By Victorita Dolean
Figure 1. Principle of microwave imaging. Image courtesy of EMTensor.
Cerebrovascular accidents (CVAs) or strokes are caused by a perturbation in the blood supply of the brain, leading to a quick loss of cerebral functions that is very often lethal. There are two categories of CVAs: ischemic strokes (80% of cases), resulting from the occlusion of a cerebral artery, and hemorrhagic strokes (20% of cases), provoked by a bleeding vessel. From a medical point of view, the detection and characterization of CVAs are crucial for patient survival.
Continuous monitoring of the brain requires an image taken every fifteen minutes. Nowadays physicians use two imaging systems of the brain: magnetic resonance imaging (MRI) and cerebral tomogram (CT) scans. Even when these techniques are very precise, their use is not well adapted to efficient medical care. Moreover, they can be harmful, as in the case of continuous monitoring with CT scans that measure X-ray absorption by tissues.
Our research team, which won the Bull-Joseph Fourier Prize in 2015, carried out its work in collaboration with EMTensor, an Austrian innovative SME dedicated to biomedical imaging. For the first time ever, we have demonstrated on synthetic data the feasibility of a new imaging technique based on microwaves (see Figure 1). This technique allows for the characterization of CVAs, beginning with the very first instance of patient care in an ambulance and extending to continuous patient monitoring during hospitalization.
Figure 2. Measurement chamber (above) and corresponding mesh (below) for numerical simulation (diameter: 28.5 cm). Image courtesy of EMTensor.
How does it work? Electric properties of biological tissues are a great indicator of the tissues’ functional and pathological condition. Microwaves can image them, on the basis of differences in their dielectric properties. In such a system, a patient’s head is equipped with a helmet consisting of electromagnetic antennas that transmit data to a high-performance computing (HPC) center, which sends images of the brain to doctors at the hospital where the patient will be treated. This type of imaging requires a reduced data acquisition phase with a satisfying spatial resolution; it is less harmful than using a mobile phone. These characteristics make microwave imaging very appealing. From a computational point of view, microwave imaging solves an inverse problem and subsequently a fast solution of Maxwell equations. To prove the feasibility of such a technique, we have developed a HPC approach that generates brain images in less than 15 minutes.
In order to develop a robust and precise methodology for microwave imaging, one must master a few distinct research fields: optimization, inverse problems, approximation, and solution methods for the simulation of the direct problem modeled by Maxwell equations. The precise simulation of a direct problem for a complex and highly heterogeneous medium is a challenge in itself. We used a few tools previously developed by the team’s researchers: the HPDDM library for domain decomposition and its interface with the FreeFem++ software (finite elements).
EMTensor’s experimental system to be simulated consists of an electromagnetic reverberating chamber surrounded by five layers of 32 antennas each, able to work alternately as emitters or receptors (see Figure 2). The object to be reconstructed is introduced in the chamber. Alternately, each of the 160 antennas emits a signal at a fixed frequency, typically 1 GHz. The electromagnetic field propagates into the chamber, which allows the correct reconstruction (what we call imaging) of its dielectric properties. The other 159 antennas record the total field in the form of complex transmission, and the inversion algorithm reconstructs a brain image on the basis of this data. Our first step involved successfully comparing the measure of data acquisition made with EMTensor’s system with those numerically performed by the resolution of Maxwell equations on a 3D mesh.
Figure 3. Reconstruction time of an image regarding the number of computing cores. Mesh of the computational domain was generated by FreeFem++, software developed by Dolean’s research group.
In the next step, we created synthetic data on a brain model coming from scan sections (362x434x362 voxels) and then simulated a hemorrhagic CVA. Lastly, we designed and tested an inversion algorithm for monitoring the evolution of the CVA, reconstructed by successive slices. Here, a slice corresponds to one layer of 32 antennas equipping the experimental system. The use of parallelism allows the reconstruction of each layer to be generated independently, and the inversion algorithm uses 4,096 computing cores to reconstruct an image in less than two minutes (94 seconds). Figure 3 depicts this reconstruction. The restitution time, which can be further refined, already fits the physicians’ objective to receive an image every fifteen minutes to efficiently monitor the patient.
The medical and industrial challenge of this work cannot be emphasized enough. It is the first time that such a realistic study has demonstrated the feasibility of microwave imaging. Although the technique is less precise than MRI or CT scans, its low price, reduced size, and lack of adverse effects even with continuous use could make microwave imaging of the brain the equivalent of echography (ultrasound imaging) on other parts of the human body.
More details can be found in the preprint “Microwave Tomographic Imaging of Cerebrovascular Accidents by Using High-Performance Computing.”
French National Research Agency
Acknowledgments: This work has been supported in part by ANR1 through the project MEDIMAX (led by C. Pichot from LEAT of the University of Nice). Large-scale numerical simulations have been possible thanks to the technical support and computing hours on large supercomputers: Curie (CEA, Bull) and Turing (CNRS, IBM) via GENCI (allocations 2016-067519 and 2016-067730) or PRACE calls. Collaborators on HPC: F. Hecht, F. Nataf, P.H. Tournier (University Pierre and Marie Curie University, France), and P. Jolivet (University of Toulouse, France).
Victorita Dolean is currently a reader in the Department of Mathematics and Statistics at the University of Strathclyde, Glasgow, United Kingdom. She has co-authored around 50 research papers and conference proceedings and is part of the team that received the 2015 Bull-Joseph Fourier Prize, awarded yearly in France for important algorithmic advances in high performance computing.