10 April 2012
Affordable imaging technology could help to unlock the brain's secrets
Affordable imaging technology could help to unlock the brain's secrets
Medical imaging is one of the underrated miracles of modern medicine. Only a few decades ago, the field consisted of little more than X rays; today, electronic and other advances have created a whole series of techniques that have transformed our ability to see what is happening inside the body and, in particular, the brain.
Yet, despite these advances, we still cannot watch our brains at work in detail – assemblies of neurons actually firing. But with a relatively unsung imaging technique – Electrical Impedance Tomography (EIT) – we soon may be able to do just that.
What's more, EIT combines a series of benefits that make it unusually attractive in many applications. It is completely safe, easily portable and hence widely useable, unlike other brain imaging techniques such magnetic resonance imaging (MRI).
EIT produces images by capturing the internal electrical impedance of a subject using rings of ECG like electrodes on the skin. EIT systems typically comprise a box of electronics and a pc. The electrodes are placed in a ring or rings on the body part of interest (not exclusively the brain). Because everything can sit on a movable trolley, recordings can be made in a clinic or an out patient department.
A single impedance measurement forms the basis of the data used to build an image. For each measurement, a constant current, usually at tens of kHz, is applied to two electrodes and the resulting voltage is recorded at two others. This minimises the errors due to electrode impedance. The current applied, up to 5mA at 50kHz, is insensible and has no known ill effect.
Tomographic images are reconstructed from multiple impedance measurements made from different combinations of electrodes. Typically, up to 64 electrodes are used in a ring or rings and the several hundred impedance measurements needed to produce a single image are collected and reconstructed several times a second. In future more electrodes will be used – 128 or even 256 is possible.
The quantity measured is the electrical impedance, which comprises two components from resistance and capacitance. The resistance corresponds mainly to current passing through tissue by ionic conduction in salt solutions and therefore the extracellular space. The reactance is due to current flow through tissue capacitances and so is due mainly to current passage through interfaces which store charge, such as cell membranes.
A pioneer of the field is Professor David Holder, Professor of Biophysics and Clinical Neurophysiology at University College London. A neurologist by training, Prof Holder's work centres on using EIT for brain imaging.
There are two ways in which EIT can be used to image the brain. One is to see brain cells swelling, or blood volume, as is done already by MRI. While MRI produces better images, EIT wins out in many situations because it is far more portable – it can be used at the bedside or taken in an ambulance – and also cheaper.
The other approach is Prof Holder's main target, the use of EIT to image fast electrical activity in the brain. This is the Holy Grail of brain imaging – actually watching neurons at work.
"It would be like having an oscilloscope probe that you could put anywhere in the brain and look at circuits as they operated," he says. "It would be a true revolution in understanding how the brain works."
Today, his work is based on an animal model, using rats, with electrodes placed directly on the brain. The reason for the current excitement is that it looks possible that EIT can achieve imaging of fast electrical activity at various depths in the cortex, something that cannot be done with any other method.
In terms of hardware, UCL's current state of the art system is the KHU Mark 2.5, developed in conjunction with Kyung Hee University in Korea. This can record at multiple frequencies in the range 20Hz to 1MHz, has parallel data recording and can produce up to four images per second.
Currently, the best resolution that can be achieved with the state of the art hardware and existing algorithms is around 10% of the image diameter – which in the case of the brain means about 1 or 2cm.
Prof Holder notes: "With more electrodes and improved software methods, you could, in theory, get it down to a few millimetres, which would produce be extraordinarily detailed images."
As Prof Holder says, EIT works but there are still challenges to overcome.
"With animals and electrodes directly attached to the brain, good imaging is possible, as its performance with lungs shows (see below). For the brain, the problem is one of signal to noise and the fact that imaging the human brain must be done with electrodes on the skull, which dilutes the signal.
"There are a number of potential errors that can occur. You need an accurate mathematical model, we can't be sure the electrodes are in the right place and errors can occur in the performance of the electronics. We are improving on several different fronts and if we achieve a fairly modest improvement with each of them, we hope the combination will create accurate images."
There are several electronic challenges involved: designing circuits to give as high an output impedance as possible and reducing noise and crosstalk. A major problem is stray capacitance in the electrode leads, which can distort signals. To overcome this, a feedback circuit is needed to compensate for this effect.
One potential development is to use active electrodes. If you can shift the electronics to the electrodes themselves, this could reduce stray capacitance significantly.
"Unfortunately, this is technically difficult to do, particularly from a power angle – if you have gone to all this trouble to eliminate wires, then have to have cables carrying power, it can be self defeating," Prof Holder says.
Another major development in the EIT imaging of fast neural activity has happened at Manchester University, which has built a system it calls fEITER – functional EIT by Evoked Response. The team has used fEITER to create images of brain response to a visual stimulus, a 50ms flash of red light, with image capture occurring within 100ms.
The key differences between fEITER and conventional EIT is that the former achieves a combination of low noise and high speed.
"We are measuring data at 100frame/s, with 546 measurements being made every 10ms," says Hugh McCann, Professor of Industrial Tomography at Manchester University.
fEITER had to meet some stringent requirements. Building images from the 1% impedance changes in the brain caused by neural firings demands a measurement sensitivity of around 80dB, monitoring them needs the 100frame/s speed. It then has to comply with the stiff electrical and electronic requirements demanded by any device used in operating rooms or intensive care units. Results so far are promising.
"We are confident that we are seeing at least the blood response to neural processing, as MRI does," Prof McCann says. "It's harder to make the claim that we are seeing the fast electrical functioning on a time scale of, say, a tenth of second. We are trying to measure things no one else has done before, apart from experiments with cats, for example, using electrodes directly on the brain, which we obviously cannot do with humans."
Invaluable imaging technique
In fact, fEITER is certain to be invaluable for MRI type use as it is far more portable and cheaper. With that in mind, Prof McCann's team is working with stroke specialists at Hope Hospital in Salford.
Technically, Prof McCann says two things have enabled fEITER. One is the technique of parallel voltage measurements between neighbouring electrodes, with fEITER using around 34 a/d converters to digitise voltage continuously.
"This gives us the speed needed," he says, "but that is no use if you can't control the noise levels. We have achieved this with an intricate marriage between digital and analogue technology, to provide very low noise current excitation, all under the control of a Xilinx Virtex-4 SX35 fpga.
"We have also developed an analogue, voltage driven current driver with very low noise and other properties that make it ideal for EIT. We also use the fpga for digital demodulation measurement of the voltages being applied."
To make progress, further reductions in noise levels are possible.
"In fact, we are running the system so far quite conservatively, for example at a single frequency of 10kHz," Prof McCann says, "and we intend to explore different frequencies, as some believe sensitivity will vary with frequency. Also, we are using only one third of the current level allowed under medical standards, so that could be increased."
The hope is that fEITER – and similar EIT systems – will have wide ranging medical applications, from diagnosing and monitoring brain ailments like strokes, Alzheimer's and Parkinson's disease, to observing the brain during anaesthesia.
fEITER's development was partly prompted by Prof McCann's department being asked by the Manchester Medical School if fEITER could be used to show what is happening during anaesthesia in patients on the operating table. Last year, Brian Pollard, a professor of anaesthesia at Manchester Medical School, did just that for the first time, creating a 3d film of a patient's brain reacting to an anaesthetic.
EIT is by no means restricted to imaging the brain. Indeed, Prof Holder says one of the most exciting advances is in imaging lung function.
"Having been a purely academic subject for 20 years, some of the major medical device companies have became interested in the last few years and, for the first time, it is now being marketed by Draeger Medical for imaging lung function. A major benefit is that you can adjust ventilator settings based on the images."
When patients are on a ventilator, accurate settings are vital to ensure the lungs are kept open so they do not collapse, but not too much, which could cause damage.
"Although a drawback to EIT is that images are fuzzy, if they are good enough for a particular clinical purpose, its portability and safety make it unique. There has been no way of doing this before – you can't put someone in a CT scanner while you are ventilating them for days."
According to Draeger, the PulmoVista 500 'makes it possible to visualise regional air distribution within the lungs, in a way that is non invasive, in real time and directly at the bedside'.
"Using EIT for lung function monitoring has become established only relatively recently, but is now seen as an extremely valuable clinical tool," Prof McCann concludes. "I hear clinicians at conferences saying they are saving lives using EIT, in particular by helping them determine the settings of ventilators."