A couple of days ago I received an E-mail message from Biomch-L
with a draft of the report on "Recommendations for Standardization in
the Reporting of Kinematic Data", from the ISB Committee for
Standardization and Terminology. Almost at the same time, I received
through regular mail ("snail mail") a hard copy of the same report
(Draft version 4.0 of March 23, 1992, ISB Newsletter 45:5-9, 1992.)

I welcome the drive to standardize terminology and conventions,
although I think it is going to be difficult to come up with reference
frames and conventions that will apply in a satisfactory way to a joint
at all points of its range of motion.

The plan is to develop a standard set of conventions for the
presentation of data in the refereed literature in Biomechanics.
Therefore, the final result of this project will affect ALL OF US, and
I am glad that Peter Cavanagh, Chair of the Committee for
Standardization and Terminology is asking for input. I think Biomch-L
would be a good place to establish a public dialogue on this topic,
because the "criss-crossing" of comments may stimulate the thinking of
readers that might otherwise remain passive. Since Peter is a Biomch-L
subscriber, the comments will reach him too. My hope is that the
process will be constructive.

Being the lazy bum that I am, I would much prefer to watch "from
the sidelines" as this whole question unfolds. However, I feel that
this issue is much too important for that, and I have no option but to
contribute my own "two-cents' worth". So here it goes.

(My comments will refer to the hard-copy draft mentioned above,
rather than to the electronic copy distributed through Biomch-L,
because the former showed drawings which made it a better copy for me
to read, but the text sems to be identical in both.)


(COMMENT A)

In "Part 1: Definition of a global reference frame", the proposal
recommends that a global XYZ reference frame be adopted, with the X
axis horizontal and pointing forward (in the general direction of the
motion), the Y axis vertical, and the Z axis horizontal and pointing
toward the subject's right. The rationale is that this will permit
two-dimensional (2D) studies to use an XY reference frame and
three-dimensional (3D) studies to use an XYZ reference frame, so the X
and Y axes will be common in the 2D and 3D studies. I don't think this
is a good idea.

I would propose that in 3D studies the X axis should be horizontal
and pointing toward the right, the Y axis horizontal and pointing
forward, and the Z axis vertical. In 2D studies, a YZ reference frame
should be used, with the Y axis pointing forward and the Z axis
vertical.

Rationale: The vertical axis is a special axis from a mechanical
standpoint, because this is the only direction in which gravity acts.
Therefore, we could classify the axes of the global reference frame
into two categories: the horizontal axes in one group, and the vertical
axis (the "special" axis) forming the other group (by itself). When we
choose names for these axes, it would seem cleanest to give consecutive
letters to the axes of each group, so X and Y for horizontals and Z for
verticals would make sense.

Why is the committee proposing instead X and Z for horizontals,
and Y for verticals? I think it is because in 2D analysis X has
traditionally been used for the horizontal and Y for the vertical, and
the committee is (very correctly) trying to produce consistency in 2D
and 3D terminology. But why have people generally used X and Y to
label the reference axes in 2D analyses? Why not A and B? Because, to
minimize confusion, it is better to use letters that are less common,
such as the last letters of the alphabet. Why then do the people that
do predominantly 2D analysis use X and Y instead of Y and Z? Because
although they are studying a 2D problem, they realize that there does
exist a third dimension, so they keep the last letter (Z) "reserved"
for the remaining dimension, even though their problem is strictly 2D
and no use at all will be made of the third dimension. Occasionally, a
person that generally uses 2D analysis will need to make a 3D analysis.
Since this person is used to seeing X and Y on the plane of the writing
paper, it is only natural for that person to keep them that way, and to
define Z as an axis normal to the plane of the paper, i.e., pointing
horizontally if the drawing shows a side view of the subject.

Present trends indicate that in the future, research work will be
more and more in 3D rather than in 2D. And motion in reality is, of
course, 3D and not 2D. So, why should we let the inertia of 2D customs
dictate what our terminology will be in 3D analysis? Why let the tail
wag the dog? If we want to have consistency between 2D and 3D
analyses, I believe that it is the conventions of 2D analysis that
should be modified to use Y for the horizontal and Z for the vertical.

-----------------------------------------------

(COMMENT B)

In "Part 1: Definition of a global reference frame", the axes of
the global reference frame, fixed to the ground, and with origin at
point O, are called XG, YG and ZG. (Sorry, I can't do subscripts in
E-Mail!) I realize that the "G's" stand for "global", but the letter
"G" is often used to designate the center of mass (or center of
gravity), and therefore this terminology could be confusing. I would
propose instead XO, YO and ZO.

-----------------------------------------------

(COMMENT C)

In "Part 2: Definition of segmental local center of mass
reference frames", the local X axis is assigned to the
anterior/posterior direction; the local Y axis to the proximo/distal
direction; and the local Z axis to the medio/lateral direction. I
don't think this is the best choice.

As with the axes of the global reference frame (see "COMMENT A"
above), we could classify the local axes of most segments into two
categories: the anterior/posterior and medio/lateral axes in one group,
and the proximo/distal axis (by itself) forming the other group. This
is because the moments of inertia of the segments are generally large
and similar to each other with respect to the anterior/posterior and
medio/lateral axes, and much smaller with respect to the proximo-distal
(longitudinal) axis. Therefore, when we choose names for these axes,
it would seem cleanest to give consecutive letters to the axes of each
group, so X for medio/lateral; Y for anterior/posterior; Z for
proximo-distal would make sense. This terminology would also fit well
with the terminology proposed in "COMMENT A" above for the global
reference frame.

-----------------------------------------------

(COMMENT D)

In "Part 2: Definition of segmental local center of mass
reference frames", the committee puts forward two alternative opinions
with respect to the definition of the segmental local axes:

The "majority opinion" in the committee proposes that, for all
segments, the medio/lateral axis should point toward the right, the
anterior/posterior axis forward, and the proximo/distal axis in the
proximal direction;

The "minority opinion" (Prof. John Paul) proposes that left-handed
reference frames be used for the segments on the left side of the body.

I feel that this is an important point, and that more thought
needs to be given to it. I have not been able to decide which is the
best way to go in this, but here are some of my thoughts on this
question. The problem with the proposal of the MAJORITY OPINION is
that a positive torque (or angle change) in the internal/external
rotation direction will mean "internal rotation" in one leg, but
"external rotation" in the other leg. The same problem will occur with
abduction/adduction, while flexion/extension will be OK. With respect
to LINEAR kinetics, positive forces in the medio/lateral direction will
mean medial forces in one leg, but lateral forces in the other; forces
in the other two directions will be consistent in both legs.

What would happen if, instead, we sought the goal of Prof. John
Paul's MINORITY OPINION, in which adduction (and presumably also
internal rotation, although he does not mention this explicitly), as
well as flexion, would have the same sign in both legs? For this (and
using X for medio/lateral; Y for anterior/posterior; and Z for
proximo-distal as proposed in COMMENT C above), the right thigh would
have an X axis pointing toward the right, a Y axis pointing anteriorly,
and a Z axis pointing from the right knee toward the right hip, while
the LEFT thigh would have an X axis pointing (AGAIN!) toward the right,
a Y axis pointing posteriorly, and a Z axis pointing from the left hip
toward the left knee.

The surprise is that the reference frame just defined for the left
leg would still be RIGHT HANDED! So left-handedness does not seem to
show up as a problem at all!!

However, while I can't see a problem with respect to angular
kine(ma)tics, I do see one with respect to LINEAR kine(ma)tics. For
instance, a force in the positive Z direction would be pointing
proximally in the right thigh, but distally in the left leg; similar
inconsistencies would also occur in the X and Y directions.

I have not found a perfect solution, so I think we have a hard
choice. But I think I would rather use the modified reference frame
last described, because with that reference frame the three torques
will be consistent in both legs, and in my opinion, joint torques are
more important parameters than joint forces. If there is a sacrifice
in consistency, I would rather make it in the joint forces than in the
joint torques. Also, to make the forces be consistent in the right and
left side segments, all that would be needed would be to reverse the
signs of all three components (X, Y and Z) in the left side segments.
That may actually be cleaner than reversing only the sign of the forces
in the medio/lateral (X) direction, which is what would be needed if we
used the original reference frame (see 4th paragraph before this one).

-----------------------------------------------

(COMMENT E)

In "Part 1: Definition of a global reference frame", the proposal
recommends that the horizontal axes of the global reference frame
should point respectively to the right and forward. It states that
"All directions are given for the subject facing the direction of
working or travel that is of most interest to the particular activity".
Such a definition makes sense for a gait analysis situation or a
javelin throw, for example. However, it will not always be appropriate
for other activities, such as a Fosbury-flop high jump, where the
athlete at the end of the run-up will be moving in a diagonal direction
with respect to the bar. In the high jump, a reference frame oriented
to fit with the final run-up direction can sometimes be useful, but
often a reference frame aligned with the high jump BAR is still more
useful. Therefore, I think the requirements for the directions of the
horizontal axes should be redefined to admit occasionally directions
that are orthogonal but not aligned with the direction of working or
travel of the subject.

-----------------------------------------------

(COMMENT F)

In "Part 1: Definition of a global reference frame", the
committee proposes that the origin of the global reference frame be
located at an important point for the activity being analyzed (for
instance at the center of one of the force plates being used). I agree
with this, but in our laboratory we have added an extra "twist" that we
find very useful.

In our laboratory, we typically add 10.00 meters to all
horizontal coordinates. In effect, this puts the origin of the global
reference frame at a point located at a nice, round, number of meters
away from the "important point". Using a reference frame where Z is
the vertical, this makes the "important point" have coordinates (10.00,
10.00, 0.00). This eliminates negative coordinates, and we feel that
it facilitates the interpretation of the data. For instance, instead
of saying that the center of mass moved from position X1 = -0.12 m to
position X2 = +0.08 m, and finally to position X3 = +0.28 m, our data
would say that the center of mass moved from position X1 = 9.88 m to X2
= 10.08 m, and finally to X3 = 10.28 m. Our way of expressing the data
still allows us to see clearly that the c.m. passed over the X
coordinate value of the "important point" (it is just as easy to see
that the c.m. goes through the X = 10 meters position as it would be to
see that it goes through the X = 0 meters position), but our way of
expressing the data also allows us to see very easily in this example
that the changes in position in each of the two intervals are constant
delta X values of 0.20 meters (9.88 ... 10.08 ... 10.28), while using
the other method, with the important point at (0.00, 0.00, 0.00), this
would be harder to see (-0.12 ... 0.08 ... 0.28). So we feel that the
change in sign associated with the important point being at position
(0, 0, 0) produces some confusion which can be avoided by making the
important point be at (10, 10, 0) and thus keeping all coordinates
positive.

-----------------------------------------------

(COMMENT G)

In "Part 5: Relative attitudes", I believe that the labels of
vectors X2 and Y2 in Figure 3 (ISB Newsletter) were mistakenly
interchanged.


-----------------------------------------------

(COMMENT H)

I believe that the biggest bone of contention is going to be in
"Part 5: Relative attitudes", in the formulation of the joint rotation
conventions for each joint of the body. That is going to be a tough
nut to crack. We'll cross that bridge when we get to it ...


-----------------------------------------------

(COMMENT I)

In "Part 6: Joint Moments", I believe that the word "moment"
should be replaced by "torque". The word "moment" must be the most
overused one in mechanics. What would you rather see, "The force
exerted a large moment at that moment because of its large moment
arm.", or "The force exerted a large torque at that instant because of
its large moment arm." ? I would hate to see the word "moment"
enshrined as the official term for "torque" in biomechanics.


-----------------------------------------------

(COMMENT J)

In "Part 6: Minority report", "With reference to part 1:", it is
proposed that the letters XYZ be avoided. The rationale is that many
equipment manufacturers have already formatted data according to their
own XYZ system, and awkward transpositions, such as Yisb = Zkistler
could be avoided by using other letters in the proposed ISB
standardization, for instance the letters RST. I don't agree. I don't
think ISB should relinquish to the manufacturers the "exclusive" right
to use XYZ!!


Jesus Dapena
Department of Kinesiology
Indiana University
Bloomington, IN 47405
U.S.A.

dapena@ogre.cica.indiana.edu