On Thursday 24 September 2020, Martin Lind Ommen will defend his PhD thesis "Micromachined2D transducers and phantoms for superresolution ultrasound imaging".
Time: 14:30
Place: Zoom, please register here: https://dtudk.zoom.us/meeting/register/u50ocO-vrzooGtDI4hi39g3Sdd2IdpBkQ1dj
Principal supervisor: Professor Erik V. Thomsen
Co-supervisor: Professor Jørgen A. Jensen
Co-supervisor: Professor Niels Bent Larsen
Examiners:
Associate Professor Emil Kromann, DTU Health Tech
Professor Alfred C. H. Yu, University of Waterloo
Senior Specialist Carsten Christensen, Grundfoss Holding
Chairperson at defence:
Associate Professor Matthias Bo Stuart
Abstract:
Super-resolution ultrasound imaging (SRUS) is a new ultrasound imaging technique which aims to visualise the smallest branches of the vascular system, namely sub-100 μm arterioles and venules and 5-9 μm capillaries. The technique breaks the conventional diffraction limited resolution through super-localisation of micro-bubble contrast agents injected into the vascular system. The method has been demonstrated to obtain resolutions of only a few tens of micrometres which should be compared to 500 μm of conventional diffraction limited ultrasound system. The goal of this project has been to develop tools to improve the SRUS techniques and transfer them from 2D to 3D imaging through development of capacitive micro-machined ultrasonic transducer (CMUT) fabrication processes and 3D printed phantom fabrication for improved validation.
A continuous goal in ultrasound transducer fabrication is to create larger transducer arrays for increased field of views (FOVs), combined with larger operating frequencies for increased resolution. The increased size of arrays means that even the smallest sample contamination might ruin the few devices available. The fabrication process optimisation presented in the thesis is about fusion bonding. Fusion bonding conducted directly in hand without a wafer bonder has been shown to provide a wafer bond of comparable quality to fusion bonding performed in dedicated wafer bonders. Handbonding allows for forming the bond directly after cleaning the wafers, minimizing the risk of particle contamination, therefore improving the processing yield.
To properly develop the SRUS techniques, suitable phantom structures need to be made. A stereolithography (SLA) 3D printing solution for fabrication of ultrasound phantoms is presented in the thesis. Conventional phantom fabrication methods consist of tubes suspended in water which can be perfused by micro-bubble-containing liquids. However, these methods are incapable of providing feature control on the scale required for SRUS and suffer from limited three-dimensional feature placement capabilities.
The printed structures are hydrogels, water-containing polymer networks, printed with a voxel size of 10.8 μm × 10.8 μm × 20 μm. The acoustic and structural properties, as well as potential ways to manipulate them are presented. The phantoms have an average speed of sound of 1577 m/s and an average density of 1.045 g/ml. The printed phantoms swell approximately 2.6% post printing, making compensation of design features necessary when using the phantoms as reference structures.
A new type of phantom was developed based on printed cavities which have been shown to reflect sound. By keeping the cavities smaller than the imaging wavelength, they can be used as stable point targets for repeated imaging. Design optimisation of the scatterers has been con- ducted in terms of actual printed size and reflected intensity, modelling different sizes, shapes and local overexposure schemes. The “Single pixel” scatterers, printed with a single voxel wide local overexposure around each cavity, yielded the highest reflected intensity.
Calibration of a 3D SRUS pipeline imaged with a row-column addressed (RCA) probe using a scatterer phantom containing eight randomly placed scatterers showed high accuracy of the pipeline. The localisation precision was found to be smaller than 27.6 μm in all directions, which is less than 1/18th of the imaging wavelength used in the experiment. The high precision allowed for detection of distortion in the beamforming on a micrometre scale. This would not have been possible to discover using conventional tube phantom setups.
A series of flow phantoms were created to perform well controlled SRUS experiments with micro-bubbles. A fiducial marker grid layout was presented, which allow for easy alignment of the ultrasound probe to the phantom features. A flow phantom was created to demonstrate super- localisation of micro-bubbles in 3D using a RCA array. The localisation precision was estimated using the flow phantom, evaluated based on the micro-bubble distributions in the flow channel, with the estimates being in line with the precision estimates based on the scatterer phantom. Finally, flow phantoms for demonstrating true resolution of the SRUS pipelines were developed, utilizing the 3D fabrication freedom of the 3D printing technique.
The results illustrate the great obtainable achievements with a high resolution 3D printing phantom fabrication method, but only scratches the surface of the potential solutions that the phantom printing method provides. The printing method allows for three-dimensional freedom of design and an unparalleled control of phantom feature placement and feature size control.