Devices for Neurocontrol and Neurorehabilitation (DeNeCoR)

Call 2012-1 (Call 6) - ENIAC, 7^th Framework Programme for Research

Coordinated by PHILIPS MEDICAL SYSTEMS NEDERLAND BV, Netherlands
Other partners: Sapiens, Universitari Medisch Centrum Utrecht, Kempenhaeghe, Eindhoven University of Technology, ST Microelectronics, Politecnico di Torino, Università degli Studi di Firenze, Università degli Studi di Pavia (UniPV), Università degli Studi Roma Tre, Austrian Institute of Technology, Guger Technologies, Plessey Semiconductors, University of Sussex, Magstim, Institute for Micro-electronic Application, Brno University of Technology, LEITAT, Guger Technologies, Philips Forschungs Labor, Fraunhover - Elektronische Nano Systeme, MRComp, Polydiagnost

The primary aim of this project is to resolve this top issue such that electronic neuromodulation therapy becomes mainstream and full market potential can be obtained. Acceptance by neurologists will be increased by demonstrating designed co-existency between therapy devices and diagnostic systems.
The second aim is to extrapolate the ‘bilateral’ agreements between AIMD and MRI manufactures in IEC/ISO 10974 to other diagnostic systems and the interaction between diagnostic systems, like EEG and UltraSound (US), and non-invasive electronics based neuromodulation therapies (paving the way for image guided neuromodulation therapy). Also in this case technical challenges have to be solved. In particular DeNeCor targets:

• focused spatial localization with a Transcranial Magnetic Stimulator (TMS), which requires the design of a coil array with compact distributed power electronic modules and control electronics;
• arrays for neural sensing, invasive and non-invasive, based on new sensor arrays and packaging technology, with connection to modulation and rehabilitation devices;
• replacing piezo-electric US transducers by Capacitive Micromachined Ultrasound Transducers (CMUT) including 3D packaging, ASIC development and integration in an endoscopic system.

The third aim is to develop test methods such that the development of therapy devices and diagnostic systems can be decoupled which increases the efficiency of the demonstrator developments and enables technical evolution of all systems at their own pace. In addition test methods are currently missing from the Technical Specification IEC/ISO 10974 and are essential for its transformation in to an international safety standard.

In particular, UniPV has contributed to the definition of the specifications for the hardware design of the neurosurgical ultrasound probe with (1) a system-level simulation tool, purposely developed in Matlab, to support the design of the ultrasound probe transmission/reception front-end electronics and (2) by means of preliminary circuits simulations to investigate design trade-offs and performance metrics of the low-noise integrated receiver front-end.
Different amplifier alternatives have been investigated to implement low-noise integrated CMUT receivers. A first test chip has been designed with stand-alone amplifiers and with a complete single channel transceiver comprising a low-noise amplifier, T/R switch and high-voltage pulser. First experimental results proved successful operation with performances beyond state of the art. Simulation tools and software procedure have been developed.
A new test-chip has been designed and delivered to STI for fabrication where the front-end has been fully customized for the final demonstrator. In particular, the design has been scaled according to the lower equivalent capacitance of the CMUT in the final transducer matrix. Power dissipation has been reduced in order to fit constrain of less than 1.5W for the probe. A high performance class-AB buffer featuring high linearity and low power has been purposely developed to drive the cables connecting the active probe to the processing unit.
Matlab scripts have been developed to simulate the acoustic pressure field generated by the CMUT probe when exciting the transducer elements with differently shaped waveforms. An analysis of the transmission focusing delays has also been carried out for the dimensioning of the digital electronics for delays generation. The activation pattern of the pulsers over time has been simulated and analysed as well.
Concerning 3D imaging, studies have been performed about the treatment and filtering of noisy point clouds with the purpose of surface reconstruction. In particular, some interesting results have emerged about reversing the effect of signal convolution with unknown noise models. In addition, software procedures have been further developed for reconstruction and tracking of surfaces from time-varying spatial samples in input. Testing of these software procedures is currently under way, using existing images from medical datasets.
The procedures to be adopted for the demonstration of CMUT technology-based US probe for real-time imaging were analyzed thoroughly in the light of the most recent technical specifications for the upcoming US probe and a complete revision and update of these same procedures were performed. A new high-voltage (200V) pulser with for driving the CMUT transducer has been designed and tested experimentally. The latter allows programming the rise and fall times, pulse width and duty cycle. This flexibility was introduced to implement predistortion of the driving waveforms in order to compensate for the high 2nd harmonic emission of the CMUT. Experimental results demonstrated second harmonic acoustic emission 50dB below the fundamental component, allowing the use of CMUT for harmonic imaging.
Matlab tools have been developed to support the design of the front-end electronics of the CMUT probe ASIC. Considering the distribution of elements in the CMUT sparse-array, which is based on a Fermat’s spiral geometry, these tools implement ad-hoc developed algorithms to optimize the placement of the TX/RX channels and the routing of timing-critical signals in the ASIC. A beamforming algorithm, called Filtered-Delay Multiply and Sum, has been developed and tested for use in ultrasound B-mode imaging. The high performance of this beam former, in terms of image contrast and resolution, have been demonstrated, also when used in conjunction with high frame-rate imaging techniques. First tests on images acquired on the ex vivo bovine brain, used in the above mentioned experimental protocol, have also shown promising results.
The design of analog front-end for the ultrasound demonstrator in BCD technology has been completed. The front-end comprises a (1) low-noise amplifier that can be reconfigured to use capacitive or resistive feedback, allowing both low-noise and wide bandwidth operation, (2) class-AB buffers, with unity gain, to drive the probe cables, (3) a T/R switch with improved insertion loss, (4) a high voltage pulser with programmable voltage waveforms for reducing the harmonic emission.
An experimental protocol, that can be used to evaluate the performance of US probes based on CMUT technology in application to brain imaging, has been designed and validated in the practice. A full 2D/3D dataset of ultrasound images has been acquired using both a prototype CMUT probe and commercial US probes. Acquisitions were performed on an ex vivo bovine brain fixated in formalin and equipped with reference glass spheres for calibration. US images were acquired by means of the ULA-OP system and using an optical motion tracking system; this made it possible also to register US images with MR images of the bovine brain through ad-hoc developed software routines.
Software tools have been developed in both Matlab and Python for Paraview to support the processing of MR and US images together with positioning data to obtain two-way registration between the MR and US image spaces, to provide real-time feedback to operators during the acquisition of US images and to allow a visual, comparative analysis of images acquired in the two modes. These software tools are part of the experimental protocol above. Datasets with the CMUT probes have been recorded on an ex-vivo bovine brain. Furthermore, MRI scans have been made of the same bovine brain. A dedicated neuro-navigation setup has been designed in order to allow registration to the MRI dataset.

Duration: June 2013 - December 2016

Project website