Heart of the problem

Heart patient care could improve if research succeeds in boosting the quantity and quality of scan data. New techniques for streamlining magnetic resonance imaging and producing more useful information are to be studied so that the potential of scanning as a diagnostic tool is improved.

The London-based investigations will focus on making the most of the latest generation of scanners so that the problems experienced with many of the current models are eliminated.

The first problem is caused by a patient's natural desire to breathe during a scan. Movement blurs the picture and, as the heart sits on top of the diaphragm, patients are asked to hold their breath so that the heart stays reasonably still.

This is not always easy because patients may be anxious about entering the enclosing scanner, or their medical condition may make it difficult. Most have to be coached by the radiographers to hold their breath, which itself takes time, and even then they are usually asked to sustain it for only 20 seconds.

This means multiple scans are needed to acquire sufficient data and leads to further challenges because the patient is rarely able to stop breathing in exactly the same place during each respiration. The heart is then in slightly different positions in the series of 12 scans and the images have to be married carefully to maximise useful data.

Two-dimensional images

Another drawback of the current method is that all but the newest machines take 2D images so radiographers must be trained to plan the scans, making sure the right 'slices' of the heart are imaged to enable the cardiologists to see the area of concern. Planning the scan is a time-consuming process.

Fortunately scanners are now being developed that have up to 32 receiver channels, instead of the usual five or six. On these 'highly parallel' machines the coils are smaller, can obtain more localised information and can be run quicker. One benefit is that they can capture data in 3D. This, however, leads to another challenge - how can the large quantity of data be manipulated quickly to produce the best-quality images?

All of these challenges are to be studied by a team from

,

and

. It has received a £357,000 grant from the

to spend three years working out how to make the most of the newest scanners.

'We are supported by

which is providing help and expertise, along with a 32-channel scanner,' said principal investigator Dr David Atkinson at UCL's Centre for Medical Image Computing.

'We want to make it easier to do the scanning and to get the information you can't currently get, such as 7D flows.'

In essence, this is a 3D film of the heart that includes blood flow. The 7D refers to the three dimensions of space plus that of time and the three directions of velocity. The team also wants to characterise and image the walls of the blood vessels.

Computing cluster

'Fortunately at UCL we have access to a computing cluster, where there are 60 computer nodes, each of which is a 64-bit machine, so we can look at strategies for reconstructing these images from the data quickly and optimally,' said Atkinson. A scan can easily produce several hundreds of megabytes of data.

With the correct algorithms in place for data manipulation, the new scanners should allow patients to breathe normally during a scan. The solution will also free the radiographer from the burden of having to plan different scan routines for each patient. Instead, the procedure will become much simpler.

'This will involve complicated mathematics to determine the optimal balance between prior information and parallel imaging principles,' said Atkinson. When the best strategy is found it will speed up the whole scanning process so that more patients can benefit.

'The use of cardiac MR internationally has not been as great as one might have hoped, considering its potential as a non-ionising imaging technique,' said Atkinson.

He hopes the project may change that by helping it to become as simple as point and click, so many more cardiac patients can receive the care they need.