The limitations of biomechanics: You are more than cylinders, cones, and spheres

As a clinician I used to hold onto the idea that you could do a biomechanical analysis of an athlete and be given a print out of what was wrong with the movement pattern. I thought you would be able to use that information to instantly correct a motion and thus eliminate an injury. 

I have recently begun to dabble in biomechanics as part of my graduate studies. Through my journey I have discovered that I could not have been further from the truth when it came to my faith in an imaginary biomechanical analysis to fix something that we had determined was an issue with "mechanics." 

When referring to limitations of biomechanics, I should really be referring to limitations of motion capture. The study of biomechanics has so much more to it than tracking markers and 3D representations of movement. However, this is the cool stuff that the clinician in me was pulled towards. 

I think that this idea comes from training in movement screens and movement analysis. I have taken several classes regarding movement screening. Movement screens consist of taking an athlete through a specific movement or set of movements while a clinician watches and evaluates. These evaluations may be a simple pass/fail or a rating system. They are designed to give information to the clinician to then address the limitation. Movement screens aren't so different from coaching. A coach visually observes a movement and gives feedback to adjust and improve the outcome. 

Like anything, there are downfalls and benefits of movement screening. Can the human brain comprehend a complex movement and see it's flaws by watching? Can coaches and clinicians agree that there is a correct form to simple and complex movements to compare their observations to? Can an athlete repeat movements and is that actually the desired outcome?

The limitations continue and are perhaps better suited for a different post. The point here is to appreciate the magic that a biomechanical analysis promises to someone with no training regarding motion capture. Imagine if any movement could be viewed by computers, analyzed, and an output could be given to fix the issue. 

Motion capture markers

To understand the limitations, you need to understand the process of motion capture. In full honesty, I am currently learning to understand the process as motion capture is very new to me. A motion capture system has a set of high speed cameras that are positioned around an area where the motion will be evaluated. Most systems use marker sets that are small reflective balls attached to various, specific points on the body. Some motion capture systems are marker-less meaning that they are able to identify and track specific points without the reflective markers. 

Example Marker Set

The markers are placed on specific landmarks across the body that encompass a marker set. Marker sets vary based on the movement and body part being examined as well as the preference of the researcher. Once attached, the subject stands in the motion capture area while the cameras take a brief snapshot of the markers. The cameras are positioned around the motion capture area. The cameras triangulate the position of each marker with relation to the origin of the motion capture area to provide x,y,z coordinates for each marker. This provides the computer the ability to know where any given marker is at any given time using the (x,y,z) coordinates.

The next step is to build a model, which gives the markers meaning. The computer recognizes markers as individual points floating around in space. Building a model assigns the markers to specific segments to represent parts of the body. For example, markers placed on the forearm, wrist, and elbow are used to create a representation of the forearm. Once a model is completed, the model can be applied to any movement recorded with the motion capture system. 

This is where the error begins. First, the markers are assumed not to move in relation to the skeletal or anatomical landmarks they represent. However, anyone that has seen a pitcher throw in slow motion can see the movement of the skin around the elbow. Extra movement of markers is termed artifact and slightly alters the accuracy of the marker data. Secondly, in building the model, the body parts are represented as cones, cylinders, and spheres for the simplicity of geometrical calculations. However, the body is much more than simple geometry. Next, the weight of the segments are estimated by the subject's height and weight. There is not a clear way to measure the weight of a subjects forearm without cutting the body part off and weighing it, which undermines the injury prevention aspect of biomechanics. The weight is therefore estimated. Any estimations again will adjust the accuracy of the data.

The model does not do a good job of accounting for accessory motion. Our joints are modeled as hinges and ball in sockets but in reality there are accessory motions such as gliding and sliding that differentiate our elbows for example from simple door hinges. A large portion of the manual therapy that I use with my athletes focuses on accessory motion. In terms of the throwing motion, motion capture typically models the thorax as one singular cylinder that does not account for spinal extension and rotation nor does it account for the scapulothoracic joint.  Joint centers are estimated based on marker placement and in highly mobile or deep joints these centers may not represent the true joint center.

Finally, limitations in technology, such as sampling frequency, limit the data accuracy. A video is simply a series of photos and the sampling frequency is the rate at which those photos are taken. The motion capture system I have briefly worked with captures at a rate of 240 Hz or 240 times per second. This sounds fast but consider that the pitching motion may be less than a second from start to finish and any motion that occurs between frames will not be directly recorded. It is similar to poor GPS tracking for exercise applications. If you run through a neighborhood with lots of turns and your GPS does not have a high sampling frequency, then the GPS path won't accurately depict the route you took. 

With all of that being said, this post is not meant to disregard motion capture biomechanics. Rather, I believe understanding the process and limitations helps to better inform the information going forward. 

So what can motion capture be used for? I believe motion capture works well for slower lower body motion because the anatomy and rate of movement plays into the limitations mentioned above. Most of the research I read regarding motion capture is for upper body and pitching specifically. 

I think motion capture can be used for intra-subject comparison which is the comparison of the same subject before and after an implementation. For example, a subject could be evaluated during short toss and long toss via motion capture analysis to look at the difference. This in fact has been done and the difference in arm path has been established. 

A lot of work of motion capture has been used to establish norms for various levels when it comes to kinetic and kinematic measurements for the throwing motion. Kinetic refers to force measurements and their derivatives while kinematic refers to measurements of motion. There are established norms for how fast an arm moves while pitching and what the force measurements are while pitching for various skill levels. However, these are subject to the same limitations previously mentioned and norms are difficult to use since each athlete is unique. 

Driveline in Washington has used motion capture to evaluate the pitching motion in an attempt to unlock velocity. I won't try to encompass their entirety of work in a basic blog post but I do think they have done a good job of using motion capture to make clinical decisions regarding exercise programs and implementations. From what I understand, they have used motion capture to evaluate kinetic and kinematic measurements for efficiency of movement within the pitching motion and designed exercises geared at enhancing movement that correlates to velocity. 

As technology improves, I think that we will get better and more accurate data to make better clinical decisions. I think these technologies should be used to confirm or refute traditional knowledge regarding baseball and movement in general. For instance, curveballs are thought to be more stressful on the elbow. However, motion capture studies have not shown that to be the case. This is an example where traditional thinking has been evaluated by modern technology. That is not to say that every 8 year old should throw a curveball, it simply adds information to the debate. 

I hope to continue to work with motion capture in the future and find ways to use it for clinical decision making. I am very intrigued to see what new studies come out and how technology is being improved. 

That's all for now...from the training room.