Home | Site Map | Watch | FAQ | History | Store | Contact

Using a Circular Graph for Periodic Data

One way to enhance the visualization of a heart sound is to use different means to express different attributes. In a simple waveform display (such as the one here), all the information is present, but it's all shown the same way: as wiggles in the line. This means that a viewer has to do a lot of interpretation to extract the various attributes. This could be done automatically in software. For example, here are some possible attributes:

These overlap, of course (for example, when the amplitude is zero, none of the other attributes can have meaningful values), so we have to be careful about what algorithms we use to distill out the various attributes, so as not to lose anything, or have one pollute another.

I'm not sure about the implications of frequency variations in these sounds; I'll have to learn more about that. I looked at some FFT's of various sounds, and the frequency spectrum information always seemed less interesting, less characteristic, than other attributes.

Amplitude and rhythm go together, obviously: rhythm is caused by variations in amplitude. It occurs to me that it's hard to see (and characterize) small differences in rhythm; for example, this:

Two pulses

looks pretty much like this
Two pulses a little further apart

and if you didn't have them side by side to compare, you might not notice. However, if we rearrange the period around a circle, the differences are clearer:

Two sets of pulses shown on circles