Skip to main content

volumetric medical x-ray imaging at extremely low dose

Periodic Reporting for period 3 - VOXEL (volumetric medical x-ray imaging at extremely low dose)

Periodo di rendicontazione: 2017-12-01 al 2019-05-31

The ultimate goal of VOXEL is to provide an alternative to tomography with a disruptive technology enabling 3D X-ray imaging at a very low dose, by transposing Light Field Imaging technology to the X-ray regime, achieving 3D X-ray images while decreasing the dose by a very limited number of views (<10) for disambiguation. VOXEL aims at prototyping new cameras that will combine the X-ray penetration and nanometre spatial resolution, easiness to use, afforded by avoiding the rotation of the source or the sample, and extremely low dose for maximum impact on medicine and biology. VOXEL integrates the trans-disciplinary fields in medical imaging, optics, X-ray physics, applied mathematics and value to society through foreseeable commercialization.
VOXEL aims at making two prototypes. The first works at moderate X-ray energies, in the so-called “water window” range where contrast naturally occurs. This microscope will allow imaging cells in 3D in real-time. Later, a hard X-ray 3D camera for biomedical applications will be prototyped.
The plenoptic camera design, whatever the wavelength, relies heavily on an interplay between the target specifications defined by the biophysicists, the mathematical advances that are necessary in order to adapt tomographic reconstruction to tomographic data, and the engineering of the camera itself, starting with the metrology of soft and hard x-ray optics to be used.
Many developments made in parallel, during this last reporting period, have contributed to this achievement. As advised by the reviewers in RP2, we “focus[ed] on the planned core developments to make sure that the project goals are reached”, and in RP3 we finalized all the testbeds, software developments and prototypes aligned with the main goals of the Action.
By pursuing these objectives, VOXEL partners have successfully demonstrated X-ray plenoptic imaging at DESY, and are also currently operating two prototypes and two demonstrators for plenoptic X-ray cameras.
Work on the prototypes was guided throughout the project by research on plenoptic algorithms (as foreseen in DoA, “Guide the design through modeling”).
In the final RP, work on the software progressed to the point in which algorithms are now being explored for the analysis of plenoptic images. Progress in the software allowed obtaining 3D information of images obtained with the FleXray scanner, and is being used to analyze images obtained at the DESY demonstration campaign. We now have 2 open-source algorithms to process plenoptic images
The work performed on the first year of the project laid out the foundations for these three main pillars: soft and hard x-ray optics design, identification of the numerical framework for the special case of x-ray plenoptic imaging, and integration of both worlds in a coherent design.
In order to further these goals, we have designed and built the metrology stations where this data will be acquired. In parallel, we have installed small demonstration setups to check some particular technical aspects in the imaging process. In the course of this work, we developed a new wavefront sensor, a new X-ray optic, a new 3D microscope and the first stereoscopic imaging from two coherent microscopy images in the soft X-rays. While doing so, we kept pushing the boundaries of phase contrast imaging in X-rays, a technique to be incorporated in VOXEL at a later stage in the project.
The main blocks of plenoptic reconstruction were put in place: a mathematical equivalence between plenoptic imaging and limited angle tomography was found (CWI); a first algorithm for treatment of plenoptic
photographic data was built and post-processed by tomographic techniques (IO/LOA/CWI); a full-field transport code was put in place in advance of phase contrast goals (UPM) and ray-trace code was implemented for
plenoptic imaging (UPM/IO); and training of the VOXEL team for nano-imaging goals with direct feedback from end-users (CNR) started at facilities such as synchrotrons.
We have made great advances towards the goals of the project, and to summarize the main highlights, we can identify:
1) Prototype 1: The plenoptic camera prototype in the “water window”
2) Prototype 2: The plenoptic camera prototype in the “hard X-rays” (for small biological sample imaging)
3) Demonstrator 1: A plenoptic system working with coherent soft X-rays is being tested at IST.
4) Demonstrator 2: An emulation of the plenoptic system at hard X-rays has been recorded at CWI
We now have also 2 open source algorithms to process plenoptic images
5) Plenotomos at CWI – https://github.com/cicwi/plenoptomos). Plenoptomos is an open-source python package that enables advanced reconstruction of light-fields in a wide range of light wavelengths, from infra-red to X-rays.
6) EMcLAW simulation package, coupled with the AMReX library, is an open source code that allows to model the propagation of pulses over distances several orders of magnitude greater than their wavelength.
While we were advancing toward the major goal of building an integrated plenoptic x-ray camera, we pursued innovative technological solutions, increasing the potential impact of the project beyond the VOXEL camera itself.
The most visible result was the virtual unrolling of Herculaneum papyrus from Alessia Cedola’s team in CNR, an achievement highlighting the importance of non destructive testing. However, many more aspects can be highlighted already as an impact in society. Building an advanced X-ray metrology station in Portugal is impacting the University of Lisbon, from students to researchers, with adaptations to the curriculum towards education in advanced imaging techniques. Developing reconstruction techniques is helping us bring to the prototyping stage the 3D nanoscope at CEA and IO. LOA and IO have now a common platform to launch novel technological solutions, from which novel optical solutions are already being produced. UPM and CWI, thanks to the new VOXEL cluster, are progressing toward a computational solution to 3D reconstruction, from which any improvement will translate, in the future, to a lesser dose to patients.
Schematic drawing of the equivalence between the tomography and plenoptic geometries
3D image of a phytoplankton by the CEA-IO test microscope
Image of the X-ray light field emulation done in the FleX-ray lab at CWI
Virtual unrolling of Herculaneum papyrus using advanced tomographic techniques (Bukeeva et al, Sci R