DigiScope could improve diagnosis of heart conditions

Because the DigiScope also works like a normal stethoscope, it will be especially useful for training purposes and for health professionals not familiar with chest medicine.

For trained physicians, the stethoscope still plays an important role in initial clinical assessment, quickly flagging up possible abnormalities for more detailed examination with, for example, electrocardiograms (ECGs).

However, auscultation (listening to internal sounds within the body via a stethoscope) is a hard skill to master. Heart sounds are of low frequency and the intervals between events are in the order of milliseconds, making it difficult to distinguish the subtle changes between a normal and a pathological heart sound.

A team of academics from Portugal led efforts to develop the DigiScope and called on sound engineers from Queen Mary University London (QMUL) to process the audio waveforms.

‘We usually do things on musical sound analysis but this sounded like an interesting project where we might be able to help, using some of the same sorts of ideas,’ QMUL’s Prof Mark Plumbley said. 

To process the sound waves, they used a technique called independent component analysis (ICA) to pick out the individual sounds from the four heart valves opening and shutting and line them up into a single output.

‘Normally we try to separate sounds that happen at the same time. So if you think of listening to a piece of music, you’ve got different instruments that are playing at the same time and you might want to separate those out. But the idea [is] that the heart sounds repeat and you can stitch those together — that’s the new bit to us,’ Plumbley said.

The output from the DigiScope can be transmitted wirelessly to a laptop or tablet computer where the sounds can be played back with the corresponding graphical visualisation. Thus, trainees can immediately review their assessments while more experienced GPs might look for confirmation of what they suspected.

‘If you really want something to be used, it’s got to be as unobtrusive and as close to what doctors normally do as possible if it’s going to be part of their everyday activity,’ Plumbley added.

The overall DigiScope initiative is a three-year, £120,000 project funded by Portugal’s Fundação para a Ciência e Tecnologia.

QMUL’s input has been undertaken as part of the five-year Machine Listening Using Sparse Representations project, which is receiving EPSRC funding of slightly more than £1.2m.