Running video analysis can be a wonderful tool at the right time and in the right hands. However, as with any tool, it is only as good as the person using it. Here I will present my take on what constitutes thorough, accurate, and responsible video gait analysis. Suggestions and stories of any gait analysis experiences would be most welcome.
What is a video gait analysis?
It’s exactly what it sounds like: recording video of someone running and then analyzing said video for “stuff”. There is a wide range of things called video gait analysis, and I’ve worked with a pretty full spectrum of methods and technologies so I can fill you in on what they are.
Back in my undergrad days, it was hand digitizing frame-by-frame off actual videotape. Yes, it was as tedious as it sounds. By the time I was doing my masters, we were using a real-time fully 3-d motion capture system similar to what they used to make Gollum come to life in LOTR. I think we had 9 infrared cameras set up with our Vicon system, several force platforms, and heaps of esoteric software. Yes, it was as complicated as it sounds.
The first is quaint and outdated, the later is generally inaccessible and super expensive. I hit the sweet spot in my post-Vicon job working in an orthotics lab that specialized in runners and military personnel. We were back to 2d recording, but had 3 or 4 cameras synchronized, getting views from the front, back, and both sides. This is what the majority of running video analysis consists of in the world of running nowadays. It is NOT 3d, although by recording from 3 positions (front, back, side) it is a pretty good substitute.
Why Should You Do A Gait Analysis?
Obviously, if you are getting injuries all the time, then video analysis is a great tool to maybe see what is causing it. However, having video gait analysis done to check your running form is a good idea even if nothing is “wrong”. Proper recording and analysis can show small inefficiencies or imbalances that are all below the threshold for causing injuries, but may be impairing your speed, compromising your efficiency, or both. In fact, the sub-clinical “injuries-in-waiting” could well become real issues if you ramp up training intensity and/or volume, so that is another scenario that would benefit from analysis.
Who Does Your Running Gait Analysis Matters
Pretty much anyone with a phone or laptop can toss up a virtual sign saying “Video Gait Analysis”. Ideally, you’d want someone with both knowledge and experience with running and biomechanics. Some runners or running coaches have heaps of practical experience but have no knowledge of the science needed to identify and correct issues. On the other end of the spectrum you can have gait lab geeks who can do inverse dynamics in their sleep but have no practical experience with running.
If you say “closed kinetic chain” and they go “huh?, then you have the former. If you talk about MAF or LTHR paces and get blank looks, then you have the latter.
Where Can I Get A Gait Analysis Done?
Many universities and colleges have some sort of gait lab, and will provide analysis services. A Rolls Royce example would be Dicharry’s SPEED lab at the University of Virginia. Some people will provide the service in person, meaning you’d get together somewhere and they would record video of you and then prepare and send you their analysis. Finally, there is the SIY option (shoot it yourself).
With modern cell phones, pretty much anyone has the technology in their hands to record useful video for analysis. The key word is “useful”, so for SIY to work you need to be provided with detailed and correct instruction on how to shoot video of yourself. You then send the video files to the person doing the analysis, and they’ll send you a completed analysis. The above are basically in descending order of both price and quality, which vary quite a bit.
How Is A Video Gait Analysis Done Properly?
Before anything else, an analyst needs to take both a running history and an injury history of each runner. Running history should cover things like the length of time running, range of distances/times, weekly distances/times, surfaces, and footwear. Injury history should not be limited to running injuries i.e. “I broke both femurs skiing” is likely going to be relevant. Once this is done, the actual video part of things comes next.
I would always do 2 recordings: one after a suitable warmup, and one after the runner has tired themselves out. Many people look flawless when they are fresh. It’s when fatigue sets in that things can start to fall apart. Since this is when many injuries occur, it only makes sense to see where the weakest links are in terms of what, if anything, goes belly-up when the runner gets tired.
Treadmill familiarization can be an issue for some people. Since the overwhelming majority of running analysis is done on treadmill, it’s a good idea to practise on one until you feel comfortable on it. Otherwise you won’t be showing your “true” running form as you are wobbling along on the accursed treadmill that you swore you’d never run on.
Very few analysts will have instrumented treadmills or force platforms, but this doesn’t mean that forces and pressures are closed books. Useful force and pressure information can be gleaned by examining used training and racing shoes. Both the insoles and the outsoles will show distinctive wear patterns that can throw up red flags to someone who knows how to read them.
The most important and difficult part is figuring out what, if anything, is wrong, whether or not to correct it, and how to go about doing that safely and effectively. More on that later.
5 Requirements for Good Video Analysis
QUALITY VIDEO
Shooting good quality video is vital, but is only the first step in good gait analysis. For starters you need to shoot somewhere bright, because without adequate lighting, you cannot get clear images at the frame-rates and resolutions required. Those should be no lower than 60 frames-per-second (fps), with 120 fps being a solid and accessible number for most people. Resolution is trade off between detail and frame rate, but it should be 720p or higher.
Running happens in 3 dimensions, so unless you are in a motion capture studio or a Vicon-equipped biomechanics lab, you’re going to need 3 cameras, or (for those on a budget) to shoot from 3 angles (front, back, and side) in sequence. Camera placement needs to have the runner centered, and camera orientation needs to be at 90 degrees to the plane of movement being captured, otherwise you get perspective errors. Since they are stuck on a treadmill they won’t be able to wander in front of or behind the plane of motions, so you can set up as close to the subject as possible without cropping off important bits. You cannot measure angles and displacements accurately if you bugger the positions up, so you should set them up thusly-ish.
KNOWING WHAT TO MEASURE
The second step is to know which aspects of running form to pay attention to. Unless you are the proud owner of a force platform or instrumented treadmill, you don’t have to worry about those pesky Ground Reaction Forces. In fact, almost all non-research-driven running analysis only measures kinematic data (movements) and leaves kinetics (forces, torques, work, power, and the like) to the biomechanists. Coz inverse dynamics is a bitch.
What anatomical landmarks will you be tracking? Which movements are important? What about posture? With movements at the foot, ankle, knee, hip, spine, shoulder, and elbow, as well as postural considerations, the data can get overwhelming if you just “collect everything”. In addition, there are different schools of thought about what determines “good” running form, and what measurements they require.
At a minimum, you need to track the ankles, knees, hips, and pelvis. I would ideally add markers to track lumbar lordosis, CoM vertical oscillation, shoulders, elbows, wrists, and head. This is best done with actual markers on the skin. The reason is that without markers you’ll be eyeballing joint centres and anatomical landmarks with every measurement you try to make. This is hugely imprecise. Having markers means that each measurement of, for example, knee angle is comparable because you KNOW you are using the same exact points of reference for hip, knee, and ankle.
The above markerset is an example of overkill for what we are talking about.
IS THIS GOOD OR BAD?
The third step is to analyse the selected variables with respect to the specific athlete and decide if they are “good” or “bad”. It is one thing to look at elite runners as the templates for optimal running form - which IMHO is a solid approach - but quite another to indiscriminately and inflexibly apply them to every athlete you analyse.
It’s a bit like ballet: there are definite positions and movements required for optimal running, just as there are very specific positions and movements in ballet. The problem is that, as with ballet, many people simply cannot get their body safely into those positions or perform those movements. In dance, the response to this was the advent of “modern dance”, for the 99% of us who don’t have the body shape and joint mobility to perform Swan Lake. In running analysis, it requires stepping away from the cookie-cutter approach and incorporating individual athlete body structure and running/health/injury histories into the analysis.
DOES ANYTHING NEED CHANGING?
The penultimate step is to take anything and everything that is “bad” or even sub-optimal and decide whether they are worth changing. Just because something “looks” bad, wonky or inefficient doesn’t mean that it needs changing.
Nobody who knows what they are doing would grab Meb (Boston Marathon winner and King of Heel Striking) and tell him to change his footstrike. The same with Paula Radcliffe’s head bob. She has the fastest ever marathon time for women at 2:15:25, so I’d say it’s a case of “leave well enough alone”. Haile Gebrselassie’s epic pronation sure didn’t hurt his running performances or career longevity, contrary to what you keep hearing about overpronation.
TOOLS AND FOLLOW-UP TO ADDRESS ISSUES
Analysis just tells you if there are issues, what they are, and what the possible consequences are. This is nice but doesn’t help if you don’t get a detailed plan for how to correct, minimize, or work around said issues. Mobility, strength, drills, a change in footwear, a change in training type/volume/intensity are all possible parts of the solution. These need to applied, tweaked and then some form of follow-up and ongoing support needs to be provided. Finally, a re-assessment should be done after a suitable interval to see if the interventions have been effective.
What are common mistakes/red flags of a video gait analysis?
The biggest challenge to developing anything like a standardized system for running analysis is defining what “optimal form” is. There is such a wide range of acceptable forms that having a model that represents “Run Like This” is hopeless and impossible. By “acceptable” I mean form that is not increasing the likelihood of injury and not impairing performance. By “performance” I mean both pace and economy.
My solution to this challenge is to take the opposite approach and develop boundaries that separate “good” form from “bad” form. In other words it will be based on “Don’t Run Like This”. Regardless of the level of runner, the most important thing for them is to not get injured. The doctrine of “do no harm” should form the basis of running analysis. I’m currently in the process of doing a big literature search on running injuries and their mechanisms in order to come up with sensible boundaries between “good” and “bad”.
Should You Change Running Form?
Probably the most damaging error I come across is that of being bound and determined to find something to fix, come hell or high water. You then end up trying to fix something that isn’t broken, and at best wind up wasting time and effort. At worst you actually cause problems instead of fixing them.
Actively correcting running form is a bit of a gamble, even if you have isolated a genuine issue that needs to be fixed and prescribed the appropriate drills or cues to correct it. This is because many times there are good reasons why the runner is running the way they are: protecting or compensating for a weak muscle, synergist dominance, restricted mobility, etc. Using the “top-down” approach of changing how you run means you are basically assuming/hoping that the physical structures involved are up to the task They may not be, and this is when you can cause injuries despite having got everything “right”.
Again, I think that turning the whole paradigm on its head is the way to go. This “bottom-up” approach involves using running analysis as a screening tool for movement dysfunction. You then look to test and correct the dysfunction rather than trying to alter the way people run. Often treating the dysfunction will allow “better” running without them ever having to think about it. This way you ensure that the muscles and joints will be up to the task, and that you don’t risk causing injuries.
But wait...there is one more hurdle: dysfunctions can be either neutral, adaptive, or negative, and you need to figure out which one you’re dealing with.
Neutral dysfunctions look ungainly but cause no harm either through increased risk of injury or decreased performance. Leave them alone. Adaptive dysfunctions are compromises in posture and/or movement to account for specific and individual structural challenges. They would likely be negative in the general population but are the optimal way for this specific runner to move. Leave them alone.
Negative dysfunctions will result in an increased risk of injury and/or decreased or limited performance. These are the ones that need to be assessed further.
The following photos are examples of either neutral or adaptive dysfunctions.
These might look horrible at first glance but both would be cases I’d probably leave alone, for 2 reasons. For starters, the mis-alignments are with the swing leg, so the loads and risk of injury are miniscule. For another, both of these ladies are very successful world class runners, so whatever they are doing has been working pretty darn well.
Here is a simplified run-through of the process:
Lack of hip extension is a very common finding. You then go and do basic mobility assessments on the single and double-joint joint hip flexors, as they tend to be the prime culprits. Combining that with assessment of activation and strength in the primary and secondary hip extensors can point you to a comprehensive and progressive program of mobility and activation that doesn’t have them do a thing directly about their running.
In addition to excessive or insufficient mobility, asymmetry is likely to be a significant player in terms of injury risk. This means that even good patterns could be bad if the R - L sides are at opposite extremes of the “good” zone.
I think that there are very few people out there doing gait analysis who have the knowledge and skills to do the whole thing on their own, start to finish. Myself included. That is why I am partnering with physical therapists, because their knowledge and experience in assessing and addressing movement dysfunctions and injuries is way better than mine. An ideal scenario would be 2 people teaming up, one with knowledge of biomechanics, one with knowledge of physical therapy, and both with knowledge of running.
If you have any questions about this, please feel free to contact me:
petedobos@gmail.com, or on Facebook
About the Author
Peter Dobos has his B.Sc in Human Kinetics, completed coursework for his M.SC in Biomechanics, and has extensive adventure racing and off-road multisport experience.
He is affiliated with the Endurance Project and works with many elite OCR athletes.
