This is about a device for monitoring EKG signals by "listening" to them. It describes something I put together in the 60's to help real-time monitoring of EKG signals by converting them to an audible signal. Perhaps you can use it for monitoring more important signals than heartbeats.
On February 15, 1960 I wrote a memo on the "Acoustic Analysis of Satellite Signatures." I was at RCA in Moorestown, working on the Ballistic Missile Early Warning System (BMEWS) which was concerned with (among other things) detecting and analyzing the radar returns from orbiting military satellites.
The radar returns from the satellites would appear as irregular waveforms on the face of an oscilloscope. These were called satellite "signatures." The idea was to look at the signatures to determine if the Russians had launched something new into orbit. There was a need for quick assessment of the signature, to decide if a more extensive quantitative analysis was required immediately.
In my memo I suggested that we might be able to make a quick qualitative (subjective) assessment of the signatures by treating them as though they were "speech sounds." The idea occurred to me after I saw a TV documentary in which different speech sounds were being displayed on an oscilloscope (aye, eee, eye, oh, you), and someone tried to recognize the sounds by looking at their waveforms on the oscilloscope. (I think it was a project to help deaf people "recognize" phonemes by seeing them.)
As I looked at the voice patterns on the screen, they seemed similar to the signatures of satellites. I immediately realized that I could never learn to recognize an "aye" from an "eye" sound, just by looking at those irregular waveforms. But my ears had no trouble at all! They were designed for recognizing time-varying patterns, while my eye was designed for recognizing space-varying patterns.
So I suggested that we try to "listen" to the satellite signatures which looked very much like sound patterns on the screen. Unfortunately, the satellite patterns were from very low frequency signals. If you tried to listen to them directly through a loudspeaker you wouldn't hear anything.
In order to hear them, it would be necessary to move the slowly changing satellite signals up into the audio frequency range. I proposed using the satellite radar return signal as the input to a Voltage Controlled Oscillator (VCO). A VCO provides an output signal whose frequency depends on the voltage at its input. In this way, the VCO would then provide an audible signal whose frequency varied with the variations in the satellite signature signal.
Unfortunately, there were no funds to proceed with the proposed approach.
On December 6, 1964 I wrote a memo to myself on "The Interpretation of Voltage Waveforms by Audio Techniques." Actually I prepared the memo to help convince a doctor at Children's Hospital in Philadelphia that the VCO technique might be useful in medical practice.
I was part of a group of engineers who volunteered to meet with doctors informally, to discuss how engineering methods might be of some help to the medical profession. During a tour of their facilities, a doctor described how he would look up at an EKG on the oscilloscope while doing heart surgery, to see if that stitch he just made caused any change in the heart pattern (signature).
I went home and made a Voltage Controlled Oscillator that would convert the changing EKG voltage patterns into changing audible patterns. I reasoned that if doctors could recognize the acoustic patterns in the heart signals, they wouldn't have to take their eyes away from the surgery they were performing.
I made a pitch to the doctors who then agreed to listen to my "Audioscope." (I intended it to be a marketable product that would make me rich!) In my pitch about the Audioscope I also made the point that if the sounds could properly identify the EKG patterns, it was possible to use another device to "automatically" analyze those sounds. (An electronic cochlea device had recently been developed at RCA to do such acoustical pattern recognition.)
During my demonstration we connected the Audioscope to an intern (a "volunteer patient") and listened to his EKG signature. As we listened we also watched the heart signals on the oscilloscope monitor. The doctors played with the knobs to get a good center frequency, and frequency deviation, for an audibly recognizable signal. We heard distinguishably different sounds that were characteristic for each of the different EKG leads. A female doctor reported better discrimination when the center frequency was raised higher.
While the group was discussing the experiment, there was a sudden change in the Audioscope sound output. Everyone looked at the intern to see if he had a heart attack! Nobody had been looking at the oscilloscope, so that sudden signal change would have otherwise gone unnoticed. Happily, there was no heart attack, just an artifact, as the intern brushed his arm against the leads.
That incident seemed important to me, since it showed the value of acoustic monitoring of a vital body function signature. Doctors could stitch up hearts without being distracted by continuously looking up to see if something changed in the EKG patterns.
After the demonstration we discussed the results. It was conceded that the Audioscope could help identify important changes in EKG patterns. But then the doctors made the following diagnosis:
1. If the audioscope provides the same information as the oscilloscope, then there is no real advantage in using it (except as an audio alarm).
2. If it provides "better discrimination" of signal features than the oscilloscope (i.e., the ear can hear things that the eye can't see), then maybe doctors will be trying to interpret meaningless changes.
3. The Audioscope might have value if it were used to collect audio recordings on large groups of children, and then submit those recordings for rapid automatic audio analysis at the hospital (using the putative electronic cochlea to make such a mass analysis cost-effective).
4. Unfortunately, since doctors spend many years learning how to interpret EKG patterns (from paper plots of EKG signals), they would not be enthusiastic about spending yet more time to learn how to interpret them by listening to them.
5. However, if the project should be carried further, they suggested that it would be best to start with automatic recognition of the T-wave.
Not long after the effort to get support from the doctors at Children's Hospital, I thought of trying a commercial approach.
I visited a well-known Philadelphia firm that made medical electronics equipment (along with its share of military electronics). They immediately appreciated the technical value of the Audioscope. However, their final evaluation was negative on the following basis:
1. It would take funds to design the product and manufacture it. They were short on funds.
2. In their experience, medical electronic equipment production was primarily "onesie, twosie." It seems that when doctors wanted equipment, they want it to be made with this or that special feature, while other doctors would be interested in it if it had a different set of features.
3. So you might be able to make one or two of the devices for a few doctors, but it would have to be made "different" to be marketable to one or two different doctors. On that basis, the cost of making medical devices was too high for them to consider the Audioscope.
I offer you the Audioscope concept as a project that you might want to develop further. It has technical merits and might even make you some money if you know how to work the medical market.
But, hey listen, the medical market is only one market. You can come up with other markets once you appreciate the basic value of using sounds to analyze time-varying patterns that would otherwise go unnoticed.
You should consider such applications as feedback devices for monitoring muscle activity during physical rehabilitation exercises. Here the patients can hear their muscle activity as they make efforts to produce certain movements. The sounds would guide them in making changes in their efforts to produce proper responses.
Similar feedback applications might apply to other important therapies whose results can be converted into slowly changing voltages that are then presented to the patient as slowly changing sounds to guide their further efforts.
How about the acoustic analysis of ...
[ Home Page ] [ Perspectives ] [ Projects ] [ Speculations ] [ Books ]
Comments or suggestions? mortgale@comcast.net
Copyright © Mort Gale, 2003 All rights reserved.