitted in one direction at a time. That implies sequential image acquisition limited by the speed of sound.
This makes it very difficult to acquire 3D images fast and makes it impossible to acquire a full volume of data for blood flow in the heart. A radical break within this area has been synthetic aperture imaging (SAI), where ultrasound is emitted in all directions and the volume of data is reconstructed through focusing afterwards. SAI makes it possible to increase frame rate, and the reconstruction also makes it possible to significantly improve focusing and contrast.
The aim of this project is first to show that SA 3D imaging can be done, and to compare it with conventional ultrasound imaging. However, because of the vast amount of calculations and data throughput, SAI has very high hardware requirements.
There is a variant of SAI for 2D imaging called two stage beamforming, which produces the image in two steps instead of one. Two stage beamforming lowers the amount of calculations and data throughput needed to produce the image, and at the same time maintains most of the advantages of SAI.
The end goal of this project is to utilise the 2D two stage beamforming technique and apply it to 3D imaging. Subsequently, it has to be investigated whether two stage beamforming is the optimal way of doing 3D imaging, or if there are other clever ways to image a 3D volume.
To be completed: 2014
Mette Funding La Cour
Micromachined integrated transducers for ultrasound imaging
Modern medical ultrasound imaging is made using multi-element piezoelectric transducers.
New technology is needed for advancing ultrasound technology. cMUT transducers uses fairly standard integrated circuit fabrication technology to make very small (10-20 micron size) ultrasound transducers, which are grouped to form larger elements. The technology is still immature and can be advanced, but the potential capabilities are large. It is here possible to fabricate any number of elements in any kind of configuration. 2D arrays with a very large array count can be made and there is also the potential to integrate switching capabilities on the die. The purpose of this project is to fabricate cMUT probes that are suitable for 2D and 3D ultrasound imaging using the Danchip clean room facility and use these with the facilities at Center for Fast Ultrasound imaging's (CFU) facilities for ultrasound imaging. CFU are experts in synthetic aperture ultrasound imaging and has the SARUS experimental ultrasound scanner, which can acquire and process data from 1024 transducer elements in real time. In synthetic aperture imaging it is possible to build up the image from several emissions and different elements can be measured from emissions to emissions. Combining all the data will significantly increase the image quality.
The purpose is to make cMUT probes with a large number of elements and with integrated multiplexing. These are then interfaced to the SARUS system for real-time operation and experimentation.
To be completed: 2014
Recently completed PhD projects
Marie Sand Enevoldsen
Computational simulation of stent-graft based minimally-invasive aortic aneurysm repair
Aortic aneurysms are abnormal dilations on the aorta arising from a weakening in the musculature of the vessel wall. The cause of development of an aortic aneurysm is not completely understood. Among the risk factors are age, male gender, smoking and hypertension. The mortality of ruptured aneurysms is 85%. Most aneurysms are treated by placing a stent-graft in the aneurysm, sealing off the fragile vessel wall to avoid rupture.
The objective is to develop a new finite element-based model of patient-specific aortic aneurysms for analysis and examination of critical parameters which are in evidence of the development and treatment of aneurysms. The influence from these critical parameters on the risk of rupture is of specific interest.
The research will facilitate the identification of patient-specific critical parameters related to rupture of aortic aneurysms, the prediction of rupture and simulation of the treatment.
To be completed: 2011
Martin Christian Hemmsen
Image processing in medical ultrasound
The requirement for new medical ultrasound image processing schemes is often stated as “not worse than…”. Comparisons are made visually, and they are largely based on experience, prejudice, lighting conditions, view angle etc. The improvement of medical ultrasound images relies on an understanding of the influencing factors. By quantizing these factors, new processing algorithms can be developed improving the diagnostic value of ultrasound imaging.
The objective is to achieve an understanding of the most significant factors influencing image quality in medical ultrasound and to develop a semi-quantitative measure of image quality based on a series of pre-clinical trials. Once such a measure is developed, new processing schemes should be developed to improve the diagnostic value.
In the project, a series of processing schemes is implemented on commercial and research scanners as part of the development of a semi-quantitative measure of image quality.
To be completed: 2011
Yigang Du
Non-linear ultrasound imaging is extensively used in clinics due to its improved image quality. It has been observed that the contrast is significantly better. However, currently there is still a limited understanding of the inner workings of non-linear imaging. It is very difficult to realistically simulate the non-linear propagation and often it is not understood why the non-linear images are better.
This project seeks to gain a detailed understanding of the non-linear propagation of ultrasound waves in the human body. The goal is sought accomplished through development of simulation software, measurements of ultrasound fields in water and on phantoms as well as pre-clinical trials to reveal the benefits of non-linear ultrasound imaging.
A clear understanding of non-linear ultrasound makes it possible to optimize the image quality, which can improve clinical diagnosis.
To be completed: 2011
David Bæk
Calibrated modeling of ultrasonic fields using Field II
The Field II program simulates ultrasound’s propagation in a linear medium. It uses the concept of spatial impulse responses to accurately model emitted, scattered and received fields in ultrasound imaging for transducer surface geometry, focusing, apodization and medium attenuation. It is employed for simulating point spread functions, conventional anatomic and flow imaging as well as new methods for vector flow imaging.
The project aims at using Field II to carry out a calibrated modeling of the ultrasonic fields in front of transducers. This includes combining the Field II software with accurate impulse responses of medical transducers, used in connection with imaging procedures, and comparing actual pressure measurements with simulations. Optimization of the calculation procedure of the spatial impulse responses from double curved transducer surfaces is another goal.
The project implements transducer models allowing you to predict the volt-to-surface acceleration impulse responses whereby simulation and measurements on the transducers will be in agreement.
Completed: 2010
Klaus Scheldrup Andersen
Non-invasive ambient pressure estimation using non-linear ultrasound contrast agents
Local blood pressure measurements provide important information on the health of the human body organs and can be used to diagnose severe heart, lung and kidney diseases. The pressure is currently measured locally in arteries and organs by a pressure catheter. As this is an invasive technique, it is inconvenient for the patient, there is a risk of infection, and the catheter will inevitably introduce changes to the blood flow and, thereby, the pressure.
The purpose is to investigate the possibilities of combining the physical interaction between ultrasound and highly compressible contrast agents for detecting ambient pressure changes. An experimental measurement setup has been designed and initial laboratory experiments carried out. A theoretical simulation study has investigated the mechanisms responsible for optimizing the sensitivity of the estimation yielding two clear and novel trends.
A non-invasive, reliable approach to measure the human blood pressure locally would provide doctors with a new and convenient tool to diagnose diseases related to the blood pressure.
Completed: 2009
Henrik Andresen
3D synthetic aperture imaging
Medical ultrasound has been used for decades as a non-invasive diagnostic tool for physicians. Ultrasound systems have moved from simple 2D systems showing a single slice to 3D systems allowing full organ visualisation. Conventional systems use a focused ultrasound beam to image a single well-defined line at a time, building the 3D volume one line at a time. This method is very time-consuming, limiting the system’s frame-rate.
To consider the limitation of focused transmissions, this project focuses on applying a new technique known as synthetic aperture focusing. This method has the potential of improving both the image quality as well as the frame-rate. The main focus is to improve the image quality of 3D volumes acquired by equipment which is also used for conventional 3D imaging.
The improvements will give a better visualisation of the organs allowing more accurate diagnoses. The image-quality and ability will be closer to e.g. X-ray CT, but without radiation.
Completed: 2009
Iben Kraglund Holfort
Ultrasound imaging is used for several diagnostic purposes and is perhaps most commonly known for fetus scannings. Ultrasound images are formed from a weighted sum of several measured signals. The signals are reflected from e.g. tissue interfaces within the body. The achievable resolution and contrast are directly dependent on the choice of weights. Conventionally, a set of pre-defined, data-independent weights is used.
The scope of the project is to investigate and refine adaptive beam forming methods used in e.g. radar systems and apply these to the field of ultrasound imaging. These adaptive methods utilize the actual data to find a set of optimized data-dependent weights. In this way, each point of the image is weighted differently. The adapted weights yield increased resolution and contrast of the ultrasound images.
The adaptive beam forming methods can increase the resolution and contrast of ultrasound images. The increased image quality may improve future diagnostic purposes.