IR&D
Our independent research and development (IR&D) division aims to develop products and technologies to improve the current imaging capabilities of clinicians while minimizing the risk of harm to patients. Our research & development areas include:
Antenna Design
Antennas are the most important part of modern wireless communication systems. They have many applications including satellite systems, tactical communication systems, radars, wireless networks, base stations, cell phones, GPS, medical devices, imaging (biomedical imaging, security imaging), radio & TV broadcast systems, and etc.
Ellumen’s IR&D division specializes in antenna design mainly for biomedical microwave imaging applications, UWB antennas, base station antennas, reflect-array antennas, and more.
Biomedical Microwave Imaging
Microwave imaging is a branch of electromagnetic applications. The relatively long wavelengths of microwaves (from a millimeter to a meter) allow for penetration into many optically opaque mediums, such as living tissues. Microwave imaging is a non-ionizing and potentially low cost imaging modality with the aim of distinguishing between healthy and malignant tissues. Similar to other imaging modalities, it aids in the screening and detection of disease. Microwave imaging allows the internal structure of an object via the dielectric values to be seen by means of electromagnetic fields at microwave frequencies.
There are two primary approaches of microwave imaging: 1) pulsed radar and 2) tomography. The pulsed radar techniques seek to identify the presence and location of targets by their scattering signatures by analyzing the portion of the energy reflected back to the radar station. The most prominent pulsed-radar reconstruction methodology is confocal imaging which uses the principles of radar synthetic focusing.
In tomography, a microwave signal is transmitted by one antenna. Typically numerous antennas surround the object being imaged and receive the transmitted and/or reflected waves. However, this requires a large number of transmitting and receiving antennas surrounding the object of interest. This also poses a non-linear inverse scattering problem which is complex and very computationally intensive to solve. Excitation measurements are compared with calculations produced from numerical models to arrive at an update for the dielectric values. Different deterministic and stochastic algorithms serves as a basis for the non-linear reconstruction techniques. Among the stochastic approaches, evolutionary algorithms, offer a number of advantages including parallelism and no requirement for a differentiable objective function.
Our research is seeking innovations of microwave medical imaging to create a better understanding of both life and health.
Computational Electromagnetics
There have been several widely used numerical techniques for modeling computational electrodynamics over the past few decades. The finite-difference-time-domain method (FDTD) is one approach that covers a wide frequency range with a single simulation run, and treats disperse material properties in a natural way. Finite-difference-time-domain method and method of moment codes have been written for forward computational electromagnetics. Efforts have been taken to efficiently write the codes and parallelize them such as with a multi-core parallel approach on a high performance computer (HPC) or parallel approach with a graphics processing unit (GPU). 2D and 3D frequency dependent finite-difference-time-domain methods have been implemented as well.
Computational Intelligence
Computational intelligence approaches, such as artificial neural networks, have been explored to augment and improve non-linear microwave imaging reconstruction techniques. Different optimization methods have been implemented to train artificial neural networks and aid in microwave imaging reconstruction. The techniques were implemented to solve classification problems as a first step to serve as a baseline and the results were statistically compared. Various parametric and non-parameter tests and post hoc tests have been explored and used, along with their conditions for use, to allow for accurate statistical comparisons. This approach allows for algorithms best worthy of use and investigation in reconstruction techniques to be recognized.
High-Performance Computing
High Performance Computing most generally refers to the use of supercomputers and parallel processing techniques to solve complex science, engineering, and business problems that require low latency networking, high bandwidth, and very high compute capabilities. High Performance Computers have the ability to deliver sustained performance through the concurrent use of computing resources. The benefits of High Performance Computing are delivered through technology, focusing on developing parallel processing algorithms and systems by incorporating both parallel computational techniques and administration.