Resumen de Advanced capabilities for planar x-ray systems
The past decades have seen a rapid evolution towards the use of digital detectors in radiology and a more flexible robotized movement of the system components, X-ray tube and detector. This evolution opened the possibility for incorporating advanced capabilities in these planar X-ray systems, and for providing new valuable diagnostic information compared to the previous technology. Some of the current challenges for radiography are to obtain more quantitative images and to reduce the inherent superposition of tissues because of the 2D nature of the technique.
Dual energy radiography, based on the acquisition of two images at different source voltages, enables a separate characterization of soft tissue and bone structures. Its benefits over conventional radiography have been proven in different applications, since it improves information content without adding significant extra acquisition time or radiation dose.
In a different direction, a really disruptive advance would be to obtain 3D imaging with systems designed just for planar images. The incorporation of tomographic capabilities into these systems would have to deal with the acquisition of a limited number of projections, with non-standard geometrical configurations.
This thesis presents original contributions in these two directions: dual energy radiography and 3D imaging with X-ray systems designed for planar imaging. The work is framed in a line of research of the Biomedical Imaging and Instrumentation Group from the Bioengineering and Aerospace Department of University Carlos III de Madrid working jointly with the University Hospital Gregorio Marañón, focused on the advance of radiology systems. This research line is carried out in collaboration with the group of Computer Architecture, Communications and Systems (ARCOS), from the same university, the Imaging Research Laboratory (IRL) of the University of Washington and the research center CREATIS, France. The research has a clear focus on technology transfer to the industry through the company Sedecal, a Spanish multinational among the 10 best world companies in the medical imaging field.
The first contribution of this thesis is a complete novel protocol to incorporate dual energy capabilities that enable quantitative planar studies. The proposal is based on the use of a preliminary calibration with a very simple and low-cost phantom formed by two parts that represent soft tissue and bone equivalent materials. This calibration is performed automatically with no strict placement requirements. Compared to current Dual-energy X-ray Absorptiometry (DXA) systems, 1) it provides real mass-thickness values directly, enabling quantitative planar studies instead of relative comparisons, and 2) it is based on an automatic preliminary calibration without the need of interaction of an experienced technician.
The second contribution is a novel protocol for the incorporation of tomographic capabilities into X-ray systems originally intended for planar imaging. For this purpose, we faced three main challenges.
First, the geometrical trajectory of equipment follows non-standard circular orbits, thus posing severe difficulties for reconstruction. To handle this, the proposed protocol comprises a new geometrical calibration procedure that estimates all the system parameters per-projection.
Second, the reconstruction of a limited number of projections from a reduced angular span leads to severe artifacts when using conventional reconstruction methods. To deal with these limited-view data, the protocol includes a novel advanced reconstruction method that incorporates the surface information of the sample, which can be extracted with a 3D light surface scanner. These data are introduced as an imposed constraint following the Split Bregman formulation. The restriction of the search space by exploiting the surface-based support becomes crucial for a complete recovery of the external contour of the sample and surroundings when the angular span is extremely reduced. The modular, efficient and flexible design followed for its implementation allows for the reconstruction of limited-view data with non-standard trajectories.
Third, the optimization of the acquisition protocols has not yet explored with these systems. This thesis includes a study of the optimum acquisition protocols that allowed us to identify the possibilities and limitations of these planar systems. Using the surface-constrained method, it is possible to reduce the total number of projections up to 33% and the angular span down to 60 degrees.
The contributions of this thesis open the way to provide depth and quantitative information very valuable for the improvement of radiological diagnosis. This could impact considerably the clinical practice, where conventional radiology is still the imaging modality most used, accounting for 80-90% of the total medical imaging exams. These advances open the possibility of new clinical applications in scenarios where 1) the reduction of the radiation dose is key, such as lung cancer screening or Pediatrics, according to the ALARA criteria (As Low As Reasonably Achievable), 2) a CT system is not usable due to movement limitations, such as during surgery or in an ICU and 3) where costs issues complicate the availability of CT systems, such as rural areas or underdeveloped countries.
The results of this thesis has a clear application in the industry, since it is part of a proof of concept of the new generation of planar X-ray systems that will be commercialized worldwide by the company SEDECAL (Madrid, Spain).
