I know there are several threads about spine rhythm but I have another question:
all components of coefficient matrix of SpineRhythmDrv are less than 1; does it mean that T12-L1 rotations are greater than other joints?

You could think of the spine rhytm as a weight vector given to each joint to distribute the overall thoracopelvic angle. You could see that this vector is “sorted” and increasing towards T12L1 - but that could be thought of as Pelvis-L1 < Pelvis-L2, Pelvis-L2<Pelvis-L3, etc.

Thank you for your response. You mentioned a very good point. Actually, I am wondering if these “weights” are assigned to total rotation of each vertebra with respect to pelvis or rotation of joints (i.e. relative rotation between adjoining vertebrae).

For example: Coefficient in 12th line of Coef matrix is 5.784718e-001. So, does in mean:

It might be or be somehow related to this number - I would need to dig into code. But I believe the easiest would be to construct a small kinematic model that is driven to flex the body and simply read out T12-Pelvis vs the angle of interest. It should have a constant slop if you are using the spine rhythm.

I have the same questions as above. Based on the code, the weight should be assigned to total rotation of each vertebra with respect to adjacent vertebra. So the result is that the T12L1 has the largest motion than other joints. It doesnot make sense. Besides, where to get the coef ? is there any reference?

It is not correct to say that T12L1 has the largest rotation. The weight that you are looking at actually represents L1Pelvis angle as far as I remember.

Unfortunately we do not have a good documentation on the spine rhythm, but the coefficients were derived from solving a beam bending problem: PDF for the spine rhythm idea

Dear pgalibarov
Thanks for your reply. I am not very clear about the coef, Is it right that:
if the thorax_pelvis_flexion is 20 deg. The anybody will attribute how many degress to the T12L1Jnt_flexion, L1L2Jnt_flexion,L2L3Jnt_flexion, and etc based on the coef.
By the way, i want to use this coef to drive the five lumbar vertebras and want to write it into my paper. So if there is any paper discussed the coef, I can refer to it.

Yes, that is correct, but manipulating these weights is not a good idea.

If you want to get individual joint angles - you can simply exclude this spine rhythm from the analysis, and drive individual joints as you want. An example of exclusion is given in the SpineFixationWithForceDepKin example of the AMMR (check Model/Drivers.any file).

Unfortunately, the spinal rhythm was not formalized into a paper. The best idea to learn kinematic rhythm is to make a small 3 segment example and see how these coefficients affect individual joint kinematics between those segments.

I caculated the matrix of axes of L4_seg with respect to the matrix of the axes of L5_seg.
Then, transfered the result of matrix into Euler angle, and found that the Euler angle was the exactly the same with Jnt45.Pos. So the Jnt45.Pos represents the sagittal,coronal,horizontal angle with respect to adjacet segment.

However, the angle, JntT12L1.Pos>JntL1L2.Pos>JntL2L3.Pos>JntL3L4.Pos>JntL4L5.Pos>JntL5Scram.Pos. And this result is consistent with the explanation of the coeff of spinerhythm.

However, this result is totally unacceptable. it dosenot make sensor at all.

As i mentioned before, the spine rhythm coefficients were derived from the beam theory equations.

But in reality these coefficients will be dependent on the joint and corresponding ligament stiffness. So a stiffer joint, e.g. with pathology, would allow less motion in the spinal unit, whereas a more elastic one will be moving more. Unfortunately it is difficult to calibrate the discs, so by default these values are used. So it represents a single person with the stiffness increasing towards the sacrum, and it does make sense in this particular case, because it was defined like that. If you provide your values you could use SpineFixationWithForceDepKinematics model to define subject-specific values.

Please have a look at the abstract that describes how this behaviour could be changed:

Thanks for your reply. I use motion capture system to get the coordinate of makers placed on the spinal process of L1,L5, T12 and T3. Then I drove mocap model with these markers. So I got the lumbar joint angle. However, the increasing angle from sacrum to thoracic segment maynot be reasonable no matter for health spine or pathological spine. The hypothesis that the stiffness increasing towards the sacrum dose make sense. But the angle of each joint doesnot make sense. How can I get reasonable lumbar joint angle through mocap model driven by the coordinate of markers I colleceted.

Thanks for your reply. I use motion capture system to get the coordinate of makers placed on the spinal process of L1,L5, T12 and T3. Then I drove mocap model with these markers. So I got the lumbar joint angle. However, the increasing angle from sacrum to thoracic segment maynot be reasonable no matter for health spine or pathological spine. The hypothesis that the stiffness increasing towards the sacrum dose make sense. But the angle of each joint doesnot make sense. How can I get reasonable lumbar joint angle through mocap model driven by the coordinate of markers I colleceted.

You have 3 options:
[ol]
[li]To leave the current kinematic rhythm to drive the lumbar spine, and assume that the overall posture and muscular configuration does not differ much from the “real” one. Possibly think what kind of error is introduced by difference in IVAs, and how it affects your final measurements of interest.
[/li][li]To replace the current rhythm with a new one, which would require you to find distribution of the intervertebral angles wrt. thoracopelvic angle yourself. An idea might be to take angle values from literature, measure your own curves from different posture x-rays, or contact an entity that measures such angles, e.g. a company: OrthoKinematics, EOS, etc.
[/li][li]Use a special solver, Force Dependent Kinematics (FDK) solver, that find curvature based on the joint stiffnesses and ligament configuration, but it might be a more time/effort consuming simulation. Please check our manual for FDK.
[/li][/ol]

Thanks for your reply. Your suggestion is quite good. Then I tried to find the angle value from literature. But I found another question. I found one paper named “Which motion segments are required to sufficiently characterize the kinematic behavior of the trunk?” The angle value of T3 with respect to T12 was more than 20 degrees, that is to say, there were much motions among thoracic segments. But anybody models in AMMR all considered the thoracic segments as a rigid body. The result in that literature was similar with our experiment. I tried to add the 20 degrees to the T12L1 to represent the thoracic movement. It seemed a little large for one spinal unit. Is there any anybody model with detailed thoracic model? Or is there any other method to solve the problem of intersegmental motion of thoracic segments in anybody model?

There is an existing model of the thoracic spine developed by ETH Zurich. Unfortunately this model will appear in one or another form in the AMMR later, but at the moment it is not available - possibly you could contact the authors directly for collaboration or to get the model.

Currently you cannot kinematically drive the thorax, but you could morph the thorax using our nonlinear scaling methods to take a certain shape, which might solve your case depending what your question of interest is.