Micro detection

13 February 2014

the human skull is a complicated object to image due to its variable tissue densities

The human skull is a complicated object to image due to its variable tissue densities

stroke identification needs to be efficient and accurate to avoid treatment problems

Stroke identification needs to be efficient and accurate to avoid treatment problems

An anatomically realistic 2D head model with an embedded stroke region has been successfully imaged using microwave tomography by researchers in Germany. The nonlinear iterative Gauss-Newton algorithm has been applied to the scattered electric fields recorded at 24 points around the model and, through a blind reconstruction, the dielectric map of the numerical model and the stroke region are successfully reconstructed.

Stroke and related brain conditions are significant time-sensitive health concerns worldwide. While all strokes involve a disturbance of blood flow in the brain, the causes can vary, and this can affect treatment processes. Ischemic stroke is the consequence of a blockage, whereas hemorrhagic stroke is caused by bleeding somewhere inside the brain.

A suspected stroke is an emergency and given that the two major kinds of stroke share similar symptoms, but require totally different treatment, fast and reliable diagnosis within the first hour is vital. Typically, this diagnosis involves a neurological examination followed by imaging – either computed tomography (CT) or magnetic resonance imaging (MRI).  Although these current imaging techniques are efficient in terms of identification of stroke, they are slow, expensive, non-portable and limited to major hospitals.

Regular mapping

Microwave tomography provides a quantitative map of the dielectric properties of the object of interest, for example complex permittivity. It uses the fact that tissue affected by disease exhibits different dielectric properties from healthy tissue, so early breast cancer detection has been one of the main targets of interest for the researchers in this field and there has been significant progress in the development of 2D and 3D imaging systems.

To apply these techniques to strokes, the group reconstructed a dielectric map of an anatomically realistic head voxel model with an embedded stroke region using microwave tomography at 1 GHz. Starting from a homogeneous initial guess of the permittivity and conductivity, and without any a priori information about the shape of the layers, the head profile was successfully reconstructed after only 10 iterations and all the layers were clearly distinguished. Using a robust regularisation, it was even possible to realise the stroke region after the first iteration.

Algorithms to antennas

Malyhe Jalilvand, one of the authors of the research, explained that the group “has been expanding our research regarding this topic in several parallel directions, mainly focusing on the implementation of an efficient near-field imaging system.” For this technique, antennas are especially important as they play a key role on the efficiency of the imaging system. “So far, monopole antennas have been the main candidates for this application,” said Jalilvand, “but we are focusing on using more efficient planar/ non-planar antennas in our imaging systems. Furthermore,” she continued, “for the evaluation of our imaging system, we are also concentrating on building multilayer human-head mimicking phantoms.”

The UWB medical imaging group at the Karlsruhe Institute of Technology are working on both hardware and software.  “For example,” said Jalilvand, “the design of compact UWB antennas for biomedical imaging systems; implementation of a near-field probe-based measurement system for the characterisation of in/on-body antennas; implementation and characterisation of antenna arrays; sensitive imaging system for the early breast cancer detection; development of efficient application-based image reconstruction algorithms (mainly radar beam-forming approach and microwave tomography); development of realistic body-mimicking phantoms and matching liquids especially for breast cancer detection; stroke detection and the detection of water accumulation in human body.” Their long-term goal, she explained, is “the implementation of reliable portable microwave imaging systems that can be easily accessed in ambulances and small clinics to initiate the necessary treatment procedure with less delay.”

Overcoming scattering

The team’s microwave imaging technique has proved to be of practical clinical use.  However, unlike X-ray imaging – which only needs to consider the attenuation of the signals along pencil beams – the major difficulty regarding active microwave imaging is high scattering and diffraction of signals in human tissue. “In the last decade,” said Jalilvand, “microwave imaging has proved to have great potential to be exploited for medical applications. As a mature and cost-effective technology, microwave imaging eases the way for the implementation of efficient portable equipment that will assist in the fast and reliable diagnosis and treatment of diseases at a lower cost.”

Further reading

This article is based on the Letter: ‘Quantitative imaging of a numerically realistic human head model using microwave tomography’ (new window).

A PDF version (new window) of this feature article is also available.

Journal content

Cover of Electronics Letters, volume 50, issue 18

View all content

Browse or search all papers in the latest or past issues of Electronics Letters on the IET Digital Library.