Scientific challenges
Non-invasive brain imaging is difficult because the scull is hard to penetrate and body substances are absorbing most signals. Magnetic Resonance Imaging is the most accurate brain-imaging mode, but is expensive, large and limited in temporal resolution since creation of each image is time consuming. Less accurate methods have been explored using EEG probes or different light sources giving limited image quality. Current brain imaging technology is confined to hospital use.
Major scientific challenges of the BIR project are:
- High-resolution short-range radar imaging
- Coupled antenna design for body coupling
- RF design of pulse radar integration with antenna
- Calibration and precise timing for multi-static radar imaging
- Signal conditioning and compression for efficient data transfer
- Multi-static instrumentation design
Background
The emergence of single chip pulsed radar systems operating at microwave frequencies enable a new generation of electromagnetic sensors with interesting penetration abilities. Radio waves (RF) have different penetration properties compared to light (our mobile phone are working pretty well in a dark room) and may be used to “illuminate” otherwise invisible matters. This property is both good and bad; we are able to discover hidden stuff, but controlling the RF waves like we focus light is impossible. We do not have “RF-lenses” like in cameras. We are able to direct the RF energy somewhat by using antennas, but far from the quality feasible with light. Additional challenges occur when trying to do imaging with RF. Instead of nice and easy integration of a focused image on a photosensitive surface, RF devices need to sense backscattered RF energy in time. Time-of-flight (ToF) is used to estimate the distance to the backscattering object. Acknowledging the penetration speed of RF waves is close to the speed of light, accurate ToF measurements are really challenging.
Assuming clock frequency of digital processing systems has saturated around 3GHz we may achieve a sampling resolution of approximately 10cm for measuring ToF. For brain imaging this is not good enough. However, the potential speed of CMOS devices is much higher, typically in the order of 300GHz. By innovative design [ref to CTBV] based on research from Univ. of Oslo, Novelda AS [www.novelda.com] has commercialized a single chip pulsed radar with close to 40GHz sampling rate. The single-bit sampling architecture combined with high pulse repetition frequency (PRF) good dynamics and high sensitivity is achieved. Depth resolution would be in the order of sub centimeter while empirical measurement with the radar has confirmed 4 millimeters resolution from single reflectors in air (good SNR). Estimating an average relative permittivity in the human body of 30, the penetration speed is reduced by a factor of 5. Reduced penetration speed is traded in as increased resolution approaching 1 millimeter for strong dielectric differences.
Proposed System implementation
Based on these estimates it may be feasible to build a brain impedance imaging system based on the Novelda pulsed radar technology. As indicated in Figure 1 the first configuration might be a circular multi-static “radar-band” with radar chips mounted on a flexible PCB giving a 2D brain image. The PCB is designed with suitable Rx and Tx antennas enabling short-range operation. The antennas must be designed for coupled energy transfer and covered with suitable foam. In addition suitable wiring and footprint for a microcontroller is required. The microcontroller should control the radar chips and add connectivity for data transfer (wired or wireless). The radar band for head mounting is light-weighted and may be battery powered.
Operation:
The imaging system may be operated in different modes. but the single transmitter mode is probably suitable:
Operate a single radar transmitter at a time with the rest of the radar as receivers. Circulate the Tx and collect data for each configuration. The Novelda radar collect 256 different depths (sample reuse) and based on the known positions of the radars, scattered energy seen from different directions may be combined to make a 2D impedance image of the brain (delay-and-sum). Based on the location of the receiver in the ring, both transmissive (tomographic) and reflective energy is received. Due to the high temporal precision, calibration of the system is required. The system my also be extended for 3D brain-imaging by adding more radars in a “hat” arrangement.
Requirements/qualifications:
The Faculty of Mathematics and Natural Sciences has a strategic ambition of being a leading research faculty. Candidates for these fellowships will be selected in accordance with this, and expected to be in the upper segment of their class with respect to academic credentials.
Applicants must hold a Master’s degree or equivalent in RF/electrical engineering preferably in combination with image analysis and signal processing.
Required skills for this position is RF wideband antenna design. Background in EM propagation in body tissue is an advantage. Skills in all aspect of radar instrumentation are preferred as well as instrumentation and software development for microcontrollers. Signal processing and image analysis qualifications are also preferable as well a general system engineering skills. System characterization and all aspects of using anechoic chamber is an advantage. The fellowship is a part of the MEDIMA project and is competing with other MEDIMA activities where the best qualified candidate will be selected.