PhD student for the Nexis project (V36.3592)

PhD student for the Nexis project

PhD student to investigate image processing and machine learning algorithms for dual-layer cone-beam CT stroke detection.
Aanstellingsfunctie(s)
Promovendus
Faculteit(en)
Faculteit Electrical Engineering
FTE
1,0
Sluitdatum
09/01/2019
Aanvraagnummer
V36.3592

The NEXIS project for early stroke diagnosis

A photonics driven breakthrough in image quality and functionality of an interventional X-ray system will allow to directly diagnose stroke in the treatment suite. This will have a huge impact: enhanced work flow, reduced diagnosis & treatment time (up to 50% time reduction), which will save people lives and reduce healthcare costs. The NEXIS project (NExt generation X-ray Imaging System) aims to establish enhanced contrast Cone Beam CT imaging, while keeping high spatial resolution for 2D image guidance by an innovative spectral X-ray detector and related image processing (including deep learning). Two new key photonic components will be developed: 1) A thin-foil-based image sensor which has a (semi-) transparent TFT backplane, so that the photodiode array can receive light from both the top and the bottom side. This (semi-) transparency will be optimized, so that the image sensor can collect light effectively from top and bottom scintillator layers at the same time. 2) A 3D-printed pixelated CT-like scintillator with high spatial and temporal resolution to enable fast Cone Beam CT imaging without image artefacts. The usability and applicability of the new spectral NEXIS X-ray system for stroke imaging will be clinically validated in a European top hospital. The project brings together a multidisciplinary consortium, involving the full value chain (photonics R&D, medical system integrator, application owner, supply chain and equipment manufacturing). NEXIS will strengthen European competitiveness by developing a spectral Detector-on-Foil technology that meets the needs of the European and global X-ray image detectors market. NEXIS initiates the transition of standard (black & white) to spectral (color) X-ray detectors, which will improve performance and functionality of X-ray imaging systems.

Spectral image processing and machine learning

At the center of the system architecture, the raw spectral image signal needs to be processed such that it can be used for stroke accurate detection. The new dual-layer detector developed within the NEXIS project generates two sonograms at different energy levels. This introduces both challenges and opportunities for the associated signal processing. First, the image quality of these signals must be ensured by (automatically) finding and correcting possible imperfections and artifacts that arise from the novel detector. For this purpose, advanced signal/image processing and machine learning methods are investigated to localize the errors in time or spectral domain and correct them. The imaging process can be simulated using a Monte Carlo simulator and several numerical phantoms, which allows the generation of training data for the employed machine learning methods. The dual-energy signal also offers the opportunity to perform material separation based on the different response of the different energy levels, thereby significantly reducing the complexity of the analysis in the following stages of the processing pipeline. Furthermore, the additional information in the dual-energy sinogram may offer sufficient information to perform an initial analysis of the scan content with e.g. convolutional neural networks (prior to 3D reconstruction), which leverages the need for computationally expensive 3D volume processing.

Project consortium

A strong consortium is built that covers all the necessary technology areas to achieve the challenging ambition of the NEXIS project. The partners in the consortium have been carefully selected and together cover the complete value chain and expertise that is required to bring the NEXIS deliverables from development phase, to product development and product commercialization.

Partners in the project will develop collectively new photonic components (TNO, IMEC, 3DCeram) based on new materials, innovative foil technology and processes including the required X-ray detector manufacturing steps (Trixell). The newly developed spectral detector and related advanced image processing (TU/e, KU Leuven), including deep learning, will be integrated (Philips) in a Philips Cone Beam CT system. Karolinska Hospital in Stockholm, Sweden, will perform clinical evaluation.

The above project is a great project for a unique PhD trajectory with first-of-a-kind opportunities, both technical and due to the close intensive industrial cooperation! We are looking for talented, motivated, and enthusiastic candidates with an MSc degree in Electrical Engineering, Computer Science, Biomedical Engineering or Physics with a background in signal processing, medical imaging and/or computer vision. The candidate should match the following profile.

  • Experience in using Matlab and/or Python;
  • Solid understanding of (cone-beam) Computed Tomography (CT);
  • Solid knowledge of Image Processing techniques, e.g. noise reduction, enhancement, transformations, etc.
  • Excellent knowledge of image (signal) processing and a basic understanding of machine learning;
  • Strong communication skills, including excellent proficiency in English (spoken and written);

As this project involves strong collaboration with industrial partners, the candidate should be able to spend part of his time at Philips Healthcare in Best, which is close to Eindhoven.

  • Challenging job in a dynamic and ambitious university and a stimulating internationally renowned environment;
  • Full-time temporary appointment for 4 years with secured budget;
  • Gross salary between € 2.266,00 per month (first year) and € 2.897,00 per month (last year);
  • Additionally, 8% holiday and 8.3% end-of-year annual supplements;
  • An extensive package of fringe benefits (e.g. excellent technical infrastructure and excellent sports facilities);

Information

For  information  concerning  employment  conditions  you  can  contact  Mrs.  Tanja van Waterschoot, t.a.m.v.waterschoot@tue.nl .

For more information about the project, please contact Fons van der Sommen (f.v.d.sommen@tue.nl) and Peter H.N. de With (p.h.n.de.with@tue.nl).

More information on employment conditions can be found here:  http://www.tue.nl/en/university/working-at-tue/working-conditions/

Application

If you are interested in this position please use 'apply for this job'-button and send:

  • a cover letter explaining your motivation and suitability for the position;
  • a detailed Curriculum Vitae;
  • portfolio with relevant work (and a sample publications if any and when space allows);
  • contact information of two references;
  • copies of diplomas with course grades;

Please keep in mind; you can upload only 5 documents up to 2 MB each!