Commenting on a recent paper about limb positioning effect — Stuttaford SA, Dyson M, Nazarpour K, Dupan SS. Reducing Motor Variability Enhances Myoelectric Control Robustness Across Untrained Limb Positions. IEEE Transactions on Neural Systems and Rehabilitation Engineering. 2023 Dec 15. [link][1] (released under CC BY 4.0).

Authors write “The limb position effect is a multi-faceted problem, associated with decreased upper-limb prosthesis control acuity following a change in arm position. Factors contributing to this problem can arise from distinct environmental or physiological sources. Despite their differences in origin, the effect of each factor manifests similarly as increased input data variability. This variability can cause incorrect decoding of user intent.”

Nah.

I don’t think it has to be all that complicated.

• What their results mainly show is a flexion exacerbated involuntary control error ¹. They discuss this in a creative way but seem to miss the point ².
• As their methods show, they place electrodes over extensor carpi radialis and over flexor carpi radialis ³.
• As a glance into anatomy books show [link], extensor carpi radialis longus together with extensor carpi radialis brevis are known to produce wrist extension and abduction (radial deviation) but in addition, extensor carpi radialis longus may help to flex the elbow joint. So once the below elbow amputee extends their imaginary fingers or wrist, the myoelectric device opens as it is customary to wire the prosthesis that way – sheer conventionalism and convenience to do things like that. Once they close that hand and once they bend the elbow after that, the device opens again but now “involuntarily” – simply because the same muscle that opens the imaginary but amputated hand that has the electrode placed on it also, anatomically and normally, flexes the elbow. You can not “train” that really nor is it a primarily statistical problem, also because that is not a technically wrong signal. It is how anatomy works, simple as that. When my extensor carpi radialis is not already atrophied to flabby fat as most other forearm stump muscle tissue that I have there, it has one and only one reason: because it always helps flex the elbow for the simple reason that anatomy in most people provides for that. Basic textbook knowledge. As myoelectric prostheses require good signals, with high SNR (signal-to-noise ratio), the extensor carpi radialis is super popular among prosthetists or myoelectric device researchers, and as it is an extensor that us arm amputees formerly used to help open our fingers, one tends to link these electrodes to the “open” function of the myoelectric device because the muscle is so well built it gives a good signal. An arm amputee with a non-atrophied stump muscle should alert you, generally it should. So when I flex my elbow, hand opens? That is what the “limb positioning effect” is all about. The explanation is anatomy and what goes with it.

Figure 6 from source [1]. 

Logically, if one flexes elbow joint, anatomy is predominant reason why electrode over that extensor carpi radialis muscle also activates myoelectric hand.

A far more tricky and difficult question to these researchers may be why electrodes are not placed over the extensor carpi radialis muscle in above elbow amputees? ⁴ 

In practice, extensor carpi radialis is used to place myoelectric electrodes over it, because it is relatively large and thick and not atrophied and gives nice signal, SNR (signal-to-noise) wise. Conversely, the flexor carpi radialis (FCR) does not also activate elbow joint motion in a comparable way. Mine is clearly more stringy and atrophied, for super simple reasons (I am a below elbow amputee and the muscle doesn’t have daily tasks to perform). So its myoelectric signal is not that good, always requires a bit of a boost, less SNR, more jitter, normal. If there is a Mexican stand-off between my ECR and FCR, the ECR tends to win simply because the ECR as daily task always gets to help flex the elbow to stay in shape. Or look at it from that angle: assume you were these muscles. You, yourself. Now, explain your life. How do you do. What have you done today. With only a bit of empathy, you could put two and two together, or maybe one plus one.

So, if one wants to know why a below elbow amputee such as me has a large and thick extensor carpi radialis, then the answer is in the question: that muscle is well and trained and not atrophied only because it helps bend the elbow.

But it is certainly not in a good shape “because I open/close my hand/fingers”, or “because I move my wrist” so much – and why is that the case? The profane reason is that I do not have these any more. These were all amputated. If you paide attention at the title of this blog you could have inferred the non-atrophied ECR and the atrophied FCR at once as that is what goes with the (remaining) territory. So again, the only reason extensor carpi radialis is not atrophied in a below elbow amputee such as me is because we flex our elbows quite a bit. Next time you see someone with a really large well trained muscle you will know that there was elbow grease that went into that. “Elbow” grease, bwahaha. 

If you accept that this is interesting, then consider that it is really a twisted situation and won’t get better particularly if you think it through. The reason why myoelectrodes can create particularly useful signal is because the muscle is not atrophied entirely. If it is not atrophied in an amputee it clearly has a role beyond the amputated limb, so look proximally.

Myoelectric control thus requires muscles that are fit, so to say. Sure, the aficionados pride themselves in saying this stuff is not body powered – but those are people that also write such papers, right? The only muscles that are fit generally are those that are used every day, often, regularly, all the time. So there is a physiological reason why also a below elbow amputee has a well trained muscle formerly mainly used to open / abduct fingers/hand, i.e., the extensor carpi radialis: because it (also, and in a below elbow amputee, only ⁵) flexes the elbow. One cannot possibly be surprised that a muscle that flexes the elbow activates the myoelectric device upon elbow flexion when daily elbow flexion was the only underlying reason that muscle had not atrophied and therefore, exactly therefore, was chosen for myoelectric electrode placement at best signal site. Which is the characterizing element here.

But it does not stop there.

In order to maintain any muscle (volume, strength, myoelectrode SNR) – imagine even installing TMR (targeted muscle reinnervation) as surgical work in a below elbow amputee! –  also, ongoing further steady daily activity will be required.

In order for myoelectric signals to work, muscle is needed that is trained or built up and not just entirely replaced with fat.

And for that to occur, something is relevant that you certainly do not want to hear, leave alone from me: for that, constant training and constant activity of muscle, also TMR-related muscle, requires permanent availability of the prosthesis and regular if not daily use are the single defining key elements. Body, body, body, right?

For that muscle training to be constantly and not sporadically occurring, the whole prosthetic arm system (myoelectrodes, battery, motor, mechanical parts, grip, hand, glove cover) has to work reliably, comfortably and provide really good function. Otherwise, muscle is likely to atrophy, and then also its signal won’t be that good any more. That is the exact reason why I prefer a functioning and tolerable simple prosthesis over anything else: keeps me going. No matter whether you wear a “myoelectric” (i.e., body powered muscle activation leveraged electrode triggered) or “body powered” (direct body power to actuate the thing mechanically) prosthesis – you rest, you rust. Must come at a surprise, right.

And we all know that this constant prosthetic availability simply is not the case with the typical myoelectric prostheses: myoelectric parts are fragile, they are insanely expensive AF, we know that skin electrodes can cause electrochemical burns that may take weeks to heal [link], that myoelectric suspension can cause some interesting skin injuries just after 1 day of office work [link] and that glove covers easily fall part even before usage [link], and that grip angles for bionic hands are problematic [link] and center of gravity is off [link] all of which warrants non-use.

The only way that I can work with any prosthetic arm also to sustain a muscle is, if, from the get go, from hour zero, the prosthesis I am supposed to control via that muscle, the prosthetic device, the components and all, so, if (conditional if) that device works and sustains and if it is reliable [error rate < 0.03%] and if it is comfortable and and if it is also affordable, none of which really are the case for myoelectric prostheses. 

Then it is  also about skin in the game.

You will certainly not even give one iota about this, now that you read this. Who cares, right? Who gives a rat’s behind about elbow, muscle and myoelectric? You write papers or build parts wherever and you can afford to not care one bit. It shows. I read that paper and I know. Zero skin in the game, big surprise the device opens. Multifactorial, sure. So one researcher whose experimental myoelectric arm I was wearing in his lab said his thing was perfectly reliable. Perfectly! I asked him if I was safe to use it at home for anything? Even fragile or difficult? He said by all means. Then I picked up a water filled paper cup and asked him, can I hold this while hovering over your laptop? He went a bit pale and looked frightened and then said no. Later he said that he learned more about the reality of prosthetic arms in that second than in all the years before. If you want I can come to anyone of you and visit you, hold a water filled paper cup over any of all of y’all’s laptops if that is necessary to inject the feeling you may lack to understand wat exactly I am trying to say here. On the other hand, you can do that experiment with anyone living nearby. Just ask someone with a myoelectric arm over and have them do elbow flex test with water filled papercup over your expensive electronic device whatever it is. Or have them drop your 7000 USD camera for you, from 1.5m onto hard concrete floor, just to get the feeling of why I think reliability is a thing. Nope, you did not feel distressed at the above elbow amputee question above. Can you teach empathy to robots, maybe? i digress…

With any body powered or myoelectric (myo means “muscle”, you knew that, right), all control is leveraged “via the body” in some way. Muscle is body. Muscle power, body power. So to create a dichotomous label in an attempt to label “myoelectric” as somewhat more modern than (if not different from) a thing labeled as “body-powered” may certainly constitute an interesting attempt at gas lighting, but not at good engineering, which as we can see also (but not only) in this study, which seems to be published in a great journal without a lick of relevant anatomy consideration.

Further research certainly could be helpful, particularly if it is well informed.

One may investigate what forearm stump muscles or muscle parts are not activated by elbow motion, before trying to reduce limb positioning effect. During the myosite testing, how is whole body motion and elbow flexion or extension as having an impact on myoelectric device control best evaluated? As far as I remember, for any myosite evaluation, one sits at a table and the setup is quite experimental and impractical with short cables and sticky electrodes and so on. That way, they probably never find out what elbow flexion/extension does in terms of involuntary device opening/closing. One should also test how much signal in any individual is generated by elbow flexion or extension, and list this as a separate statistic to then see how that can be individually improved.

If one wants to shorten all this, one could say this about the extensor carpi radialis (ECR) muscle in below elbow amputees:

Essentially, for myoelectric control of a prosthetic hand, researchers rely on a particular muscle (ECR) in below elbow amputees, where it is activated/used/trained and kept in good shape almost exclusively by elbow flexion (and that only in non-amputees also controls wrist and hand/fingers), which causes “involuntary” hand control problems once the elbow is flexed, with researchers then wondering why that muscle is activated by elbow activity. They may want to try to better understand the reality of below elbow amputation, and anatomy.

The limb positioning effect thus may better be called “function of muscle ignored by researcher/operator” effect. Which, in essence, it is.

That is why, on myoelectric prosthesis, (tips corner, right?) I had my flexor and extensor electrodes reversed so extensor closes the myoelectric device upon elbow flexion (instead of dropping whatever I try to hold on to). Still sucks, but at least things do not drop just as often when I flex my elbow. Myoelectric control, in a real practical sense, is body-powered AF, if ever you needed a concept spill-over to help you understand better. That is why leveraging body power, trunk motion, stop / freeze maneuvers, to steer a heavy cumbersome and fragile insanely expensive uncomfortable type of technology, is essentially junk [link], particularly seeing as if I can alternatively just use a well built body power directly, in direct mechanical control. With a well built body-powered arm, I neither get harness issues, plexus compression, reach space restriction or any other control issues they try to tag on wrongly-built body-powered arms.

Of course I just may start out by apparently “just claiming” that body powered is good and myoelectric is junk, which results in people getting possibly upset – but there are many distinct ice-cold well founded engineering and medical arguments why I am right. The myoelectric curse particularly for below elbow amputees seems to be based on deep wounds instanced in the aftermath of the Russian arm [link] which still is what is used to this day.

But as society in general looks down on and devalues amputees (for which we all also collect cases, examples, as we go, if we want or not), I may take my time with this. Not everyone in society thinks that amputees are best not seen or heard, there are some very relevant voices that express a different opinion, but as things are, I can say whatever I want because those people in power of spending tax money on research are not interested in mundane vulgar everyday aspects including (but not restricted to), say, anatomy.

So I can mention “anatomy” as much as I want,  researchers/prosthetists/operators etc still will keep sticking electrodes on extensor carpi radialis muscles, have the amputee bend the elbow and see how they drop the object with involuntary prosthetic hand opening up, with everyone going “oops did not see that coming”. They spent over 50 million Euro on dead horse research, so, go figure [link]. So everyone chill, no one knows that extensor carpi radialis has been flexor of elbow to begin with.

For the moment, you’d really have to be an engineer to understand [link] – but, there you go, it is not even Easter yet and already we have free tips – if nothing else, try to use extensor to close the myoelectric device, and flexor for opening. In the long run, this is not a technical debate – not if you build a body-powered arm right – it is a sociological tragedy of ill-understood concepts of vulnerability. Thereby, true engineers and true prosthetic arm users are privileged and own, share, develop and cherish what seems to be (but is not) sacred secret knowledge, which exposes any outsider to vulnerabilities related to the absence of understanding, performance and affordability in a degree of reliability and constant availability they still dream of. Ignoring reality comes at that price.

@article{stuttaford2023reducing,
  title={Reducing Motor Variability Enhances Myoelectric Control Robustness Across Untrained Limb Positions},
  author={Stuttaford, Simon A and Dyson, Matthew and Nazarpour, Kianoush and Dupan, Sigrid SG},
  journal={IEEE Transactions on Neural Systems and Rehabilitation Engineering},
  year={2023},
  publisher={IEEE}
}