Agenda

Signal Processing Seminar

• Thursday, 11 October 2018
• 13:30-14:30
• HB 17.150

manufacturing defect detection

Signal Processing Seminar

• Thursday, 4 October 2018
• 13:30-14:30
• HB 17.150

Machine learning in physical sciences

Machine learning (ML) is booming thanks to efforts promoted by Google. However, ML also has use in physical sciences. I start with a general overview of ML for supervised/unsupervised learning. Then I will focus on my applications of ML in array processing in seismology and ocean acoustics. This will include source localization using neural networks or graph processing. Final example is using ML-based tomography to obtain high-resolution subsurface geophysical structure in Long Beach, CA, from seismic noise recorded on a 5200-element array. This method exploits the dense sampling obtained by ambient noise processing on large arrays by learning a dictionary of local, or small-scale, geophysical features directly from the data.

Signal Processing Seminar

• Thursday, 27 September 2018
• 13:30-14:30
• HB 17.150

Signal processing algorithms for acoustic vector sensors

Symposium MRI for Low-Resource Setting

• Thursday, 27 September 2018
• 13:30-15:00
• Science Center, Mekelzaal 2, Mijnbouwstraat 120, 2628 RX Delft

Sustainable Low-Field MRI Technology for Point of Care Diagnostics in Low-Income Countries

MRIs are expensive and require sophisticated facilities to use them, which is why many people worldwide do not have access to this diagnostic service. Luckily an interdisciplinary team of scientists and medical professionals from the Netherlands (TU Delft and LUMC), the US (Penn State) and Uganda (MUST) is working on creating an affordable and simple MRI scanner. Want to know more? Come to the symposium!

In the talk, Steven Schiff of Penn State University (USA) and Johnes Obungoloch of Mbarara University (Uganda) will explain how their Ultra-low field MRI technology is a response to the need for appropriate medical technologies in countries like those in Sub-Saharan Africa. Together with TU Delft and LUMC they intend to use simple non-cooled magnets that can do the job by combining them with algorithms and a reference set of advanced MRI images. The ultimate goal is to create an MRI scanner that is inexpensive and that can be assembled, operated and maintained in developing countries. Their initial focus is hydrocephalus – the most common condition in children worldwide that requires brain imaging and neurosurgical treatment.

Microelectronics Colloquium

• Tuesday, 25 September 2018
• 03:00-05:00
• EWI, Snijderszaal

Tenure track colloquium

Wideband phased arrays for future wireless communication terminals, Daniele Cavallo (TS group)

Wireless data traffic will grow exponentially in the next years, due to the proliferation of user terminals and bandwidth-greedy applications. To address this demand, the next generations of mobile communication (5G and beyond) will have to shift the operation to higher frequencies, especially to millimetre-wave (mmWave) spectrum (30-300 GHz), that can provide extremely high-speed data links. To enable mm-wave wireless communication, mobile terminals such as smartphones will need phased arrays antennas, able to radiate or receive greater power in specific directions that can be dynamically steered electronically. However, to cover the different 5G mm-wave bands simultaneously (28, 39, 60 GHz, …) and to achieve total angular coverage, too many of such antennas should be on the same device: the main bottleneck is the insufficient space available to place all antenna modules. Therefore, I propose to investigate novel phased array antenna solutions with very large angular coverage and ultra-wide frequency bandwidth, to massively reduce the overall space occupation of handset antennas and overcome the current limitations of mobile terminal antenna development.

Towards smart organs-on-chip, Massimo Mastrangeli (ECTM Group)

Organs-on-chip are microfluidic systems that enable dynamic tissue co-cultures under physiologically realistic conditions. OOCs are helping innovating the drug screening process and gaining new fundamental insights in human physiology. In this talk, after a summary of my past research journey, I will describe how the ECTM group at TU Delft is envisioning the use microfabrication and materials science to embed real-time sensing and actuation in innovative and scalable OOC platforms.

Emerging electronic materials: from lab to fab, Sten Vollebregt (ECTM group)

Due to their nm-size features and often unique physical properties nanomaterials, like nanotubes and 2D materials, can potentially outperform classical materials or even provide functionality which cannot be obtained otherwise. Because of this, these nanomaterials hold many promises for the next generation of devices for sensing & communication and health & wellbeing.

Unfortunately, many promising applications of nanomaterials never reach sufficient maturity to be implemented in actual products. This is mostly because the interest in the academic community reduces once the initial properties have been demonstrated, while the risk for industrialization is still too high for most companies to start their own R&D activities. My goal is to bridge these two worlds by investigating the integration of novel nanomaterials in semiconductor technology and demonstrating the scalability of novel sensing devices. In this talk, I will give examples on how carbon nanotubes, graphene and other emerging nanomaterials can be used in the next generation of sensing devices.

Signal Processing Seminar

• Thursday, 20 September 2018
• 13:30-14:30
• HB 17.150

Tutorial on: Sum-of-squares Representation in Optimization and Applications in Signal Processing

MSc SS Thesis Presentation

• Thursday, 30 August 2018
• 10:00-10:45
• HB 01.230

Vessel Layer Separation of X-ray Angiographic Images using Deep Learning Methods

Percutaneous coronary intervention is a minimally-invasive procedure to treat coronary artery disease. In such procedures, X-ray angiography, a real-time imaging technique, is commonly used for image guidance to identify lesion sites and navigate catheters and guide-wires within coronary arteries. Due to the physical nature of X-ray imaging, photon energy undergoes absorption when penetrating tissues, rendering a 2D projection image of a 3D scene, in which semi-transparent structures overlap with each other. The overlapping structures make robust information processing of X-ray images challenging. To tackle this issue, layer separation techniques for X-ray images were proposed to separate those structures into different image layers based on structure appearance or motion information. These techniques have been proven effective for vessel enhancement in X-ray angiograms. However, layer separation approaches still suffer either from non-robust separation or long processing time, which prevent their application in clinics.

The purposes of this work are to investigate whether vessel layer separation from X-ray angiography images is possible via deep learning methods and further to what extent vessel layer separation can be achieved with deep learning methods.

To this end, several deep learning based methods were developed and evaluated to extract the vessel layer. In particular, all the proposed methods utilize a fully convolutional network (FCN) with two different architectures, which was trained by two different strategies: conventional losses and an adversarial loss.

The results of all the methods show good vessel layer separation on 42 clinical sequences. Compared to the previous state-of-the-art, the proposed methods have similar performance but runs much faster, which makes it a potential real-time clinical application. In addition, the proposed methods were assessed for low-contrast / low-dose scenarios with synthetic X-ray angiography data, and the results showed robust performance.

MSc SS Thesis Presentation

• Thursday, 30 August 2018
• 10:00-10:45
• HB 17.150

Automatic Interferer Selection for Binaural Beamforming

Spatial cues allow a listener to determine the direction sound is coming from. In addition, recognising spatially separated sound sources facilitate the listener to focus on specific sound sources. Because of this, preservation of spatial cues in multi-microphone hearing assistive devices is important to the listening experience and safety of the user. A number of linearly-constrained-minimum-variance-based methods exist for this purpose. Most of these are limited in the number of interfering sources for which they can preserve the spatial cues. In this thesis, a method of selecting the most important interfering point sources using convex optimisation is proposed.

The method is presented based on two different convex relaxations, which are compared, using simulation experiments, to existing, exhaustive search and randomised methods in terms of noise suppression and localisation errors. 

Both methods are shown to improve the performance of the joint binaural linearly constrained minimum variance beamformer, an existing method for simultaneous noise reduction and spatial cue preservation, by giving it more degrees of freedom for noise reduction and allowing it to handle a larger number of (virtual) sources present in the scene.

MSc SS Thesis Presentation

• Wednesday, 29 August 2018
• 13:00-13:45
• HB 17.150

Audio-visual authentication for mobile devices

Authentication is becoming an increasingly important application in the connected world and is driven by the growing use of mobile and IoT devices that use an increasing number of applications that require transactions of sensitive data. Security usually relies on passwords and/or two-factor authentication which are too intrusive for daily use. Biometric solutions such as fingerprints, voice, iris and retina are a good alternative to overcome previous problems.

In this project an audio-visual identity verification is presented, where the use of multiple modes that can already be captured from most IoT devices (microphone and camera) make authentication robust to adverse conditions. End-factor analysis (i-vectors) with cosine distance is implemented as the main classification algorithm which takes into account variations within and between speakers. Mel Frequencies Cepstrum Coefficients (MFCC) are used as audio features, 2D-DCT coefficients of a single snapshot and  Motion Vectors (MV) of the lips are extracted for visual features. Improvements combining different modes are shown using VidTimit dataset where the proposed algorithm achieves 0.7% of Half Total Error (HTER) in the test set outperforming single modes audio and visual by 9.5% and 6.4%, respectively.

MSc SS Thesis Presentation

• Wednesday, 29 August 2018
• 13:30-14:15
• LB01.010 Snijderszaal

Automatic Initialization for 3D Ultrasound CT Registration During Liver Tumor Ablations

Ablation is a medical procedure to treat liver cancer where a needle-like catheter has to be inserted into a tumor, which will then be heated or frozen to destroy the tumor tissue. To guide the catheter, Ultrasound(US) imaging is used which shows the catheter position in real time. However, some tumors are not visible on US images. To make these tumors visible, image fusion can be used between the inter-operative US image and a pre-operative contrast enhanced CT(CECT) scan, on which the tumors are visible. Several methods exist for tracking the motions of the US transducer relative to the CECT scan, but they all require a manual initialization or external tracking hardware to align the coordinate systems of both scans. In this thesis we present a technique for finding an initialization using only the image data. To achieve this, deep learning is used to segment liver vessels and the boundary of the liver in 3D US images. To find the rigid transformation parameters, the SaDE evolutionary algorithm was used to optimize the alignment between the blood vessels and the liver boundary between both scans.

MSc TC Thesis Presentation

• Friday, 24 August 2018
• 14:00-14:45
• HB 17.150

Phase Domain Ranging for Narrowband ISM Radio Bands

In this thesis, we study ranging algorithms in an indoor environment using narrow-band industrial, scientific, and medical (ISM) radio bands at 2.4 GHz. Previously, a phase difference approach has been implemented for this problem. However, the distance estimation is rather inaccurate for indoor ranging, mainly due to multipath and noise. This thesis studies several direction of arrival (DOA) techniques such as matched filter (MF), minimum variance distortionless response (MVDR), and multiple signal classification (MUSIC) to reduce the impact of indoor multipath. Forward-backward smoothing as well as the Akaike information criterion (AIC) and the minimum descriptive length (MDL) are also proposed to diminish the multipath effect further and estimate the number of separable multipath in the channel. Besides, a MUSIC-like method is discussed to prevent incorrect estimation of the number of sources. We test the proposed algorithm under different channel parameter values, compensate the bias, and show the related performance improvement as the absolute bias value is reduced up an order of magnitude.