Ultrasound non-destructive testing

We are looking for students who have a strong background in computational mathematics and/or (computational) physics, and an affinity with imaging and machine learning

Project description

Non-destructive testing (NDT) refers to techniques to analyze objects and materials without damaging them. It is for example used in testing evaluating welds of pipelines.

A particular setup that employs ultrasonic (US) waves is shown in figure. In this setup, a large volume of data is collected, from which a number of images (modes) can be produced. The various modes highlight different aspects of the structure, as shown in the figure below, and are used by inspectors to evaluate the weld.

Mathematically, the imaging procedure is an inverse problem; given (noisy) data we want to compute an image such that it fits the data through the non-linear forward operator. Through a series of approximations, the image is computed from the data by a linear imaging step.

The project aims to improve the accuracy and robustness of this imaging step. The projects are a collaboration between CWI, the Imaging Physics Department at TU Delft, and Applus RTD in Rotterdam.

Supervision & focus areas

Supervisor : Tristan van Leeuwen (CWI), Koen van Dongen (TU Delft)

Keywords : ultrasound imaging, inverse problems