vsjserge39

09-28-1995, 08:23 PM

Dear Biomch-L Readers,

It seems that a new "hot subject" is raising on our list. It is also an

interesting one.

When I read the proposal in J. Biomch, I thought: "it's absolutly not the

system we are using now, we will have to alter all our routines or to make

supplementary transformations ...". So , I was not too much enthusiastic to

switch (I'm quite lazy), but was ready to do it to be up-to-date. The

ISB RECOMMENDATIONS also sound quite official.

Now after a few email messages on the list, it appears that the official

proposal does not make everybody happy.

So, I wrote this note just to give another example where the unofficial

(as described in Dct Hatze's note)

but widely accepted axes system would be replaced by the the new one without

giving any advantages (on the contrary!).

When building anatomical models from medical imaging using mathematical

surface interpolation (e.g., B-splines), atomical structure contours

are first processed before building the whole volume.

Medical images (e.g., CT-Scan) taken in a transversal plane are especially

usefull to get nice contours.

The contours on each image are bi-dimensionnal with the third dimension

remaining constant. Very logically, both dimension of the plane are given as

X and Y by the CT system, the constant dimension being Z.

After contours are interpolated, the volume patches are obtained in the

third dimension Z to get the description of the full object.

Now, if you want to make a motion simulation on the model it's easy:

flexion-extension along the transversal X, abduction-adduction along the

saggital Y and axial rotation along the axial Z.

It's a very natural convention. Furthermore, the implementation of such a

coordinate system is really easy to implement in a computer graphics

environnement which coordinate system convention are:

X-Y plane parallel to the screen, Z along the screen depth.

You see the whole system is really in harmony: surface interpolation axes system

fits into the kinematic axes system that fits into the computer graphics axes

system (though some graphics system use a left-hand system, but the XY plane

is still parallel to the screen).

Adopting the new convention means for our project:

- Transform the medical image system coordinates into the new system.

- Then to display the data, transform the new data into the

computer graphics system.

It means two supplementary manipulation of the data (who said "error increases

with the number of operations a data has to go through"? ) and time

processing also increases.

Now, I'm really opened to any new ideas, but still waiting for a solid argument

to adopt the new convention.

A new convention will never satisfy everybody (those who will have to adapt).

The questions of H. Rassoulian are interesting: is a convention really

necessary?

It seems to me there are so many different fields in biomechanics using

so many different conventions (often depending of the hardware) that

a standard will be really difficult to use for ALL applications.

Serge VAN SINT JAN

Department for Human Anatomy (CP 619)

Faculty of Medicine

University of Brussels (ULB)

808, Lennik Street

B-1070 Brussels

Belgium

voice: int + 32 2 555 6325

fax: 6378

email: sintjans@ulb.ac.be

It seems that a new "hot subject" is raising on our list. It is also an

interesting one.

When I read the proposal in J. Biomch, I thought: "it's absolutly not the

system we are using now, we will have to alter all our routines or to make

supplementary transformations ...". So , I was not too much enthusiastic to

switch (I'm quite lazy), but was ready to do it to be up-to-date. The

ISB RECOMMENDATIONS also sound quite official.

Now after a few email messages on the list, it appears that the official

proposal does not make everybody happy.

So, I wrote this note just to give another example where the unofficial

(as described in Dct Hatze's note)

but widely accepted axes system would be replaced by the the new one without

giving any advantages (on the contrary!).

When building anatomical models from medical imaging using mathematical

surface interpolation (e.g., B-splines), atomical structure contours

are first processed before building the whole volume.

Medical images (e.g., CT-Scan) taken in a transversal plane are especially

usefull to get nice contours.

The contours on each image are bi-dimensionnal with the third dimension

remaining constant. Very logically, both dimension of the plane are given as

X and Y by the CT system, the constant dimension being Z.

After contours are interpolated, the volume patches are obtained in the

third dimension Z to get the description of the full object.

Now, if you want to make a motion simulation on the model it's easy:

flexion-extension along the transversal X, abduction-adduction along the

saggital Y and axial rotation along the axial Z.

It's a very natural convention. Furthermore, the implementation of such a

coordinate system is really easy to implement in a computer graphics

environnement which coordinate system convention are:

X-Y plane parallel to the screen, Z along the screen depth.

You see the whole system is really in harmony: surface interpolation axes system

fits into the kinematic axes system that fits into the computer graphics axes

system (though some graphics system use a left-hand system, but the XY plane

is still parallel to the screen).

Adopting the new convention means for our project:

- Transform the medical image system coordinates into the new system.

- Then to display the data, transform the new data into the

computer graphics system.

It means two supplementary manipulation of the data (who said "error increases

with the number of operations a data has to go through"? ) and time

processing also increases.

Now, I'm really opened to any new ideas, but still waiting for a solid argument

to adopt the new convention.

A new convention will never satisfy everybody (those who will have to adapt).

The questions of H. Rassoulian are interesting: is a convention really

necessary?

It seems to me there are so many different fields in biomechanics using

so many different conventions (often depending of the hardware) that

a standard will be really difficult to use for ALL applications.

Serge VAN SINT JAN

Department for Human Anatomy (CP 619)

Faculty of Medicine

University of Brussels (ULB)

808, Lennik Street

B-1070 Brussels

Belgium

voice: int + 32 2 555 6325

fax: 6378

email: sintjans@ulb.ac.be